Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions

Challenge results

Task description

The scope of this task is to identify whether the sound emitted from a target machine is normal or anomalous via an anomaly detector trained using only normal sound data. The main difference from the DCASE 2020 Task 2 is that the participants have to solve the domain shift problem, i.e., the condition where the acoustic characteristics of the training and test data are different.

More detailed task description can be found in the task description page

Teams ranking

Table including only the best performing system per submitting team.

Rank Submission Information Evaluation Dataset Development Dataset
Submission Code Technical
Report 		Official
Rank 		Official
Score 		ToyCar
(AUC) 		ToyCar
(pAUC) 		ToyTrain
(AUC) 		ToyTrain
(pAUC) 		Fan
(AUC) 		Fan
(pAUC) 		Gearbox
(AUC) 		Gearbox
(pAUC) 		Pump
(AUC) 		Pump
(pAUC) 		Slider
(AUC) 		Slider
(pAUC) 		Valve
(AUC) 		Valve
(pAUC) 		ToyCar
(AUC) 		ToyCar
(pAUC) 		ToyTrain
(AUC) 		ToyTrain
(pAUC) 		Fan
(AUC) 		Fan
(pAUC) 		Gearbox
(AUC) 		Gearbox
(pAUC) 		Pump
(AUC) 		Pump
(pAUC) 		Slider
(AUC) 		Slider
(pAUC) 		Valve
(AUC) 		Valve
(pAUC)
DCASE2021_baseline_task2_AE DCASE2021baseline2021 51 56.375 65.93 52.32 68.51 57.56 60.68 50.50 65.49 56.86 58.30 50.98 57.22 51.41 51.87 50.07 62.49 52.36 61.71 53.81 63.24 53.38 65.97 52.76 61.92 54.41 66.74 55.94 53.41 50.54
Tozicka_NSW_task2_4 TozickaNSW2021 19 61.186 68.23 59.40 51.71 51.79 73.17 62.02 64.70 57.86 78.65 65.16 69.89 59.61 53.93 54.22 74.40 58.22 73.94 58.88
Asai_PFU_task2_1 AsaiPFU2021 40 57.845 46.66 50.66 53.67 51.94 74.41 61.15 52.39 51.20 73.87 62.54 77.15 60.06 57.25 55.72 61.84 55.79 65.63 56.02 61.65 61.44 65.59 53.09 64.58 57.40 70.85 58.78 74.20 67.47
Bai_LFXS_task2_2 BaiLFXS2021 43 57.040 49.39 57.51 42.53 49.75 75.15 61.32 58.61 55.55 58.12 51.21 83.09 71.86 55.29 53.34 56.32 55.29 64.12 59.71 69.13 61.73 66.67 61.65 70.54 65.07 72.93 65.50 79.39 71.64
Liu_CQUPT_task2_1 LiuCQUPT2021 64 53.837 44.15 50.38 69.57 63.60 56.40 50.79 64.43 56.04 51.71 51.22 52.20 51.21 51.47 50.79 64.52 63.77 62.12 60.36 61.40 60.98 67.10 66.22 62.62 61.69 66.62 64.69 51.32 51.08
Narita_AIT_task2_2 NaritaAIT2021 26 60.445 58.81 58.97 54.88 56.19 68.86 55.79 73.28 63.45 72.33 61.17 67.72 57.49 53.07 53.76 83.29 71.50 73.75 60.65 68.87 61.12 87.04 76.77 72.94 63.07 73.67 64.12 71.57 64.37
Deng_THU_task2_1 DengTHU2021 28 60.172 52.46 59.78 43.22 48.44 88.09 70.84 63.04 54.78 80.22 69.07 73.89 58.83 57.04 54.06 73.86 56.51 65.90 60.85 72.05 69.26 72.16 61.29 71.08 60.10 68.43 61.36 84.86 72.24
Li_CQUST_task2_1 LiCQUST2021 55 55.430 44.96 52.39 47.89 49.43 66.97 58.68 51.46 53.02 66.40 61.66 73.67 60.61 52.83 51.07 64.13 62.00 64.42 59.90 69.52 68.73 68.25 65.52 66.34 64.38 63.47 58.82 69.05 68.10
Chan_NTPU_task2_2 ChanNTPU2021 73 51.925 54.77 58.83 47.67 50.37 47.82 50.56 44.10 51.72 51.86 51.07 59.02 53.30 58.64 52.07 56.34 54.90 47.46 50.71 60.51 64.67 61.53 57.30 72.20 63.26 53.90 59.80 62.83 55.21
Zhang_NJUPT_task2_1 ZhangNJUPT2021 37 58.340 42.45 54.84 65.91 54.16 61.11 60.94 58.74 54.88 74.46 63.51 74.60 60.70 53.11 53.64 68.19 58.38 68.24 57.60 73.26 71.35 76.67 66.03 72.83 62.04 74.33 64.51 64.81 56.69
Wilkinghoff_FKIE_task2_3 WilkinghoffFKIE2021 4 64.201 69.83 63.41 57.50 54.23 88.98 70.20 57.75 50.17 74.82 66.64 86.84 66.93 62.74 53.63 81.41 68.56 77.88 61.09 74.62 67.18 75.99 55.94 77.04 63.02 80.97 64.20 81.10 64.93
Tan_NTU_task2_1 TanNTU2021 45 56.768 68.98 55.62 67.31 56.19 57.54 50.51 64.72 56.82 59.59 51.44 58.66 51.49 53.73 50.10 62.49 52.36 61.71 53.81 63.24 53.38 65.97 52.76 61.92 54.41 66.74 55.94 53.41 50.54
Zhou_PSH_task2_4 ZhouPSH2021 14 62.239 58.67 54.81 53.43 52.06 82.65 64.70 57.20 50.34 85.54 77.77 83.76 67.61 60.54 52.54 77.69 64.20 70.90 58.98 72.97 69.03 79.71 54.00 71.37 62.67 80.88 68.45 69.38 57.20
Wang_NTU_task2_4 WangNTU2021 41 57.420 63.94 51.99 62.93 54.86 73.10 60.22 66.46 58.43 56.53 51.73 60.39 53.06 49.28 50.87 70.54 57.10 66.48 54.79 62.27 62.85 72.47 57.56 65.21 56.04 67.41 58.21 57.42 54.42
Morita_SECOM_task2_3 MoritaSECOM2021 2 64.956 60.88 58.22 45.60 49.64 86.48 72.39 67.45 56.65 85.04 74.77 83.05 70.51 71.49 60.05 79.50 67.87 66.44 58.60 80.56 72.59 81.13 70.00 78.24 65.96 77.16 71.00 86.42 73.02
Lopez_IL_task2_4 LopezIL2021 1 66.798 75.27 59.71 69.15 59.91 61.01 60.79 63.07 61.56 86.76 81.55 83.18 63.60 65.36 60.15 87.45 78.37 77.56 70.48 81.22 80.25 86.13 76.35 79.58 67.90 82.87 69.25 90.32 77.24
Abe_RLB_task2_1 AbeRLB2021 62 54.307 54.68 51.64 58.68 51.97 58.66 53.55 63.33 54.51 55.64 52.28 56.98 51.74 49.21 50.75 57.49 54.02 61.57 50.70 56.06 51.17 54.30 52.17 59.16 55.89 65.18 54.41 52.32 51.39
He_XJU_task2_4 HeXJU2021 17 61.480 70.60 62.03 48.24 49.71 87.68 66.91 56.56 52.94 72.54 65.67 76.66 61.10 60.70 53.29 56.85 57.85 62.28 58.90 57.68 66.16 70.56 61.32 67.33 55.50 60.59 58.60 71.49 57.42
Cai_SMALLRICE_task2_2 CaiSMALLRICE2021 20 60.966 53.81 58.40 47.49 49.09 90.68 79.99 58.00 54.54 77.82 67.66 77.34 63.68 53.53 54.20 74.33 59.63 72.00 63.10 75.18 68.58 78.22 63.42 78.80 66.22 78.66 66.62 71.26 65.90
Sakamoto_Fixstars_task2_1 SakamotoFixstars2021 12 62.593 73.32 67.03 61.71 54.51 68.98 52.10 67.74 55.08 71.87 57.65 79.88 58.32 63.73 57.70 84.37 64.34 78.22 65.34 70.79 57.83 78.39 57.73 71.09 58.60 74.23 61.56 84.94 70.10
Wang_UCAS_task2_1 WangUCAS2021 48 56.509 66.67 55.59 66.51 56.71 59.18 50.76 65.04 56.85 57.48 51.05 58.30 51.62 52.17 50.38 65.94 53.43 67.26 55.19 62.60 53.42 66.61 52.83 62.18 54.76 66.78 56.18 54.46 50.51
Jalali_AIT_task2_1 JalaliAIT2021 58 54.983 44.06 53.99 45.94 48.53 51.41 50.10 57.61 53.95 77.20 66.80 80.28 61.85 49.09 53.18 58.91 53.94 69.22 59.80 60.35 63.30 65.46 58.48 68.55 61.08 71.31 62.24 76.76 66.97
Lu_UESTC_task2_3 LuUESTC2021 50 56.390 53.83 53.10 65.78 55.70 63.29 51.74 65.57 56.81 60.87 51.89 60.22 51.80 54.73 50.38 67.22 53.66 71.56 58.89 66.51 54.43 70.04 53.67 60.80 54.49 67.58 56.37 59.05 50.55
Yamashita_GifuUniv_task2_2 YamashitaGifuUniv2021 44 56.787 61.04 59.76 72.95 63.94 60.63 52.57 61.57 55.14 49.76 50.32 60.43 55.00 51.02 50.31 56.26 51.21 71.06 57.14 60.04 52.77 60.57 52.17 54.50 52.65 67.55 58.11 60.00 54.38
Primus_CPJKU_task2_4 PrimusCPJKU2021 27 60.221 55.71 54.10 51.97 50.97 90.22 71.19 59.68 54.49 74.71 67.17 75.13 60.05 49.75 53.74 79.12 63.61 65.63 61.62 78.72 74.39 74.86 60.31 79.99 64.68 76.34 65.62 77.52 65.90
Dini_TAU_task2_1 DiniTAU2021 68 53.226 54.45 55.37 60.35 51.66 55.55 52.44 56.14 49.74 53.28 50.66 56.60 53.56 48.26 49.69 62.94 51.24 53.98 52.14 62.59 52.05 69.93 50.75 60.58 52.26 69.29 56.28 56.96 50.10
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2021 9 63.745 61.70 55.58 61.79 57.07 66.60 66.17 62.53 54.92 74.60 61.69 86.27 74.12 62.36 60.05 84.31 69.10 79.15 64.84 79.68 77.03 84.09 67.73 77.15 67.22 84.03 73.22 94.42 81.03


Supplementary metrics (recall, precision, and F1 score)

Rank Submission Information Evaluation Dataset
Submission Code Technical
Report 		Official
Rank 		ToyCar
(F1 score) 		ToyCar
(Recall) 		ToyCar
(Precision) 		ToyTrain
(F1 score) 		ToyTrain
(Recall) 		ToyTrain
(Precision) 		Fan
(F1 score) 		Fan
(Recall) 		Fan
(Precision) 		Gearbox
(F1 score) 		Gearbox
(Recall) 		Gearbox
(Precision) 		Pump
(F1 score) 		Pump
(Recall) 		Pump
(Precision) 		Slider
(F1 score) 		Slider
(Recall) 		Slider
(Precision) 		Valve
(F1 score) 		Valve
(Recall) 		Valve
(Precision)
DCASE2021_baseline_task2_AE DCASE2021baseline2021 51 66.67 100.00 50.00 66.67 100.00 50.00 38.29 29.56 54.36 47.53 37.47 64.97 40.41 31.74 55.62 42.36 33.79 56.74 27.80 19.01 51.73
Tozicka_NSW_task2_4 TozickaNSW2021 19 66.67 100.00 50.00 49.82 47.43 52.47 58.20 53.28 64.12 42.46 29.69 74.55 65.84 62.56 69.48 56.90 51.70 63.25 33.94 24.62 54.63
Asai_PFU_task2_1 AsaiPFU2021 40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 17.52 10.73 47.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Bai_LFXS_task2_2 BaiLFXS2021 43 58.26 68.25 50.82 59.63 74.38 49.76 65.79 63.83 67.87 51.51 50.15 52.93 36.08 26.10 58.46 37.18 25.44 69.05 33.73 24.07 56.37
Liu_CQUPT_task2_1 LiuCQUPT2021 64 66.67 100.00 50.00 62.44 63.73 61.21 55.19 56.36 54.06 47.92 38.06 64.67 65.09 90.25 50.90 45.77 41.21 51.46 66.66 98.63 50.34
Narita_AIT_task2_2 NaritaAIT2021 26 0.00 0.00 0.00 64.44 92.98 49.31 61.25 60.99 61.52 68.20 74.17 63.12 64.26 61.37 67.43 68.39 75.49 62.50 49.60 47.65 51.73
Deng_THU_task2_1 DengTHU2021 28 51.06 47.76 54.85 46.18 46.33 46.02 81.37 81.73 81.01 54.56 48.98 61.58 53.58 38.95 85.82 60.16 51.13 73.05 52.45 48.14 57.61
Li_CQUST_task2_1 LiCQUST2021 55 58.32 64.46 53.24 58.10 70.13 49.60 45.64 36.39 61.20 6.77 3.61 54.90 37.32 24.27 80.71 6.78 3.54 81.11 0.00 0.00 0.00
Chan_NTPU_task2_2 ChanNTPU2021 73 57.14 66.67 50.00 66.67 100.00 50.00 65.48 95.25 49.89 58.92 79.39 46.84 64.02 86.31 50.88 65.59 86.59 52.79 65.37 92.22 50.63
Zhang_NJUPT_task2_1 ZhangNJUPT2021 37 0.00 0.00 0.00 65.73 83.05 54.39 6.15 3.21 71.84 40.85 31.24 59.02 60.89 52.71 72.07 32.11 20.05 80.46 23.17 14.35 60.06
Wilkinghoff_FKIE_task2_3 WilkinghoffFKIE2021 4 52.09 51.03 53.19 60.44 77.05 49.72 69.95 61.90 80.41 36.54 28.09 52.29 66.73 66.06 67.40 74.48 69.13 80.72 37.73 30.47 49.52
Tan_NTU_task2_1 TanNTU2021 45 66.67 100.00 50.00 67.08 72.75 62.22 54.91 52.78 57.22 47.29 36.31 67.80 39.76 30.78 56.14 48.45 42.06 57.13 31.71 22.69 52.64
Zhou_PSH_task2_4 ZhouPSH2021 14 66.70 100.00 50.04 67.03 99.31 50.58 73.25 92.90 60.46 65.72 80.80 55.38 69.48 92.83 55.52 72.22 96.05 57.86 67.73 88.72 54.77
Wang_NTU_task2_4 WangNTU2021 41 0.00 0.00 0.00 42.73 31.99 64.36 0.00 0.00 0.00 50.51 39.74 69.29 25.07 16.03 57.58 66.48 100.00 49.79 0.00 0.00 0.00
Morita_SECOM_task2_3 MoritaSECOM2021 2 54.74 55.04 54.43 62.02 80.54 50.42 51.24 37.23 82.19 60.24 55.36 66.08 73.83 67.22 81.89 0.00 0.00 0.00 62.49 60.79 64.29
Lopez_IL_task2_4 LopezIL2021 1 71.97 98.02 56.86 35.01 23.94 65.12 38.52 26.83 68.30 52.08 45.52 60.84 66.59 52.28 91.68 62.24 49.96 82.54 26.04 16.34 64.12
Abe_RLB_task2_1 AbeRLB2021 62 63.85 87.01 50.43 49.90 44.35 57.05 56.89 55.63 58.21 50.05 42.99 59.88 46.38 39.86 55.45 49.01 43.56 56.00 35.04 26.78 50.67
He_XJU_task2_4 HeXJU2021 17 66.67 100.00 50.00 63.72 86.85 50.31 47.20 34.75 73.56 53.58 56.24 51.16 57.79 55.09 60.76 44.58 31.54 76.03 47.27 42.97 52.52
Cai_SMALLRICE_task2_2 CaiSMALLRICE2021 20 56.47 55.82 57.13 46.59 45.33 47.92 77.13 66.14 92.49 51.64 47.08 57.19 64.98 54.62 80.21 64.46 55.95 76.04 36.02 26.30 57.15
Sakamoto_Fixstars_task2_1 SakamotoFixstars2021 12 64.57 65.45 63.71 55.83 56.50 55.17 60.96 56.09 66.77 58.00 51.99 65.57 63.21 58.18 69.19 73.09 72.55 73.64 61.09 60.29 61.91
Wang_UCAS_task2_1 WangUCAS2021 48 0.00 0.00 0.00 31.17 20.07 69.66 45.70 38.50 56.23 53.08 46.43 61.96 48.49 42.73 56.06 54.11 52.11 56.28 32.98 24.26 51.47
Jalali_AIT_task2_1 JalaliAIT2021 58 60.84 80.63 48.85 63.88 88.99 49.83 24.83 16.46 50.53 59.06 60.82 57.41 62.96 58.79 67.76 68.56 70.23 66.96 37.10 28.44 53.35
Lu_UESTC_task2_3 LuUESTC2021 50 67.53 100.00 50.98 66.88 95.70 51.40 68.17 82.83 57.92 64.84 73.75 57.86 63.20 73.51 55.43 67.05 85.96 54.96 58.42 64.62 53.31
Yamashita_GifuUniv_task2_2 YamashitaGifuUniv2021 44 47.33 38.28 61.99 59.32 52.11 68.86 51.00 46.91 55.88 52.93 47.56 59.66 42.02 36.86 48.86 54.30 49.47 60.16 44.80 39.24 52.19
Primus_CPJKU_task2_4 PrimusCPJKU2021 27 67.46 90.23 53.86 67.00 100.00 50.38 80.78 94.66 70.44 0.00 0.00 0.00 66.71 79.52 57.46 72.09 96.58 57.51 0.00 0.00 0.00
Dini_TAU_task2_1 DiniTAU2021 68 53.71 53.71 53.71 57.65 57.65 57.65 53.35 53.35 53.35 56.46 56.51 56.41 51.52 51.52 51.52 55.71 55.79 55.63 48.64 48.64 48.64
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2021 9 55.51 55.46 55.56 60.06 59.97 60.16 63.39 63.39 63.39 59.57 59.42 59.72 67.90 67.90 67.90 79.22 79.36 79.07 59.84 59.84 59.84



Systems ranking

Rank Submission Information Evaluation Dataset Development Dataset
Submission Code Technical
Report 		Official
Rank 		Official
Score 		ToyCar
(AUC) 		ToyCar
(pAUC) 		ToyTrain
(AUC) 		ToyTrain
(pAUC) 		Fan
(AUC) 		Fan
(pAUC) 		Gearbox
(AUC) 		Gearbox
(pAUC) 		Pump
(AUC) 		Pump
(pAUC) 		Slider
(AUC) 		Slider
(pAUC) 		Valve
(AUC) 		Valve
(pAUC) 		ToyCar
(AUC) 		ToyCar
(pAUC) 		ToyTrain
(AUC) 		ToyTrain
(pAUC) 		Fan
(AUC) 		Fan
(pAUC) 		Gearbox
(AUC) 		Gearbox
(pAUC) 		Pump
(AUC) 		Pump
(pAUC) 		Slider
(AUC) 		Slider
(pAUC) 		Valve
(AUC) 		Valve
(pAUC)
DCASE2021_baseline_task2_AE DCASE2021baseline2021 51 56.375 65.93 52.32 68.51 57.56 60.68 50.50 65.49 56.86 58.30 50.98 57.22 51.41 51.87 50.07 62.49 52.36 61.71 53.81 63.24 53.38 65.97 52.76 61.92 54.41 66.74 55.94 53.41 50.54
DCASE2021_baseline_task2_MNV2 DCASE2021baseline2021 59 54.770 42.73 55.93 42.91 50.55 64.96 58.14 51.14 51.59 67.97 60.28 72.92 60.71 53.13 51.52 56.04 56.37 57.46 51.61 61.56 63.02 66.70 59.16 61.89 57.37 59.26 56.00 56.51 52.64
Tozicka_NSW_task2_1 TozickaNSW2021 61 54.397 51.80 50.69 54.00 54.15 60.35 52.14 54.30 49.97 56.87 53.01 67.35 55.59 53.57 52.05
Tozicka_NSW_task2_2 TozickaNSW2021 39 58.110 50.22 50.32 51.71 51.79 73.17 62.02 58.86 51.09 78.65 65.16 69.89 59.61 53.93 54.22
Tozicka_NSW_task2_3 TozickaNSW2021 49 56.484 68.23 59.40 70.57 61.17 57.77 51.21 64.70 57.86 50.90 50.85 57.57 51.50 48.72 50.90 74.40 58.22 51.32 49.60 57.76 51.94 73.94 58.88 56.59 53.54 64.66 54.26 55.15 52.07
Tozicka_NSW_task2_4 TozickaNSW2021 19 61.186 68.23 59.40 51.71 51.79 73.17 62.02 64.70 57.86 78.65 65.16 69.89 59.61 53.93 54.22 74.40 58.22 73.94 58.88
Asai_PFU_task2_1 AsaiPFU2021 40 57.845 46.66 50.66 53.67 51.94 74.41 61.15 52.39 51.20 73.87 62.54 77.15 60.06 57.25 55.72 61.84 55.79 65.63 56.02 61.65 61.44 65.59 53.09 64.58 57.40 70.85 58.78 74.20 67.47
Bai_LFXS_task2_1 BaiLFXS2021 54 55.514 39.40 53.28 39.65 48.24 74.60 63.42 52.29 53.89 72.45 64.79 81.24 66.47 51.98 51.54 64.28 57.41 59.50 51.11 72.38 70.58 48.04 49.87 72.09 64.47 65.78 61.89 78.02 66.34
Bai_LFXS_task2_2 BaiLFXS2021 43 57.040 49.39 57.51 42.53 49.75 75.15 61.32 58.61 55.55 58.12 51.21 83.09 71.86 55.29 53.34 56.32 55.29 64.12 59.71 69.13 61.73 66.67 61.65 70.54 65.07 72.93 65.50 79.39 71.64
Bai_LFXS_task2_3 BaiLFXS2021 77 36.679 6.98 53.97 50.21 49.84 55.81 54.40 49.26 51.91 60.72 55.59 67.69 61.62 51.45 51.97 67.11 62.78 65.02 56.99 73.57 69.71 76.52 61.91 72.09 64.47 73.82 65.73 82.82 69.82
Bai_LFXS_task2_4 BaiLFXS2021 65 53.739 27.44 58.41 37.70 48.59 76.30 62.57 53.16 52.12 74.93 66.15 83.14 67.13 53.60 53.43 67.11 62.78 65.02 56.99 73.57 69.71 76.52 61.91 72.09 64.47 73.82 65.73 82.82 69.82
Liu_CQUPT_task2_1 LiuCQUPT2021 64 53.837 44.15 50.38 69.57 63.60 56.40 50.79 64.43 56.04 51.71 51.22 52.20 51.21 51.47 50.79 64.52 63.77 62.12 60.36 61.40 60.98 67.10 66.22 62.62 61.69 66.62 64.69 51.32 51.08
Narita_AIT_task2_1 NaritaAIT2021 32 59.548 59.32 58.16 59.42 55.56 61.26 53.59 75.35 64.54 67.97 56.89 66.96 55.81 53.17 53.98 81.56 68.34 73.44 61.96 66.85 58.79 83.40 74.94 68.97 60.65 71.17 61.46 67.66 62.95
Narita_AIT_task2_2 NaritaAIT2021 26 60.445 58.81 58.97 54.88 56.19 68.86 55.79 73.28 63.45 72.33 61.17 67.72 57.49 53.07 53.76 83.29 71.50 73.75 60.65 68.87 61.12 87.04 76.77 72.94 63.07 73.67 64.12 71.57 64.37
Deng_THU_task2_1 DengTHU2021 28 60.172 52.46 59.78 43.22 48.44 88.09 70.84 63.04 54.78 80.22 69.07 73.89 58.83 57.04 54.06 73.86 56.51 65.90 60.85 72.05 69.26 72.16 61.29 71.08 60.10 68.43 61.36 84.86 72.24
Li_CQUST_task2_1 LiCQUST2021 55 55.430 44.96 52.39 47.89 49.43 66.97 58.68 51.46 53.02 66.40 61.66 73.67 60.61 52.83 51.07 64.13 62.00 64.42 59.90 69.52 68.73 68.25 65.52 66.34 64.38 63.47 58.82 69.05 68.10
Chan_NTPU_task2_1 ChanNTPU2021 75 49.305 37.03 49.76 47.48 48.89 51.74 50.82 43.47 51.65 50.85 51.03 54.06 51.63 56.04 52.64 54.65 56.63 52.36 50.08 64.27 65.75 59.41 53.04 64.94 59.16 59.65 56.51 58.51 55.26
Chan_NTPU_task2_2 ChanNTPU2021 73 51.925 54.77 58.83 47.67 50.37 47.82 50.56 44.10 51.72 51.86 51.07 59.02 53.30 58.64 52.07 56.34 54.90 47.46 50.71 60.51 64.67 61.53 57.30 72.20 63.26 53.90 59.80 62.83 55.21
Chan_NTPU_task2_3 ChanNTPU2021 74 51.107 51.03 52.46 47.09 48.91 49.80 51.16 43.00 51.72 51.77 51.28 57.24 53.00 58.00 52.67 58.54 55.81 49.77 50.90 62.93 65.55 66.53 60.93 70.75 62.32 63.47 56.96 60.82 55.21
Chan_NTPU_task2_4 ChanNTPU2021 76 37.176 47.86 59.45 40.51 50.89 15.38 49.07 34.82 48.41 18.68 49.03 37.51 48.87 51.25 52.01 51.58 60.00 45.61 51.53 30.26 52.78 40.28 49.22 20.28 49.22 27.45 50.73 36.27 49.06
Zhang_NJUPT_task2_1 ZhangNJUPT2021 37 58.340 42.45 54.84 65.91 54.16 61.11 60.94 58.74 54.88 74.46 63.51 74.60 60.70 53.11 53.64 68.19 58.38 68.24 57.60 73.26 71.35 76.67 66.03 72.83 62.04 74.33 64.51 64.81 56.69
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2021 6 63.831 68.22 62.28 57.65 54.14 88.93 69.84 57.86 52.78 74.15 66.10 85.88 65.50 60.13 53.06 81.43 68.62 77.89 61.11 74.80 67.41 76.49 58.19 77.08 63.05 81.07 64.29 81.60 66.16
Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2021 7 63.793 68.22 62.28 57.65 54.14 88.77 70.57 56.92 49.52 74.15 66.10 85.88 65.50 63.49 54.20 81.43 68.62 77.89 61.11 74.19 66.37 73.32 53.81 77.08 63.05 81.07 64.29 80.02 62.11
Wilkinghoff_FKIE_task2_3 WilkinghoffFKIE2021 4 64.201 69.83 63.41 57.50 54.23 88.98 70.20 57.75 50.17 74.82 66.64 86.84 66.93 62.74 53.63 81.41 68.56 77.88 61.09 74.62 67.18 75.99 55.94 77.04 63.02 80.97 64.20 81.10 64.93
Wilkinghoff_FKIE_task2_4 WilkinghoffFKIE2021 8 63.747 68.15 62.29 57.64 54.45 88.91 69.87 57.88 52.83 74.24 66.13 85.91 65.51 58.21 53.37 81.31 68.50 78.03 61.27 74.83 67.40 76.49 58.26 77.06 63.04 81.10 64.22 78.10 66.40
Tan_NTU_task2_1 TanNTU2021 45 56.768 68.98 55.62 67.31 56.19 57.54 50.51 64.72 56.82 59.59 51.44 58.66 51.49 53.73 50.10 62.49 52.36 61.71 53.81 63.24 53.38 65.97 52.76 61.92 54.41 66.74 55.94 53.41 50.54
Zhou_PSH_task2_1 ZhouPSH2021 25 60.479 46.33 53.61 50.14 51.28 79.56 63.36 57.13 50.40 86.80 80.69 83.48 66.74 61.59 52.66 76.12 62.65 72.00 59.92 76.61 71.45 80.58 55.32 71.66 62.74 81.67 68.16 70.11 56.23
Zhou_PSH_task2_2 ZhouPSH2021 16 61.792 55.89 55.30 50.60 52.24 79.27 62.10 57.41 50.59 87.11 80.42 84.02 67.40 61.76 52.80 77.34 62.81 71.42 60.43 76.26 70.91 80.68 55.59 72.73 62.89 81.48 68.50 69.82 56.41
Zhou_PSH_task2_3 ZhouPSH2021 15 62.221 59.19 54.72 51.66 52.66 78.57 62.84 57.81 50.53 86.62 79.44 84.46 69.12 61.56 52.41 78.03 65.37 71.30 60.49 75.10 70.58 80.26 55.22 71.73 62.71 81.65 68.82 69.99 56.83
Zhou_PSH_task2_4 ZhouPSH2021 14 62.239 58.67 54.81 53.43 52.06 82.65 64.70 57.20 50.34 85.54 77.77 83.76 67.61 60.54 52.54 77.69 64.20 70.90 58.98 72.97 69.03 79.71 54.00 71.37 62.67 80.88 68.45 69.38 57.20
Wang_NTU_task2_1 WangNTU2021 46 56.650 66.44 53.40 64.98 56.55 59.84 50.77 66.46 58.43 56.53 51.73 61.23 53.40 50.20 50.72 70.54 56.77 64.97 52.67 62.81 52.04 72.47 57.56 65.21 56.04 66.36 58.47 57.94 50.63
Wang_NTU_task2_2 WangNTU2021 67 53.429 51.96 51.96 45.90 49.78 73.10 60.22 33.64 49.47 70.41 61.27 73.55 57.60 49.28 50.87 51.73 50.91 52.42 50.35 62.27 62.85 41.72 48.46 58.81 57.50 62.87 57.87 57.42 54.42
Wang_NTU_task2_3 WangNTU2021 47 56.552 63.17 55.78 62.93 54.86 62.27 50.83 66.83 58.66 58.62 52.33 60.07 52.41 49.55 50.35 67.66 54.16 66.48 54.79 64.29 53.75 71.64 55.63 64.31 55.33 67.57 57.77 57.10 50.23
Wang_NTU_task2_4 WangNTU2021 41 57.420 63.94 51.99 62.93 54.86 73.10 60.22 66.46 58.43 56.53 51.73 60.39 53.06 49.28 50.87 70.54 57.10 66.48 54.79 62.27 62.85 72.47 57.56 65.21 56.04 67.41 58.21 57.42 54.42
Morita_SECOM_task2_1 MoritaSECOM2021 5 64.128 58.47 57.41 43.99 49.84 83.82 71.64 70.12 59.01 85.86 75.19 77.65 65.07 71.47 59.83 72.54 65.90 64.60 58.20 80.93 71.47 79.52 68.35 79.05 64.37 67.45 62.85 76.57 66.15
Morita_SECOM_task2_2 MoritaSECOM2021 3 64.424 60.51 58.21 46.52 49.54 86.20 67.59 66.28 54.96 84.33 74.46 83.57 69.72 71.49 60.05 78.85 66.52 65.01 55.74 78.60 68.35 82.08 69.87 78.62 67.04 77.57 70.84 86.42 73.02
Morita_SECOM_task2_3 MoritaSECOM2021 2 64.956 60.88 58.22 45.60 49.64 86.48 72.39 67.45 56.65 85.04 74.77 83.05 70.51 71.49 60.05 79.50 67.87 66.44 58.60 80.56 72.59 81.13 70.00 78.24 65.96 77.16 71.00 86.42 73.02
Lopez_IL_task2_1 LopezIL2021 11 63.146 50.25 56.20 65.55 57.83 63.45 58.49 56.68 60.75 86.22 76.92 82.41 63.09 65.36 60.15 67.02 62.33 71.93 67.72 71.71 72.95 83.42 74.43 77.99 66.84 78.71 67.28 90.32 77.24
Lopez_IL_task2_2 LopezIL2021 52 56.324 25.79 55.20 52.70 54.63 61.01 60.79 58.82 52.19 85.37 78.48 84.29 65.72 63.29 55.41 58.43 56.29 66.41 56.96 81.22 80.25 71.56 59.64 75.43 65.06 71.84 62.39 72.97 62.06
Lopez_IL_task2_3 LopezIL2021 31 59.913 73.23 59.56 61.48 53.59 71.52 58.93 69.29 57.60 62.12 52.91 68.99 56.01 54.47 50.39 86.57 78.31 77.97 60.31 78.66 60.24 80.81 65.13 69.93 56.55 74.83 60.54 61.30 52.75
Lopez_IL_task2_4 LopezIL2021 1 66.798 75.27 59.71 69.15 59.91 61.01 60.79 63.07 61.56 86.76 81.55 83.18 63.60 65.36 60.15 87.45 78.37 77.56 70.48 81.22 80.25 86.13 76.35 79.58 67.90 82.87 69.25 90.32 77.24
Abe_RLB_task2_1 AbeRLB2021 62 54.307 54.68 51.64 58.68 51.97 58.66 53.55 63.33 54.51 55.64 52.28 56.98 51.74 49.21 50.75 57.49 54.02 61.57 50.70 56.06 51.17 54.30 52.17 59.16 55.89 65.18 54.41 52.32 51.39
Abe_RLB_task2_2 AbeRLB2021 72 52.030 54.81 53.07 49.52 49.58 46.52 50.12 55.82 52.83 51.78 50.98 59.66 54.89 51.03 50.43 62.13 57.58 63.18 56.55 62.26 57.15 66.30 55.44 66.60 62.74 73.19 61.87 51.72 50.47
Abe_RLB_task2_3 AbeRLB2021 70 52.413 36.24 54.67 66.46 56.13 36.68 49.82 61.37 57.16 59.97 53.50 68.43 56.12 51.98 49.49 64.83 58.68 63.89 55.71 55.18 54.68 60.75 54.23 62.89 57.78 69.97 58.43 53.02 50.84
He_XJU_task2_1 HeXJU2021 38 58.213 70.60 62.03 53.79 51.75 63.88 52.14 62.01 54.67 59.91 52.17 71.47 55.70 60.70 53.29 56.85 56.14 58.96 55.27 68.99 56.15 61.19 63.84 67.80 56.17 64.17 54.96 71.49 59.62
He_XJU_task2_2 HeXJU2021 36 58.912 46.95 51.31 49.65 53.75 79.72 69.47 57.29 53.71 79.87 66.71 76.66 61.10 52.01 52.78 60.15 53.96 62.28 58.90 57.67 66.16 70.56 61.32 67.33 57.52 63.96 58.60 66.07 59.72
He_XJU_task2_3 HeXJU2021 42 57.383 40.78 54.20 48.24 49.71 87.68 66.91 56.56 52.94 72.54 65.67 75.88 58.43 53.73 51.76 65.07 57.85 58.66 58.15 63.45 62.55 67.97 58.00 63.70 55.50 60.59 57.13 62.88 57.42
He_XJU_task2_4 HeXJU2021 17 61.480 70.60 62.03 48.24 49.71 87.68 66.91 56.56 52.94 72.54 65.67 76.66 61.10 60.70 53.29 56.85 57.85 62.28 58.90 57.68 66.16 70.56 61.32 67.33 55.50 60.59 58.60 71.49 57.42
Cai_SMALLRICE_task2_1 CaiSMALLRICE2021 29 60.149 55.03 56.42 49.09 50.04 85.98 73.42 58.25 54.72 76.65 67.72 69.17 60.92 53.53 54.20 74.33 59.63 72.00 63.10 75.18 68.58 78.22 63.42 78.80 66.22 78.66 66.62 71.26 65.90
Cai_SMALLRICE_task2_2 CaiSMALLRICE2021 20 60.966 53.81 58.40 47.49 49.09 90.68 79.99 58.00 54.54 77.82 67.66 77.34 63.68 53.53 54.20 74.33 59.63 72.00 63.10 75.18 68.58 78.22 63.42 78.80 66.22 78.66 66.62 71.26 65.90
Cai_SMALLRICE_task2_3 CaiSMALLRICE2021 22 60.867 53.31 58.05 48.97 49.22 89.14 77.63 60.67 54.70 77.21 65.44 75.47 63.27 53.53 54.20 74.33 59.63 72.00 63.10 75.18 68.58 78.22 63.42 78.80 66.22 78.66 66.62 71.26 65.90
Cai_SMALLRICE_task2_4 CaiSMALLRICE2021 21 60.874 52.13 58.28 47.76 48.95 90.68 79.65 59.02 54.79 78.19 67.33 77.04 63.53 53.53 54.20 74.33 59.63 72.00 63.10 75.18 68.58 78.22 63.42 78.80 66.22 78.66 66.62 71.26 65.90
Sakamoto_Fixstars_task2_1 SakamotoFixstars2021 12 62.593 73.32 67.03 61.71 54.51 68.98 52.10 67.74 55.08 71.87 57.65 79.88 58.32 63.73 57.70 84.37 64.34 78.22 65.34 70.79 57.83 78.39 57.73 71.09 58.60 74.23 61.56 84.94 70.10
Sakamoto_Fixstars_task2_2 SakamotoFixstars2021 24 60.527 63.45 55.22 58.02 53.28 68.49 52.05 67.74 55.08 69.94 56.46 79.88 58.32 63.73 57.70 80.46 62.54 73.41 64.02 70.33 56.73 78.39 57.73 71.09 58.20 74.23 61.56 84.94 70.10
Sakamoto_Fixstars_task2_3 SakamotoFixstars2021 23 60.810 73.32 67.03 57.77 52.95 68.98 52.10 67.14 54.98 71.87 57.65 77.40 54.55 58.52 52.36 84.37 64.34 77.89 61.61 70.79 57.83 78.03 56.78 71.09 58.60 72.68 60.04 76.55 56.86
Sakamoto_Fixstars_task2_4 SakamotoFixstars2021 18 61.308 68.75 67.66 64.89 56.18 69.03 52.42 68.07 57.68 61.64 52.56 70.12 55.54 65.62 57.22 83.11 61.18 77.47 64.33 69.41 54.82 76.94 59.66 70.00 57.69 72.77 59.90 74.72 63.75
Wang_UCAS_task2_1 WangUCAS2021 48 56.509 66.67 55.59 66.51 56.71 59.18 50.76 65.04 56.85 57.48 51.05 58.30 51.62 52.17 50.38 65.94 53.43 67.26 55.19 62.60 53.42 66.61 52.83 62.18 54.76 66.78 56.18 54.46 50.51
Wang_UCAS_task2_2 WangUCAS2021 63 54.092 47.96 51.09 67.99 57.69 57.13 50.50 64.03 56.06 55.08 50.45 54.81 51.53 50.25 49.83 61.09 51.88 61.65 53.69 61.59 51.69 64.40 53.37 60.15 53.40 64.23 54.36 52.43 50.66
Wang_UCAS_task2_3 WangUCAS2021 56 55.147 56.11 52.06 67.46 57.40 58.07 50.47 64.55 56.36 56.03 50.55 56.05 51.66 51.13 50.17 63.20 52.31 63.79 54.26 62.71 52.61 65.36 53.14 61.14 53.97 65.45 55.21 52.95 50.66
Wang_UCAS_task2_4 WangUCAS2021 71 52.176 31.62 50.85 70.89 62.52 56.69 50.52 64.64 56.96 53.23 50.52 53.96 51.47 50.42 49.76 59.49 52.83 62.01 53.08 60.41 52.16 65.75 53.06 58.84 53.64 62.53 53.31 49.38 50.12
Jalali_AIT_task2_1 JalaliAIT2021 58 54.983 44.06 53.99 45.94 48.53 51.41 50.10 57.61 53.95 77.20 66.80 80.28 61.85 49.09 53.18 58.91 53.94 69.22 59.80 60.35 63.30 65.46 58.48 68.55 61.08 71.31 62.24 76.76 66.97
Lu_UESTC_task2_1 LuUESTC2021 66 53.463 62.22 56.18 55.32 50.12 60.57 56.92 66.94 55.58 50.42 51.84 45.85 49.50 44.92 51.13 63.54 56.89 60.49 53.53 61.83 64.71 63.70 56.25 67.24 56.83 69.20 60.76 64.48 53.92
Lu_UESTC_task2_2 LuUESTC2021 57 55.015 48.50 52.44 69.80 55.82 59.49 51.51 63.98 54.94 58.25 50.79 58.65 51.96 51.86 49.86 66.04 53.98 64.18 53.86 62.76 53.48 65.38 52.64 62.43 54.75 66.49 55.14 56.50 50.33
Lu_UESTC_task2_3 LuUESTC2021 50 56.390 53.83 53.10 65.78 55.70 63.29 51.74 65.57 56.81 60.87 51.89 60.22 51.80 54.73 50.38 67.22 53.66 71.56 58.89 66.51 54.43 70.04 53.67 60.80 54.49 67.58 56.37 59.05 50.55
Lu_UESTC_task2_4 LuUESTC2021 53 56.124 51.93 50.84 68.54 60.97 60.22 51.12 65.60 57.20 59.69 51.61 58.93 51.91 54.61 50.09 66.60 57.78 67.55 57.78 64.40 53.14 68.40 53.87 61.48 54.37 67.80 56.05 57.37 50.26
Yamashita_GifuUniv_task2_1 YamashitaGifuUniv2021 60 54.764 58.59 57.58 65.18 55.87 62.33 52.37 61.42 54.24 48.55 50.09 55.65 50.91 49.79 50.38 62.87 52.68 65.06 53.67 61.48 53.95 59.04 51.62 58.12 53.69 63.62 53.99 54.39 50.84
Yamashita_GifuUniv_task2_2 YamashitaGifuUniv2021 44 56.787 61.04 59.76 72.95 63.94 60.63 52.57 61.57 55.14 49.76 50.32 60.43 55.00 51.02 50.31 56.26 51.21 71.06 57.14 60.04 52.77 60.57 52.17 54.50 52.65 67.55 58.11 60.00 54.38
Primus_CPJKU_task2_1 PrimusCPJKU2021 35 59.084 56.76 53.71 51.34 50.33 76.65 65.18 64.25 55.58 72.80 65.84 69.96 54.37 54.24 51.60 65.95 59.79 60.65 57.44 78.41 69.39 75.72 55.41 68.76 61.74 72.09 60.72 56.74 51.68
Primus_CPJKU_task2_2 PrimusCPJKU2021 34 59.196 58.70 55.78 47.39 50.36 78.52 67.09 61.39 54.46 75.11 66.99 72.01 54.00 54.60 51.74 66.94 57.82 59.14 57.28 78.58 70.21 75.74 55.75 70.72 63.86 70.06 60.76 56.26 51.86
Primus_CPJKU_task2_3 PrimusCPJKU2021 33 59.502 53.50 52.98 49.74 49.67 90.04 70.24 60.29 54.63 74.53 66.71 75.02 59.55 49.73 53.74 78.16 60.94 65.19 60.79 78.45 74.17 73.49 60.75 79.73 64.26 76.19 64.78 77.53 65.92
Primus_CPJKU_task2_4 PrimusCPJKU2021 27 60.221 55.71 54.10 51.97 50.97 90.22 71.19 59.68 54.49 74.71 67.17 75.13 60.05 49.75 53.74 79.12 63.61 65.63 61.62 78.72 74.39 74.86 60.31 79.99 64.68 76.34 65.62 77.52 65.90
Dini_TAU_task2_1 DiniTAU2021 68 53.226 54.45 55.37 60.35 51.66 55.55 52.44 56.14 49.74 53.28 50.66 56.60 53.56 48.26 49.69 62.94 51.24 53.98 52.14 62.59 52.05 69.93 50.75 60.58 52.26 69.29 56.28 56.96 50.10
Dini_TAU_task2_2 DiniTAU2021 69 52.722 53.02 54.37 58.59 51.04 56.00 50.98 54.68 49.52 51.51 51.07 54.97 52.87 50.56 50.53 62.94 51.24 53.98 52.14 62.59 52.05 69.93 50.75 60.58 52.26 69.29 56.28 56.96 50.10
Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2021 30 59.915 58.97 51.44 63.48 56.57 65.06 57.17 67.70 59.10 66.31 54.91 74.22 60.51 58.44 52.97 82.60 69.48 78.10 65.06 78.75 67.28 77.17 60.76 75.06 65.67 78.37 70.68 80.70 69.38
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2021 10 63.213 60.17 58.58 55.93 51.95 61.48 68.54 54.23 51.91 84.36 77.19 87.44 73.26 62.37 61.87 68.95 57.49 70.23 60.21 72.11 76.23 80.76 66.04 69.85 65.26 76.02 65.90 92.09 81.70
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2021 9 63.745 61.70 55.58 61.79 57.07 66.60 66.17 62.53 54.92 74.60 61.69 86.27 74.12 62.36 60.05 84.31 69.10 79.15 64.84 79.68 77.03 84.09 67.73 77.15 67.22 84.03 73.22 94.42 81.03
Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2021 13 62.263 58.97 51.44 63.48 56.57 63.25 61.37 60.72 55.57 76.55 65.30 77.87 66.51 62.37 61.87 82.60 69.48 78.10 65.06 81.25 71.80 82.31 66.51 75.17 66.07 78.37 73.44 92.09 81.70


Supplementary metrics (recall, precision, and F1 score)

Rank Submission Information Evaluation Dataset
Submission Code Technical
Report 		Official
Rank 		ToyCar
(F1 score) 		ToyCar
(Recall) 		ToyCar
(Precision) 		ToyTrain
(F1 score) 		ToyTrain
(Recall) 		ToyTrain
(Precision) 		Fan
(F1 score) 		Fan
(Recall) 		Fan
(Precision) 		Gearbox
(F1 score) 		Gearbox
(Recall) 		Gearbox
(Precision) 		Pump
(F1 score) 		Pump
(Recall) 		Pump
(Precision) 		Slider
(F1 score) 		Slider
(Recall) 		Slider
(Precision) 		Valve
(F1 score) 		Valve
(Recall) 		Valve
(Precision)
DCASE2021_baseline_task2_AE DCASE2021baseline2021 51 66.67 100.00 50.00 66.67 100.00 50.00 38.29 29.56 54.36 47.53 37.47 64.97 40.41 31.74 55.62 42.36 33.79 56.74 27.80 19.01 51.73
DCASE2021_baseline_task2_MNV2 DCASE2021baseline2021 59 50.59 45.85 56.41 0.00 0.00 0.00 38.41 28.26 59.93 15.67 9.34 48.61 5.95 3.07 92.31 41.65 29.87 68.74 0.00 0.00 0.00
Tozicka_NSW_task2_1 TozickaNSW2021 61 43.86 38.93 50.21 48.72 43.95 54.65 57.83 57.64 58.03 44.05 39.41 49.93 54.35 53.60 55.13 59.82 57.01 62.93 46.32 42.38 51.07
Tozicka_NSW_task2_2 TozickaNSW2021 39 0.00 0.00 0.00 49.82 47.43 52.47 58.20 53.28 64.12 52.55 49.89 55.51 65.84 62.56 69.48 56.90 51.70 63.25 33.94 24.62 54.63
Tozicka_NSW_task2_3 TozickaNSW2021 49 66.67 100.00 50.00 6.69 3.66 38.72 26.85 17.35 59.42 42.46 29.69 74.55 23.04 14.98 49.86 30.33 20.57 57.73 13.00 7.50 48.84
Tozicka_NSW_task2_4 TozickaNSW2021 19 66.67 100.00 50.00 49.82 47.43 52.47 58.20 53.28 64.12 42.46 29.69 74.55 65.84 62.56 69.48 56.90 51.70 63.25 33.94 24.62 54.63
Asai_PFU_task2_1 AsaiPFU2021 40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 17.52 10.73 47.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Bai_LFXS_task2_1 BaiLFXS2021 54 64.83 83.17 53.12 57.13 68.42 49.05 40.53 27.90 74.09 14.60 8.39 56.07 50.82 37.86 77.26 22.24 12.84 82.85 10.62 6.03 44.17
Bai_LFXS_task2_2 BaiLFXS2021 43 58.26 68.25 50.82 59.63 74.38 49.76 65.79 63.83 67.87 51.51 50.15 52.93 36.08 26.10 58.46 37.18 25.44 69.05 33.73 24.07 56.37
Bai_LFXS_task2_3 BaiLFXS2021 77 0.00 0.00 0.00 0.00 0.00 0.00 63.30 86.77 49.82 0.00 0.00 0.00 38.63 27.31 65.98 66.93 67.75 66.13 13.48 7.77 50.62
Bai_LFXS_task2_4 BaiLFXS2021 65 0.00 0.00 0.00 0.00 0.00 0.00 70.39 87.01 59.11 31.91 23.32 50.50 51.82 41.78 68.20 54.27 45.08 68.17 28.45 19.08 55.84
Liu_CQUPT_task2_1 LiuCQUPT2021 64 66.67 100.00 50.00 62.44 63.73 61.21 55.19 56.36 54.06 47.92 38.06 64.67 65.09 90.25 50.90 45.77 41.21 51.46 66.66 98.63 50.34
Narita_AIT_task2_1 NaritaAIT2021 32 0.00 0.00 0.00 65.75 88.23 52.40 64.56 79.66 54.27 64.70 59.00 71.61 60.94 59.78 62.15 68.11 78.39 60.21 51.90 51.39 52.41
Narita_AIT_task2_2 NaritaAIT2021 26 0.00 0.00 0.00 64.44 92.98 49.31 61.25 60.99 61.52 68.20 74.17 63.12 64.26 61.37 67.43 68.39 75.49 62.50 49.60 47.65 51.73
Deng_THU_task2_1 DengTHU2021 28 51.06 47.76 54.85 46.18 46.33 46.02 81.37 81.73 81.01 54.56 48.98 61.58 53.58 38.95 85.82 60.16 51.13 73.05 52.45 48.14 57.61
Li_CQUST_task2_1 LiCQUST2021 55 58.32 64.46 53.24 58.10 70.13 49.60 45.64 36.39 61.20 6.77 3.61 54.90 37.32 24.27 80.71 6.78 3.54 81.11 0.00 0.00 0.00
Chan_NTPU_task2_1 ChanNTPU2021 75 0.00 0.00 0.00 66.67 100.00 50.00 62.88 86.37 49.43 0.00 0.00 0.00 22.00 13.97 51.79 59.29 71.58 50.61 62.05 80.30 50.56
Chan_NTPU_task2_2 ChanNTPU2021 73 57.14 66.67 50.00 66.67 100.00 50.00 65.48 95.25 49.89 58.92 79.39 46.84 64.02 86.31 50.88 65.59 86.59 52.79 65.37 92.22 50.63
Chan_NTPU_task2_3 ChanNTPU2021 74 0.00 0.00 0.00 66.67 100.00 50.00 64.14 91.14 49.48 54.17 63.77 47.08 56.96 62.08 52.63 62.79 80.76 51.37 64.25 88.50 50.44
Chan_NTPU_task2_4 ChanNTPU2021 76 66.67 100.00 50.00 66.67 100.00 50.00 65.46 97.32 49.32 66.25 99.84 49.57 66.52 99.66 49.92 66.48 100.00 49.79 66.67 100.00 50.00
Zhang_NJUPT_task2_1 ZhangNJUPT2021 37 0.00 0.00 0.00 65.73 83.05 54.39 6.15 3.21 71.84 40.85 31.24 59.02 60.89 52.71 72.07 32.11 20.05 80.46 23.17 14.35 60.06
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2021 6 51.78 50.46 53.18 60.00 76.21 49.48 74.27 72.95 75.64 53.76 52.74 54.82 67.21 67.74 66.70 74.21 69.40 79.73 48.68 45.61 52.20
Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2021 7 51.78 50.46 53.18 60.00 76.21 49.48 66.39 56.53 80.41 9.39 5.32 40.12 67.21 67.74 66.70 74.21 69.40 79.73 17.43 11.47 36.24
Wilkinghoff_FKIE_task2_3 WilkinghoffFKIE2021 4 52.09 51.03 53.19 60.44 77.05 49.72 69.95 61.90 80.41 36.54 28.09 52.29 66.73 66.06 67.40 74.48 69.13 80.72 37.73 30.47 49.52
Wilkinghoff_FKIE_task2_4 WilkinghoffFKIE2021 8 52.84 52.85 52.83 59.80 75.93 49.32 75.39 75.50 75.29 53.18 51.89 54.54 67.14 67.74 66.55 74.54 69.52 80.33 43.07 36.58 52.37
Tan_NTU_task2_1 TanNTU2021 45 66.67 100.00 50.00 67.08 72.75 62.22 54.91 52.78 57.22 47.29 36.31 67.80 39.76 30.78 56.14 48.45 42.06 57.13 31.71 22.69 52.64
Zhou_PSH_task2_1 ZhouPSH2021 25 61.71 80.53 50.02 67.27 99.66 50.77 73.55 90.84 61.79 65.94 80.07 56.04 69.90 93.05 55.98 71.80 95.81 57.42 66.79 86.97 54.21
Zhou_PSH_task2_2 ZhouPSH2021 16 66.70 99.15 50.26 67.38 99.83 50.85 73.36 90.87 61.51 66.99 83.53 55.92 69.55 92.63 55.68 71.68 95.56 57.35 67.25 87.48 54.62
Zhou_PSH_task2_3 ZhouPSH2021 15 66.70 100.00 50.04 67.04 99.66 50.50 72.14 89.16 60.58 66.57 82.48 55.81 69.88 93.40 55.82 71.77 95.81 57.38 66.34 85.75 54.10
Zhou_PSH_task2_4 ZhouPSH2021 14 66.70 100.00 50.04 67.03 99.31 50.58 73.25 92.90 60.46 65.72 80.80 55.38 69.48 92.83 55.52 72.22 96.05 57.86 67.73 88.72 54.77
Wang_NTU_task2_1 WangNTU2021 46 0.00 0.00 0.00 65.78 76.93 57.45 16.71 9.78 57.26 50.51 39.74 69.29 25.07 16.03 57.58 61.71 60.83 62.61 33.52 25.26 49.80
Wang_NTU_task2_2 WangNTU2021 67 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Wang_NTU_task2_3 WangNTU2021 47 0.00 0.00 0.00 42.73 31.99 64.36 18.80 11.41 53.43 48.66 37.14 70.54 28.47 18.95 57.23 56.01 52.06 60.61 32.58 24.38 49.07
Wang_NTU_task2_4 WangNTU2021 41 0.00 0.00 0.00 42.73 31.99 64.36 0.00 0.00 0.00 50.51 39.74 69.29 25.07 16.03 57.58 66.48 100.00 49.79 0.00 0.00 0.00
Morita_SECOM_task2_1 MoritaSECOM2021 5 67.19 100.00 50.59 64.94 90.31 50.70 66.47 57.42 78.90 61.99 58.54 65.88 73.84 69.39 78.91 0.00 0.00 0.00 9.74 5.24 68.82
Morita_SECOM_task2_2 MoritaSECOM2021 3 54.61 53.24 56.05 62.77 77.60 52.70 51.21 37.57 80.38 57.39 51.22 65.25 69.41 59.00 84.27 0.00 0.00 0.00 62.49 60.79 64.29
Morita_SECOM_task2_3 MoritaSECOM2021 2 54.74 55.04 54.43 62.02 80.54 50.42 51.24 37.23 82.19 60.24 55.36 66.08 73.83 67.22 81.89 0.00 0.00 0.00 62.49 60.79 64.29
Lopez_IL_task2_1 LopezIL2021 11 63.47 81.90 51.81 38.67 27.67 64.20 34.14 23.53 62.20 57.87 70.09 49.28 65.45 51.43 89.97 52.99 39.10 82.18 26.04 16.34 64.12
Lopez_IL_task2_2 LopezIL2021 52 0.00 0.00 0.00 63.97 83.87 51.71 38.66 26.83 69.18 0.00 0.00 0.00 70.06 56.26 92.84 57.23 45.28 77.76 17.43 10.23 58.88
Lopez_IL_task2_3 LopezIL2021 31 0.00 0.00 0.00 31.37 22.03 54.44 68.63 99.32 52.43 43.20 30.65 73.20 0.00 0.00 0.00 35.01 23.10 72.29 11.08 6.21 51.06
Lopez_IL_task2_4 LopezIL2021 1 71.97 98.02 56.86 35.01 23.94 65.12 38.52 26.83 68.30 52.08 45.52 60.84 66.59 52.28 91.68 62.24 49.96 82.54 26.04 16.34 64.12
Abe_RLB_task2_1 AbeRLB2021 62 63.85 87.01 50.43 49.90 44.35 57.05 56.89 55.63 58.21 50.05 42.99 59.88 46.38 39.86 55.45 49.01 43.56 56.00 35.04 26.78 50.67
Abe_RLB_task2_2 AbeRLB2021 72 66.67 100.00 50.00 66.28 99.13 49.78 45.64 43.53 47.96 51.92 50.33 53.62 26.28 17.50 52.79 66.85 88.53 53.70 47.61 44.62 51.04
Abe_RLB_task2_3 AbeRLB2021 70 0.00 0.00 0.00 12.37 6.78 69.80 20.34 13.25 43.66 25.93 15.84 71.54 53.25 46.99 61.44 40.32 28.09 71.42 24.57 16.80 45.74
He_XJU_task2_1 HeXJU2021 38 66.67 100.00 50.00 66.29 99.15 49.79 30.83 20.34 63.66 59.52 63.83 55.75 43.22 33.84 59.78 72.79 89.22 61.47 47.27 42.97 52.52
He_XJU_task2_2 HeXJU2021 36 46.91 43.65 50.71 0.00 0.00 0.00 66.49 56.13 81.52 16.35 9.47 59.78 0.00 0.00 0.00 44.58 31.54 76.03 0.00 0.00 0.00
He_XJU_task2_3 HeXJU2021 42 66.42 94.32 51.26 63.72 86.85 50.31 47.20 34.75 73.56 53.58 56.24 51.16 57.79 55.09 60.76 72.74 87.55 62.21 48.98 45.33 53.28
He_XJU_task2_4 HeXJU2021 17 66.67 100.00 50.00 63.72 86.85 50.31 47.20 34.75 73.56 53.58 56.24 51.16 57.79 55.09 60.76 44.58 31.54 76.03 47.27 42.97 52.52
Cai_SMALLRICE_task2_1 CaiSMALLRICE2021 29 57.33 56.30 58.40 48.73 48.55 48.91 73.09 63.08 86.89 52.99 48.93 57.77 66.34 58.35 76.88 60.18 53.33 69.04 36.02 26.30 57.15
Cai_SMALLRICE_task2_2 CaiSMALLRICE2021 20 56.47 55.82 57.13 46.59 45.33 47.92 77.13 66.14 92.49 51.64 47.08 57.19 64.98 54.62 80.21 64.46 55.95 76.04 36.02 26.30 57.15
Cai_SMALLRICE_task2_3 CaiSMALLRICE2021 22 53.57 51.32 56.03 50.82 50.83 50.81 74.53 62.90 91.43 56.00 52.40 60.12 57.98 44.90 81.82 67.39 60.32 76.34 36.02 26.30 57.15
Cai_SMALLRICE_task2_4 CaiSMALLRICE2021 21 0.00 0.00 0.00 18.52 13.27 30.65 26.99 15.67 97.11 14.87 8.64 53.22 21.22 11.87 100.00 25.77 15.27 82.38 36.02 26.30 57.15
Sakamoto_Fixstars_task2_1 SakamotoFixstars2021 12 64.57 65.45 63.71 55.83 56.50 55.17 60.96 56.09 66.77 58.00 51.99 65.57 63.21 58.18 69.19 73.09 72.55 73.64 61.09 60.29 61.91
Sakamoto_Fixstars_task2_2 SakamotoFixstars2021 24 61.42 63.80 59.22 54.29 55.67 52.97 57.74 51.99 64.92 58.00 51.99 65.57 58.39 51.81 66.90 73.09 72.55 73.64 61.09 60.29 61.91
Sakamoto_Fixstars_task2_3 SakamotoFixstars2021 23 64.57 65.45 63.71 53.32 54.62 52.08 60.96 56.09 66.77 57.09 50.93 64.95 63.21 58.18 69.19 70.66 69.29 72.07 52.25 50.22 54.46
Sakamoto_Fixstars_task2_4 SakamotoFixstars2021 18 57.40 57.84 56.97 59.34 57.46 61.34 59.14 54.40 64.80 58.54 52.55 66.08 52.38 47.65 58.16 62.76 58.47 67.72 55.53 49.93 62.53
Wang_UCAS_task2_1 WangUCAS2021 48 0.00 0.00 0.00 31.17 20.07 69.66 45.70 38.50 56.23 53.08 46.43 61.96 48.49 42.73 56.06 54.11 52.11 56.28 32.98 24.26 51.47
Wang_UCAS_task2_2 WangUCAS2021 63 0.00 0.00 0.00 0.00 0.00 0.00 29.84 20.86 52.41 44.13 33.01 66.56 33.01 24.13 52.23 34.33 24.46 57.54 21.67 13.98 48.19
Wang_UCAS_task2_3 WangUCAS2021 56 0.00 0.00 0.00 17.92 10.16 75.66 34.67 25.89 52.47 46.61 36.82 63.49 38.98 30.54 53.86 41.28 32.65 56.11 26.03 17.67 49.45
Wang_UCAS_task2_4 WangUCAS2021 71 0.00 0.00 0.00 0.00 0.00 0.00 21.25 13.39 51.40 47.09 37.16 64.25 31.88 23.73 48.57 23.36 14.85 54.76 0.00 0.00 0.00
Jalali_AIT_task2_1 JalaliAIT2021 58 60.84 80.63 48.85 63.88 88.99 49.83 24.83 16.46 50.53 59.06 60.82 57.41 62.96 58.79 67.76 68.56 70.23 66.96 37.10 28.44 53.35
Lu_UESTC_task2_1 LuUESTC2021 66 66.67 100.00 50.00 53.98 53.93 54.04 37.25 26.97 60.20 0.00 0.00 0.00 62.83 85.57 49.64 66.48 100.00 49.79 5.60 3.05 34.39
Lu_UESTC_task2_2 LuUESTC2021 57 65.21 87.60 51.94 69.20 86.59 57.63 56.51 54.94 58.18 58.25 58.15 58.34 54.37 53.40 55.37 57.50 59.79 55.38 47.96 44.08 52.58
Lu_UESTC_task2_3 LuUESTC2021 50 67.53 100.00 50.98 66.88 95.70 51.40 68.17 82.83 57.92 64.84 73.75 57.86 63.20 73.51 55.43 67.05 85.96 54.96 58.42 64.62 53.31
Lu_UESTC_task2_4 LuUESTC2021 53 66.67 100.00 50.00 66.67 100.00 50.00 50.03 44.23 57.58 53.12 45.84 63.16 47.52 40.76 56.98 50.85 46.60 55.94 37.33 28.56 53.90
Yamashita_GifuUniv_task2_1 YamashitaGifuUniv2021 60 30.42 23.76 42.25 53.30 46.99 61.56 51.72 46.64 58.03 53.82 47.40 62.26 44.41 40.91 48.58 48.97 45.20 53.41 42.48 37.80 48.49
Yamashita_GifuUniv_task2_2 YamashitaGifuUniv2021 44 47.33 38.28 61.99 59.32 52.11 68.86 51.00 46.91 55.88 52.93 47.56 59.66 42.02 36.86 48.86 54.30 49.47 60.16 44.80 39.24 52.19
Primus_CPJKU_task2_1 PrimusCPJKU2021 35 66.67 100.00 50.00 67.57 96.39 52.01 64.30 58.96 70.71 56.43 53.14 60.16 57.43 51.70 64.59 65.88 66.97 64.82 42.50 35.57 52.78
Primus_CPJKU_task2_2 PrimusCPJKU2021 34 66.67 100.00 50.00 66.75 94.95 51.46 67.36 65.63 69.19 33.52 24.46 53.24 63.31 57.35 70.65 70.86 89.96 58.45 49.31 46.38 52.63
Primus_CPJKU_task2_3 PrimusCPJKU2021 33 66.90 87.64 54.10 67.04 100.00 50.42 80.78 94.66 70.44 0.00 0.00 0.00 66.69 79.52 57.42 72.04 96.41 57.50 0.00 0.00 0.00
Primus_CPJKU_task2_4 PrimusCPJKU2021 27 67.46 90.23 53.86 67.00 100.00 50.38 80.78 94.66 70.44 0.00 0.00 0.00 66.71 79.52 57.46 72.09 96.58 57.51 0.00 0.00 0.00
Dini_TAU_task2_1 DiniTAU2021 68 53.71 53.71 53.71 57.65 57.65 57.65 53.35 53.35 53.35 56.46 56.51 56.41 51.52 51.52 51.52 55.71 55.79 55.63 48.64 48.64 48.64
Dini_TAU_task2_2 DiniTAU2021 69 52.67 52.67 52.67 57.21 57.21 57.21 52.96 52.96 52.96 54.60 54.71 54.49 52.39 52.39 52.39 55.55 55.63 55.47 49.70 49.70 49.70
Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2021 30 54.97 54.97 54.97 62.49 62.43 62.54 61.28 61.28 61.28 61.84 61.75 61.94 61.21 61.21 61.21 68.60 68.72 68.49 56.96 56.96 56.96
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2021 10 55.07 55.07 55.07 55.02 55.02 55.02 61.67 61.67 61.67 54.83 54.69 54.97 78.24 78.24 78.24 81.94 82.09 81.78 61.32 61.32 61.32
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2021 9 55.51 55.46 55.56 60.06 59.97 60.16 63.39 63.39 63.39 59.57 59.42 59.72 67.90 67.90 67.90 79.22 79.36 79.07 59.84 59.84 59.84
Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2021 13 54.97 54.97 54.97 62.49 62.43 62.54 61.03 61.03 61.03 58.02 58.03 58.01 69.89 69.89 69.89 73.03 73.16 72.91 61.32 61.32 61.32


Domain-wise performance

Rank Submission Information Ranking Eveluation Dataset in Source Domain Eveluation Dataset in Target Domain
Submission Code Technical
Report 		Official
Rank 		Official
Score 		Harmonic mean
(AUC, source) 		ToyCar
(AUC, source) 		ToyCar
(pAUC, source) 		ToyTrain
(AUC, source) 		ToyTrain
(pAUC, source) 		Fan
(AUC, source) 		Fan
(pAUC, source) 		Gearbox
(AUC, source) 		Gearbox
(pAUC, source) 		Pump
(AUC, source) 		Pump
(pAUC, source) 		Slider
(AUC, source) 		Slider
(pAUC, source) 		Valve
(AUC, source) 		Valve
(pAUC, source) 		Harmonic mean
(AUC, target) 		ToyCar
(AUC, target) 		ToyCar
(pAUC, target) 		ToyTrain
(AUC, target) 		ToyTrain
(pAUC, target) 		Fan
(AUC, target) 		Fan
(pAUC, target) 		Gearbox
(AUC, target) 		Gearbox
(pAUC, target) 		Pump
(AUC, target) 		Pump
(pAUC, target) 		Slider
(AUC, target) 		Slider
(pAUC, target) 		Valve
(AUC, target) 		Valve
(pAUC, target)
DCASE2021_baseline_task2_AE DCASE2021baseline2021 51 56.375 64.76 76.33 51.26 69.89 55.49 66.58 51.36 67.81 55.71 62.75 51.18 64.13 50.91 51.56 50.89 57.03 58.02 53.42 67.18 59.78 55.74 49.68 63.32 58.06 54.43 50.79 51.65 51.92 52.19 49.27
DCASE2021_baseline_task2_MNV2 DCASE2021baseline2021 59 54.770 53.82 34.32 53.49 47.30 52.49 70.88 57.76 53.16 53.47 67.12 60.77 73.06 60.47 54.71 53.03 54.80 56.62 58.59 39.27 48.75 59.96 58.53 49.27 49.83 68.85 59.79 72.78 60.94 51.64 50.10
Tozicka_NSW_task2_1 TozickaNSW2021 61 54.397 58.68 58.73 49.80 56.63 52.66 62.19 51.67 56.25 50.06 56.35 54.08 69.20 57.93 53.90 53.17 54.46 46.34 51.61 51.60 55.72 58.61 52.62 52.47 49.87 57.39 51.98 65.60 53.43 53.26 50.98
Tozicka_NSW_task2_2 TozickaNSW2021 39 58.110 61.21 50.69 50.10 49.44 50.94 73.43 62.26 62.96 52.02 76.22 65.39 68.87 59.34 57.90 57.00 60.04 49.77 50.54 54.19 52.66 72.90 61.79 55.27 50.20 81.23 64.93 70.93 59.88 50.46 51.69
Tozicka_NSW_task2_3 TozickaNSW2021 49 56.484 61.24 73.69 59.71 73.42 60.75 60.45 51.76 64.14 55.39 55.89 51.50 63.17 51.54 47.06 51.55 56.46 63.52 59.08 67.92 61.59 55.32 50.68 65.26 60.56 46.73 50.21 52.87 51.46 50.51 50.27
Tozicka_NSW_task2_4 TozickaNSW2021 19 61.186 64.87 73.69 59.71 49.44 50.94 73.43 62.26 64.14 55.39 76.22 65.39 68.87 59.34 57.90 57.00 63.95 63.52 59.08 54.19 52.66 72.90 61.79 65.26 60.56 81.23 64.93 70.93 59.88 50.46 51.69
Asai_PFU_task2_1 AsaiPFU2021 40 57.845 64.93 50.00 51.91 60.77 54.14 72.33 63.12 55.46 50.58 72.75 64.91 87.08 63.36 70.05 59.63 55.86 43.73 49.47 48.05 49.91 76.62 59.30 49.65 51.82 75.01 60.34 69.26 57.09 48.40 52.29
Bai_LFXS_task2_1 BaiLFXS2021 54 55.514 58.26 40.40 51.63 47.88 47.65 74.05 63.27 55.26 54.05 73.80 65.71 79.91 67.25 59.33 53.92 51.17 38.45 55.04 33.84 48.84 75.15 63.56 49.62 53.73 71.14 63.90 82.60 65.72 46.25 49.36
Bai_LFXS_task2_2 BaiLFXS2021 43 57.040 59.32 44.16 56.29 48.42 48.52 79.83 63.12 62.52 55.32 59.77 51.47 88.67 76.68 55.03 53.94 56.04 56.03 58.79 37.92 51.05 70.98 59.62 55.16 55.79 56.56 50.95 78.17 67.61 55.56 52.74
Bai_LFXS_task2_3 BaiLFXS2021 77 36.679 54.36 35.14 58.72 48.42 48.52 64.35 59.29 54.13 53.80 67.08 62.98 83.59 74.59 53.28 53.45 18.66 3.88 49.93 52.13 51.23 49.28 50.26 45.20 50.15 55.47 49.76 56.87 52.49 49.74 50.57
Bai_LFXS_task2_4 BaiLFXS2021 65 53.739 45.67 18.23 55.29 40.56 47.37 76.77 62.51 54.51 50.97 75.87 68.62 85.43 71.73 57.15 57.16 56.19 55.43 61.91 35.21 49.88 75.83 62.64 51.87 53.32 74.02 63.85 80.97 63.09 50.46 50.16
Liu_CQUPT_task2_1 LiuCQUPT2021 64 53.837 57.30 48.45 52.81 71.64 61.29 55.99 50.58 66.01 54.68 55.80 51.36 57.11 50.69 52.30 51.27 52.14 40.55 48.17 67.61 66.10 56.81 50.99 62.93 57.48 48.18 51.09 48.06 51.74 50.67 50.33
Narita_AIT_task2_1 NaritaAIT2021 32 59.548 70.30 84.43 68.74 63.47 55.54 68.90 55.08 77.70 64.71 67.86 56.61 75.19 57.74 60.41 54.11 56.50 45.72 50.40 55.86 55.59 55.15 52.18 73.14 64.37 68.07 57.16 60.36 54.00 47.48 53.84
Narita_AIT_task2_2 NaritaAIT2021 26 60.445 70.48 80.68 65.60 60.57 56.79 73.48 57.05 75.45 63.25 72.61 61.68 75.25 59.42 60.63 53.90 57.22 46.27 53.55 50.17 55.59 64.78 54.58 71.23 63.66 72.05 60.67 61.56 55.67 47.19 53.62
Deng_THU_task2_1 DengTHU2021 28 60.172 61.87 44.32 52.43 43.56 47.46 88.42 69.51 66.84 55.00 76.16 67.85 77.20 58.59 65.49 55.90 62.08 64.28 69.52 42.87 49.46 87.76 72.22 59.64 54.56 84.73 70.32 70.86 59.07 50.52 52.35
Li_CQUST_task2_1 LiCQUST2021 55 55.430 56.05 38.81 51.17 47.58 49.09 73.17 58.51 54.70 54.73 67.46 62.16 74.54 61.05 55.82 51.83 55.90 53.42 53.66 48.21 49.76 61.74 58.85 48.58 51.42 65.38 61.17 72.82 60.17 50.14 50.34
Chan_NTPU_task2_1 ChanNTPU2021 75 49.305 45.90 35.25 48.63 48.78 49.44 47.39 49.66 41.20 51.63 49.51 50.92 54.22 51.36 50.94 51.39 49.90 39.01 50.94 46.25 48.36 56.97 52.04 46.01 51.67 52.26 51.15 53.90 51.91 62.26 53.96
Chan_NTPU_task2_2 ChanNTPU2021 73 51.925 49.23 52.36 60.62 49.26 50.38 43.78 50.78 39.05 51.14 51.47 50.95 59.61 52.14 55.11 51.09 53.84 57.41 57.15 46.18 50.35 52.68 50.34 50.66 52.30 52.25 51.19 58.44 54.51 62.65 53.08
Chan_NTPU_task2_3 ChanNTPU2021 74 51.107 47.51 45.15 48.56 48.90 48.94 44.76 50.29 38.55 51.11 49.75 51.50 56.92 51.58 53.27 52.03 54.24 58.68 57.04 45.41 48.88 56.12 52.06 48.61 52.35 53.95 51.06 57.56 54.50 63.65 53.33
Chan_NTPU_task2_4 ChanNTPU2021 76 37.176 25.37 47.49 61.11 38.54 50.88 9.78 48.21 28.53 48.10 24.80 49.63 35.36 48.64 43.50 49.75 34.63 48.25 57.88 42.69 50.90 35.99 49.97 44.66 48.73 14.99 48.44 39.94 49.09 62.34 54.49
Zhang_NJUPT_task2_1 ZhangNJUPT2021 37 58.340 63.75 43.95 52.59 69.88 51.55 73.71 62.71 62.09 56.08 76.54 64.94 73.56 59.67 60.96 56.76 55.66 41.05 57.29 62.36 57.05 52.19 59.28 55.74 53.73 72.49 62.14 75.66 61.76 47.05 50.85
Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2021 6 63.831 72.97 67.03 62.80 70.88 55.87 88.89 68.83 60.71 53.10 70.93 65.50 88.07 64.35 72.88 55.59 64.42 69.45 61.77 48.59 52.50 88.98 70.88 55.27 52.47 77.67 66.71 83.79 66.68 51.17 50.75
Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2021 7 63.793 72.97 67.03 62.80 70.88 55.87 89.03 70.65 60.36 50.23 70.93 65.50 88.07 64.35 73.24 56.23 65.11 69.45 61.77 48.59 52.50 88.51 70.50 53.85 48.83 77.67 66.71 83.79 66.68 56.03 52.30
Wilkinghoff_FKIE_task2_3 WilkinghoffFKIE2021 4 64.201 73.13 67.07 63.05 70.87 56.19 89.07 69.85 61.19 50.97 70.89 65.52 88.06 64.38 73.19 55.97 65.76 72.83 63.77 48.38 52.39 88.89 70.55 54.68 49.40 79.20 67.81 85.66 69.69 54.90 51.47
Wilkinghoff_FKIE_task2_4 WilkinghoffFKIE2021 8 63.747 72.17 66.89 62.85 70.90 56.69 88.84 68.81 60.76 53.22 71.10 65.53 88.10 64.36 67.54 56.29 64.41 69.47 61.74 48.56 52.39 88.99 70.95 55.27 52.45 77.66 66.74 83.83 66.70 51.14 50.75
Tan_NTU_task2_1 TanNTU2021 45 56.768 65.71 80.61 56.09 71.33 54.43 64.05 51.13 66.14 55.52 62.75 51.33 65.74 51.39 54.88 50.78 57.04 60.29 55.16 63.72 58.08 52.23 49.91 63.36 58.19 56.74 51.54 52.96 51.59 52.62 49.43
Zhou_PSH_task2_1 ZhouPSH2021 25 60.479 64.65 41.32 47.85 59.77 51.78 75.08 61.81 64.50 51.12 84.98 77.33 87.49 72.49 65.23 54.20 61.19 52.72 60.94 43.18 50.78 84.60 64.98 51.27 49.70 88.71 84.36 79.81 61.83 58.34 51.21
Zhou_PSH_task2_2 ZhouPSH2021 16 61.792 66.51 47.09 48.38 60.19 52.49 74.32 61.52 63.35 51.05 85.44 78.11 87.63 72.25 66.45 54.60 64.06 68.72 64.53 43.65 51.99 84.92 62.69 52.48 50.14 88.86 82.88 80.69 63.16 57.70 51.12
Zhou_PSH_task2_3 ZhouPSH2021 15 62.221 67.80 52.52 48.60 60.82 52.98 72.10 60.81 63.43 50.80 85.41 77.88 88.16 73.20 66.37 53.74 64.47 67.81 62.61 44.91 52.35 86.31 65.02 53.10 50.27 87.86 81.08 81.06 65.47 57.40 51.15
Zhou_PSH_task2_4 ZhouPSH2021 14 62.239 67.58 51.08 48.20 60.00 51.87 78.12 61.10 62.22 50.44 84.13 76.33 87.33 70.51 66.11 54.01 65.18 68.91 63.53 48.16 52.25 87.75 68.76 52.94 50.25 86.99 79.26 80.48 64.94 55.83 51.15
Wang_NTU_task2_1 WangNTU2021 46 56.650 64.89 71.28 51.95 68.34 58.16 66.34 51.28 68.43 57.68 63.99 51.89 72.68 54.98 49.67 51.01 56.26 62.21 54.93 61.93 55.02 54.50 50.28 64.60 59.21 50.63 51.58 52.89 51.90 50.75 50.44
Wang_NTU_task2_2 WangNTU2021 67 53.429 51.31 50.33 50.33 44.33 49.09 72.81 60.24 31.81 49.49 69.63 61.79 74.13 56.56 47.70 50.24 54.41 53.70 53.70 47.59 50.48 73.39 60.19 35.69 49.45 71.20 60.76 72.98 58.69 50.97 51.51
Wang_NTU_task2_3 WangNTU2021 47 56.552 64.08 65.05 51.92 67.47 56.10 68.71 51.40 68.87 58.66 65.28 52.58 70.59 52.94 48.81 51.00 56.44 61.40 60.26 58.97 53.67 56.93 50.26 64.90 58.66 53.19 52.08 52.28 51.90 50.32 49.72
Wang_NTU_task2_4 WangNTU2021 41 57.420 64.34 65.64 48.12 67.47 56.10 72.81 60.24 68.43 57.68 63.99 51.89 72.42 54.02 47.70 50.24 57.95 62.33 56.54 58.97 53.67 73.39 60.19 64.60 59.21 50.63 51.58 51.78 52.13 50.97 51.51
Morita_SECOM_task2_1 MoritaSECOM2021 5 64.128 71.94 56.47 53.08 52.23 49.60 89.53 76.27 74.48 59.61 83.15 72.07 86.25 71.22 81.74 65.72 62.62 60.61 62.50 38.00 50.09 78.80 67.53 66.24 58.42 88.75 78.59 70.61 59.90 63.49 54.92
Morita_SECOM_task2_2 MoritaSECOM2021 3 64.424 71.89 55.83 53.07 54.69 49.40 88.92 66.34 71.68 55.99 81.80 71.49 87.38 69.63 81.28 64.72 65.00 66.04 64.47 40.47 49.67 83.63 68.88 61.63 53.96 87.02 77.69 80.08 69.80 63.80 56.01
Morita_SECOM_task2_3 MoritaSECOM2021 2 64.956 71.89 56.47 53.08 52.23 49.60 89.53 76.27 74.48 59.61 83.15 72.07 86.25 71.22 81.28 64.72 65.00 66.04 64.47 40.47 49.67 83.63 68.88 61.63 53.96 87.02 77.69 80.08 69.80 63.80 56.01
Lopez_IL_task2_1 LopezIL2021 11 63.146 64.07 43.07 53.61 75.59 62.14 59.60 59.78 54.59 60.36 86.92 79.87 84.55 66.88 69.56 64.03 66.08 60.31 59.04 57.86 54.08 67.84 57.26 58.93 61.15 85.53 74.17 80.38 59.71 61.64 56.71
Lopez_IL_task2_2 LopezIL2021 52 56.324 46.65 16.93 53.68 60.01 52.81 51.45 61.70 60.15 54.45 86.90 81.49 84.35 62.97 67.10 54.90 63.08 54.11 56.81 46.99 56.57 74.93 59.91 57.55 50.11 83.89 75.67 84.23 68.73 59.88 55.94
Lopez_IL_task2_3 LopezIL2021 31 59.913 70.94 81.04 59.58 64.79 52.05 85.00 63.26 73.78 57.93 67.85 54.14 76.81 56.66 56.09 50.95 60.40 66.79 59.54 58.50 55.22 61.74 55.15 65.32 57.27 57.28 51.72 62.60 55.36 52.94 49.83
Lopez_IL_task2_4 LopezIL2021 1 66.798 71.66 81.44 59.05 77.56 62.21 51.45 61.70 63.52 61.38 88.72 82.19 85.56 66.10 69.56 64.03 70.00 69.97 60.39 62.38 57.78 74.93 59.91 62.62 61.75 84.88 80.91 80.92 61.28 61.64 56.71
Abe_RLB_task2_1 AbeRLB2021 62 54.307 61.32 53.35 50.85 70.55 51.85 71.80 58.27 67.30 56.61 60.69 54.08 60.93 51.97 51.14 51.25 52.29 56.07 52.45 50.23 52.10 49.59 49.54 59.80 52.55 51.36 50.61 53.51 51.52 47.43 50.26
Abe_RLB_task2_2 AbeRLB2021 72 52.030 51.62 46.33 48.71 48.35 49.43 50.31 50.93 54.22 51.66 52.08 51.57 59.54 54.84 52.57 50.57 53.26 67.08 58.29 50.75 49.72 43.27 49.33 57.52 54.05 51.49 50.40 59.78 54.94 49.59 50.30
Abe_RLB_task2_3 AbeRLB2021 70 52.413 57.19 47.11 55.31 68.39 53.27 44.73 50.39 61.31 57.08 68.48 56.57 69.10 56.65 53.14 49.45 46.35 29.45 54.04 64.64 59.32 31.08 49.27 61.42 57.24 53.34 50.75 67.78 55.59 50.87 49.54
He_XJU_task2_1 HeXJU2021 38 58.213 67.66 78.90 66.28 62.03 53.77 70.13 52.76 60.76 53.63 61.87 52.72 74.04 55.43 70.03 55.93 58.37 63.87 58.29 47.48 49.88 58.65 51.54 63.31 55.75 58.07 51.63 69.07 55.98 53.56 50.89
He_XJU_task2_2 HeXJU2021 36 58.912 64.18 45.73 51.65 59.32 53.82 82.53 70.09 60.87 54.72 77.86 68.71 81.62 64.05 59.89 53.76 56.72 48.23 50.97 42.69 53.69 77.10 68.85 54.10 52.75 81.99 64.83 72.26 58.41 45.96 51.83
He_XJU_task2_3 HeXJU2021 42 57.383 62.78 41.84 55.06 53.63 50.03 88.06 65.44 64.45 56.31 72.91 69.29 78.03 58.62 64.30 53.09 54.55 39.77 53.36 43.84 49.40 87.31 68.46 50.38 49.96 72.17 62.41 73.85 58.25 46.14 50.49
He_XJU_task2_4 HeXJU2021 17 61.480 71.11 78.90 66.28 53.63 50.03 88.06 65.44 64.45 56.31 72.91 69.29 81.62 64.05 70.03 55.93 60.28 63.87 58.29 43.84 49.40 87.31 68.46 50.38 49.96 72.17 62.41 72.26 58.41 53.56 50.89
Cai_SMALLRICE_task2_1 CaiSMALLRICE2021 29 60.149 62.24 47.53 49.03 46.46 48.42 88.31 76.56 61.24 56.12 72.39 66.39 78.97 64.65 63.43 55.76 61.09 65.35 66.44 52.05 51.78 83.77 70.53 55.54 53.38 81.45 69.10 61.53 57.60 46.30 52.73
Cai_SMALLRICE_task2_2 CaiSMALLRICE2021 20 60.966 63.34 49.62 53.85 45.59 48.13 92.67 83.07 61.08 55.10 74.36 68.27 82.96 66.24 63.43 55.76 61.28 58.78 63.79 49.56 50.10 88.76 77.12 55.21 53.98 81.61 67.06 72.44 61.31 46.30 52.73
Cai_SMALLRICE_task2_3 CaiSMALLRICE2021 22 60.867 64.45 50.33 53.03 47.87 48.32 92.05 79.43 64.91 56.00 74.24 64.94 80.96 65.29 63.43 55.76 60.92 56.65 64.11 50.13 50.17 86.41 75.91 56.96 53.46 80.43 65.94 70.68 61.36 46.30 52.73
Cai_SMALLRICE_task2_4 CaiSMALLRICE2021 21 60.874 63.32 47.79 53.56 46.09 47.78 92.93 82.68 62.67 55.33 74.83 67.59 82.80 66.24 63.43 55.76 61.11 57.34 63.91 49.54 50.18 88.54 76.83 55.77 54.26 81.86 67.07 72.02 61.02 46.30 52.73
Sakamoto_Fixstars_task2_1 SakamotoFixstars2021 12 62.593 73.25 81.76 71.53 66.24 53.74 70.93 53.54 68.64 55.29 71.53 58.14 83.63 58.47 73.37 63.10 65.48 66.46 63.06 57.75 55.29 67.13 50.73 66.87 54.88 72.20 57.18 76.45 58.18 56.33 53.15
Sakamoto_Fixstars_task2_2 SakamotoFixstars2021 24 60.527 69.86 59.88 53.90 66.24 53.74 69.84 52.88 68.64 55.29 71.82 57.45 83.63 58.47 73.37 63.10 63.90 67.48 56.61 51.62 52.82 67.19 51.24 66.87 54.88 68.15 55.51 76.45 58.18 56.33 53.15
Sakamoto_Fixstars_task2_3 SakamotoFixstars2021 23 60.810 70.22 81.76 71.53 59.32 52.02 70.93 53.54 67.92 54.83 71.53 58.14 80.71 54.77 64.91 53.70 64.32 66.46 63.06 56.30 53.91 67.13 50.73 66.39 55.14 72.20 57.18 74.36 54.33 53.28 51.09
Sakamoto_Fixstars_task2_4 SakamotoFixstars2021 18 61.308 73.02 81.76 71.53 74.03 57.11 70.93 53.54 69.48 56.95 64.69 53.50 77.47 57.18 75.35 61.93 61.49 59.31 64.18 57.75 55.29 67.22 51.35 66.72 58.44 58.87 51.65 64.05 54.00 58.12 53.18
Wang_UCAS_task2_1 WangUCAS2021 48 56.509 65.28 77.34 54.89 72.04 56.52 66.60 52.02 66.78 55.86 62.44 51.49 65.61 51.42 52.16 51.14 56.08 58.59 56.31 61.77 56.90 53.24 49.55 63.39 57.86 53.24 50.61 52.46 51.82 52.18 49.65
Wang_UCAS_task2_2 WangUCAS2021 63 54.092 59.20 54.33 48.10 72.96 57.16 62.34 51.03 64.97 54.24 57.72 50.32 59.45 51.37 48.63 50.49 53.15 42.93 54.48 63.66 58.22 52.73 49.99 63.11 58.01 52.67 50.58 50.84 51.70 51.99 49.19
Wang_UCAS_task2_3 WangUCAS2021 56 55.147 62.03 65.18 49.81 72.37 56.51 64.27 51.28 65.81 54.62 59.67 50.60 61.84 51.61 50.00 50.93 54.53 49.26 54.52 63.17 58.33 52.96 49.69 63.33 58.22 52.80 50.50 51.25 51.71 52.31 49.43
Wang_UCAS_task2_4 WangUCAS2021 71 52.176 55.29 38.87 47.65 73.10 62.24 60.69 50.98 65.09 54.60 55.46 50.43 57.54 51.38 50.14 50.52 48.14 26.64 54.52 68.81 62.81 53.18 50.07 64.19 59.54 51.17 50.61 50.80 51.57 50.71 49.03
Jalali_AIT_task2_1 JalaliAIT2021 58 54.983 55.50 36.07 50.65 51.20 48.47 50.25 50.16 58.14 53.55 72.73 66.50 79.62 61.34 64.71 55.64 54.69 56.60 57.80 41.66 48.60 52.62 50.04 57.10 54.35 82.26 67.09 80.94 62.36 39.54 50.93
Lu_UESTC_task2_1 LuUESTC2021 66 53.463 56.12 61.28 53.40 56.80 49.72 58.32 55.86 70.72 58.28 53.60 52.22 43.17 50.60 56.50 53.17 52.13 63.18 59.26 53.92 50.53 63.00 58.02 63.56 53.12 47.59 51.47 48.89 48.45 37.28 49.25
Lu_UESTC_task2_2 LuUESTC2021 57 55.015 60.56 52.62 51.33 71.86 56.06 63.33 51.95 65.28 54.35 59.59 50.78 62.86 52.85 53.02 50.74 55.47 44.98 53.60 67.85 55.57 56.09 51.08 62.72 55.55 56.96 50.80 54.97 51.10 50.75 49.01
Lu_UESTC_task2_3 LuUESTC2021 50 56.390 64.37 56.32 50.53 72.29 57.66 69.68 52.73 67.65 55.36 64.88 52.38 66.91 51.49 56.58 51.49 56.67 51.56 55.94 60.34 53.88 57.97 50.79 63.62 58.34 57.33 51.40 54.76 52.13 53.01 49.31
Lu_UESTC_task2_4 LuUESTC2021 53 56.124 64.27 57.92 47.37 73.14 60.18 67.29 51.92 67.18 56.13 63.94 51.82 67.14 51.57 56.50 50.78 55.32 47.07 54.86 64.49 61.79 54.49 50.35 64.09 58.31 55.96 51.40 52.51 52.26 52.84 49.42
Yamashita_GifuUniv_task2_1 YamashitaGifuUniv2021 60 54.764 58.09 53.64 58.32 79.81 62.06 66.03 51.56 64.86 54.55 47.02 49.96 60.02 51.00 48.01 50.57 55.43 64.55 56.85 55.08 50.81 59.03 53.19 58.32 53.94 50.18 50.21 51.87 50.83 51.70 50.18
Yamashita_GifuUniv_task2_2 YamashitaGifuUniv2021 44 56.787 59.72 55.32 60.04 82.66 67.12 63.29 51.73 65.65 56.85 48.38 50.10 66.03 57.64 49.52 50.43 57.89 68.09 59.49 65.28 61.04 58.19 53.44 57.97 53.53 51.23 50.55 55.71 52.59 52.61 50.18
Primus_CPJKU_task2_1 PrimusCPJKU2021 35 59.084 64.40 53.98 51.02 56.97 48.85 79.22 64.18 69.89 56.42 73.79 66.69 72.37 54.99 54.61 51.68 60.53 59.84 56.71 46.72 51.90 74.24 66.22 59.45 54.77 71.84 65.00 67.71 53.77 53.88 51.52
Primus_CPJKU_task2_2 PrimusCPJKU2021 34 59.196 65.33 65.07 58.72 50.90 50.05 80.66 66.04 67.31 54.84 75.48 67.81 73.92 56.72 55.22 51.32 59.22 53.47 53.13 44.33 50.67 76.49 68.17 56.42 54.08 74.74 66.19 70.18 51.54 54.00 52.17
Primus_CPJKU_task2_3 PrimusCPJKU2021 33 59.502 62.50 44.03 49.80 54.17 49.38 90.28 68.33 60.91 54.94 68.76 65.20 74.61 60.97 64.97 56.85 61.04 68.16 56.59 45.97 49.97 89.80 72.26 59.69 54.33 81.35 68.29 75.44 58.19 40.28 50.96
Primus_CPJKU_task2_4 PrimusCPJKU2021 27 60.221 63.74 47.13 51.29 55.95 48.96 90.44 69.36 61.13 55.08 68.88 65.40 74.61 61.04 64.97 56.84 61.49 68.11 57.25 48.52 53.15 90.00 73.12 58.30 53.91 81.62 69.05 75.67 59.09 40.31 50.95
Dini_TAU_task2_1 DiniTAU2021 68 53.226 55.57 55.66 54.36 64.05 51.42 56.34 52.56 57.63 49.73 52.04 49.65 60.45 54.56 46.43 49.59 53.92 53.30 56.43 57.06 51.91 54.78 52.33 54.71 49.75 54.58 51.71 53.21 52.59 50.25 49.79
Dini_TAU_task2_2 DiniTAU2021 69 52.722 54.26 52.90 53.42 59.53 51.09 56.42 50.20 56.31 49.49 51.23 50.70 56.85 53.28 48.30 50.27 53.88 53.14 55.35 57.69 50.98 55.58 51.78 53.14 49.54 51.80 51.45 53.21 52.46 53.04 50.79
Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2021 30 59.915 70.19 58.97 51.43 76.21 57.96 68.90 57.76 73.21 60.38 63.41 53.67 81.81 63.42 74.18 55.21 59.67 58.97 51.44 54.40 55.25 61.62 56.59 62.96 57.87 69.48 56.20 67.93 57.85 48.21 50.91
Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2021 10 63.213 67.79 62.40 66.51 59.20 51.11 64.96 67.02 58.63 53.44 85.22 77.90 91.21 78.22 65.80 66.66 61.51 58.09 52.34 53.00 52.81 58.35 70.14 50.43 50.46 83.52 76.50 83.98 68.89 59.28 57.72
Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2021 9 63.745 72.06 62.02 59.69 71.49 58.98 73.59 66.77 67.08 55.81 75.36 66.16 90.44 75.43 70.39 67.25 62.66 61.38 52.00 54.42 55.27 60.83 65.58 58.55 54.05 73.86 57.78 82.47 72.85 55.97 54.25
Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2021 13 62.263 69.70 58.97 51.43 76.21 57.96 64.96 67.02 58.63 53.44 85.22 77.90 91.21 78.22 65.80 66.66 61.71 58.97 51.44 54.40 55.25 61.62 56.59 62.96 57.87 69.48 56.20 67.93 57.85 59.28 57.72



System characteristics

Summary of the submitted system characteristics.

Rank Submission
Code 		Technical
Report 		Classifier System
Complexity 		Acoustic
Feature 		Data
Augmentation 		Decision Making System
Embeddings 		Subsystem
Conut 		External
Data Usage 		Front End
System
51 DCASE2021_baseline_task2_AE DCASE2021baseline2021 AE 269992 log-mel energies
59 DCASE2021_baseline_task2_MNV2 DCASE2021baseline2021 MobileNetV2 710067 log-mel energies
61 Tozicka_NSW_task2_1 TozickaNSW2021 KNN 1576834 log-mel energies Siamese Network
39 Tozicka_NSW_task2_2 TozickaNSW2021 KNN 1576834 log-mel energies Siamese Network
49 Tozicka_NSW_task2_3 TozickaNSW2021 AE 9053064 log-mel energies OpenL3
19 Tozicka_NSW_task2_4 TozickaNSW2021 AE, KNN 10629898 log-mel energies Siamese Network, OpenL3
40 Asai_PFU_task2_1 AsaiPFU2021 CNN, GMM, OCSVM 13405536 log-mel energies mixup
54 Bai_LFXS_task2_1 BaiLFXS2021 Transformer 1186755 log-mel energies mixup, spec augmentation
43 Bai_LFXS_task2_2 BaiLFXS2021 Transformer 2367555 spectrogram mixup, spec augmentation
77 Bai_LFXS_task2_3 BaiLFXS2021 Transformer, CNN 1186755 spectrogram, log-mel energies mixup, spec augmentation 7
65 Bai_LFXS_task2_4 BaiLFXS2021 Transformer, CNN 1186755 spectrogram, log-mel energies mixup, spec augmentation average 6
64 Liu_CQUPT_task2_1 LiuCQUPT2021 GAN 1166568 log-mel energies
32 Narita_AIT_task2_1 NaritaAIT2021 VAE, CenterLoss 128043983 log-mel energies PANNs ResNet38 pre-trained model Resampling(16kHz to 32kHz)
26 Narita_AIT_task2_2 NaritaAIT2021 VAE, CenterLoss 182304719 log-mel energies average PANNs ResNet38 2 pre-trained model Resampling(16kHz to 32kHz)
28 Deng_THU_task2_1 DengTHU2021 CNN 533000 log-mel energies mixup average
55 Li_CQUST_task2_1 LiCQUST2021 AE, ELM, GAN 894781 log-mel energies
75 Chan_NTPU_task2_1 ChanNTPU2021 MobileNetV2 710067 log-mel energies mixup
73 Chan_NTPU_task2_2 ChanNTPU2021 ResNet50V2 23570947 log-mel energies mixup
74 Chan_NTPU_task2_3 ChanNTPU2021 MobileNetV2, ResNet50V2 23642014 log-mel energies mixup average 2
76 Chan_NTPU_task2_4 ChanNTPU2021 MobileNetV2, ResNet50V2 24291004 log-mel energies mixup average 2
37 Zhang_NJUPT_task2_1 ZhangNJUPT2021 MobileNetV2, AE 2933855 log-mel energies Audio denoising, Mixup 2 train denoising model
6 Wilkinghoff_FKIE_task2_1 WilkinghoffFKIE2021 CNN, GMM, ensemble 1217299311 log-mel energies mixup sum, average 40
7 Wilkinghoff_FKIE_task2_2 WilkinghoffFKIE2021 CNN, GMM, ensemble 609044474 log-mel energies mixup sum 20
4 Wilkinghoff_FKIE_task2_3 WilkinghoffFKIE2021 CNN, GMM, ensemble 1217299311 log-mel energies mixup sum, average 40
8 Wilkinghoff_FKIE_task2_4 WilkinghoffFKIE2021 CNN, GMM, ensemble 1189557471 log-mel energies mixup sum, average 40
45 Tan_NTU_task2_1 TanNTU2021 AE 2,415,060 log-mel energies Guassian noise, frequency shifting Static classifier selection 3
25 Zhou_PSH_task2_1 ZhouPSH2021 CNN, ArcFace, ensemble 58316344 spectrogram mixup maximum 6
16 Zhou_PSH_task2_2 ZhouPSH2021 CNN, ArcFace, ensemble 58316344 spectrogram mixup average 6
15 Zhou_PSH_task2_3 ZhouPSH2021 CNN, ArcFace, ensemble 58316344 spectrogram mixup average 6
14 Zhou_PSH_task2_4 ZhouPSH2021 CNN, ArcFace, ensemble 58316344 spectrogram mixup average 6
46 Wang_NTU_task2_1 WangNTU2021 CAE 7916544 log-mel energies
67 Wang_NTU_task2_2 WangNTU2021 CRNN 663783 log-mel energies
47 Wang_NTU_task2_3 WangNTU2021 CAE 7719648 log-mel energies
41 Wang_NTU_task2_4 WangNTU2021 CAE, CRNN 31989511 log-mel energies
5 Morita_SECOM_task2_1 MoritaSECOM2021 CNN, LOF 994048 spectrogram
3 Morita_SECOM_task2_2 MoritaSECOM2021 CNN, k-NN 994048 spectrogram
2 Morita_SECOM_task2_3 MoritaSECOM2021 CNN, LOF, k-NN 994048 spectrogram
11 Lopez_IL_task2_1 LopezIL2021 classifier 3341026 STFT, log-mel energies
52 Lopez_IL_task2_2 LopezIL2021 classifier 1889899 raw waveform Teager-Kaiser
31 Lopez_IL_task2_3 LopezIL2021 normalizing flow 6574848 log-mel energies
1 Lopez_IL_task2_4 LopezIL2021 ensemble, normalizing flow, classifier 11805773 log-mel energies Teager-Kaiser
62 Abe_RLB_task2_1 AbeRLB2021 Fast Autoregressive Transformer 37897472 log-mel energies Frame sequence transformation
72 Abe_RLB_task2_2 AbeRLB2021 Fast Autoregressive Transformer 25539968 log-mel energies
70 Abe_RLB_task2_3 AbeRLB2021 Fast Autoregressive Transformer 25539968 log-mel energies
38 He_XJU_task2_1 HeXJU2021 PCA 220000 log-mel energies
36 He_XJU_task2_2 HeXJU2021 CNN 2000000 log-mel energies
42 He_XJU_task2_3 HeXJU2021 CNN 571000 log-mel energies
17 He_XJU_task2_4 HeXJU2021 PCA, CNN 2791000 log-mel energies 3
29 Cai_SMALLRICE_task2_1 CaiSMALLRICE2021 AE, CNN 4282586 log-mel energies mixup, median-filtering, timeshift, timemask pre-trained model
20 Cai_SMALLRICE_task2_2 CaiSMALLRICE2021 AE, CNN 4282586 log-mel energies mixup, median-filtering, timeshift, timemask average 5 pre-trained model
22 Cai_SMALLRICE_task2_3 CaiSMALLRICE2021 AE, CNN 4282586 log-mel energies mixup, median-filtering, timeshift, timemask maximum pre-trained model
21 Cai_SMALLRICE_task2_4 CaiSMALLRICE2021 AE, CNN 4282586 log-mel energies mixup, median-filtering, timeshift, timemask average pre-trained model
12 Sakamoto_Fixstars_task2_1 SakamotoFixstars2021 Mahalanobis distance, Section ID classification, IDNN 301315 log-mel energies 3
24 Sakamoto_Fixstars_task2_2 SakamotoFixstars2021 Mahalanobis distance, Section ID classification, IDNN 301315 log-mel energies 3
23 Sakamoto_Fixstars_task2_3 SakamotoFixstars2021 Mahalanobis distance, Section ID classification, IDNN 301315 log-mel energies 3
18 Sakamoto_Fixstars_task2_4 SakamotoFixstars2021 Mahalanobis distance, IDNN 184320 log-mel energies 2
48 Wang_UCAS_task2_1 WangUCAS2021 AE 1,854,376 log-mel energies
63 Wang_UCAS_task2_2 WangUCAS2021 VAE 1,860,008 log-mel energies
56 Wang_UCAS_task2_3 WangUCAS2021 AE,VAE 3,714,384 log-mel energies average
71 Wang_UCAS_task2_4 WangUCAS2021 IAE,IAVE 630,072 log-mel energies average
58 Jalali_AIT_task2_1 JalaliAIT2021 LeNet 391302 log-mel energies mixup
66 Lu_UESTC_task2_1 LuUESTC2021 CNN 1675971 STFT, log-mel energies average 2 STFT cut
57 Lu_UESTC_task2_2 LuUESTC2021 VAE 186528 log-mel energies
50 Lu_UESTC_task2_3 LuUESTC2021 CNN 989692 log-mel energies
53 Lu_UESTC_task2_4 LuUESTC2021 AE 185496 log-mel energies
60 Yamashita_GifuUniv_task2_1 YamashitaGifuUniv2021 VAE, CNN 1254167 log-mel energies
44 Yamashita_GifuUniv_task2_2 YamashitaGifuUniv2021 VAE, CNN 1254167 log-mel energies
35 Primus_CPJKU_task2_1 PrimusCPJKU2021 MADE 144900000 log-mel energies
34 Primus_CPJKU_task2_2 PrimusCPJKU2021 MAF 228200000 log-mel energies
33 Primus_CPJKU_task2_3 PrimusCPJKU2021 ResNet 46200000 log-mel energies
27 Primus_CPJKU_task2_4 PrimusCPJKU2021 Ensemble 421372000 log-mel energies average 3
68 Dini_TAU_task2_1 DiniTAU2021 GAN 15224193 log-mel energies
69 Dini_TAU_task2_2 DiniTAU2021 GAN 15224193 log-mel energies
30 Kuroyanagi_NU-HDL_task2_1 KuroyanagiNU-HDL2021 AE, Conformer, GMM, ID regression 400000000 log-mel energies SpecAugment average, median, maximum, raking 10
10 Kuroyanagi_NU-HDL_task2_2 KuroyanagiNU-HDL2021 ensemble, CNN, ArcFace,binary classification 71127950 log-mel energies mixup, gaussian noise, volume perturbation average ResNet34, ResNeXt50, efficientnet-b3 6 pre-trained model
9 Kuroyanagi_NU-HDL_task2_3 KuroyanagiNU-HDL2021 AE, Conformer, GMM, ID regression, ensemble, CNN, ArcFace, binary classification 471127950 log-mel energies SpecAugment, mixup, gaussian noise, volume perturbation average, median, maximum, raking ResNet34, ResNeXt50, efficientnet-b3 18 pre-trained model
13 Kuroyanagi_NU-HDL_task2_4 KuroyanagiNU-HDL2021 AE, Conformer, GMM, ID regression, ensemble, CNN, ArcFace, binary classification 471127950 log-mel energies SpecAugment, mixup, gaussian noise, volume perturbation average, median, maximum, raking ResNet34, ResNeXt50, efficientnet-b3 18 pre-trained model



Technical reports

ANOMALOUS SOUND DETECTION BY AUTO REGRESSIVE FRAME SEQUENCE MODEL

Yoshiharu Abe
Ralabo, Yokohama, Kanagawa, Japan

Abstract

The normal sound frame sequence is modeled by a base module. This base module inputs a partially masked frame sequence and predicts the masked part of the frame sequence. The anomaly score is calculated as the difference between the predicted and actual frames of the masked area. The Transformer[1] is used as the sequence model in the base module. The base module is trained with a large amount of normal sound from the source domain. A front-end module is added in front of the base module to cope with environmental changes in the target domain. The front-end module, consisted of Transformer[1], transforms a target domain frame sequence into a source domain frame sequence. The front-end module is trained with a small amount of normal sound from the target domain. The AUC for audio clips in the target domain was 51.11% for the domain-dependent model (with base and front-end modules), and 61.44% for the domain-independent model (with base module). Further investigation would be needed to determine why the performance of the domain-dependent model is lower than that of the domain-independent model.

System characteristics
Classifier Fast Autoregressive Transformer
System complexity 25539968, 37897472 parameters
Acoustic features log-mel energies
Front end system Frame sequence transformation
PDF
Source code

Sub-Cluster AdaCos based unsupervised anomalous sound detection for machine condition monitoring under domain shift conditions

Yudai Asai
PFU Limited., Kanagawa, Japan

Abstract

This technical describes our approaches for the DCASE 2021 Challenge Task 2. Our approaches are based on deep metric learning using sub-cluster AdaCos loss and outlier detection using GMM and One-Class SVM. To tackle the difficulties of domain shift conditions, first we trained our model with only source domain data, and then, fine-tuned with source and target domain data. We achieved an averaged area under the curve (AUC) of 66.12% and averaged partial AUC (p = 0.1) of 58.18% on the test data in development dataset.

System characteristics
Classifier CNN, GMM, OCSVM
System complexity 13405536 parameters
Acoustic features log-mel energies
Data augmentation mixup
PDF

DPTRANS: DUAL-PATH TRANSFORMER FOR MACHINE CONDITION MONITORING

Jisheng Bai, Zejian Wang, Mou Wang, and Jianfeng Chen
LianFeng Acoustic Technologies Co., Ltd., Xi'an, China and School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an, China

Abstract

Anomaly detection has a wide range of application scenarios in industry such as finding fraud cases in financial industry or finding network intrusion in network security. Finding anomaly condition of machines in factories can prevent causing damage. Previous works mainly focus on finding local and deep features from spectrograms of anomaly sounds. Most importantly, deep features are always obtained after deep convolutional and pooling layers. However, the details of spectrogram, which present potential anomaly information, may be lost by these operations. In this paper, we introduce DPTrans, a novel dual-path Transformer-based neural network for DCASE 2021 challenge Task2 (Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions). DPTrans learns temporal and frequency dependencies through self-attention blocks, and achieves great performance. Moreover, DPTrans takes advantages of Transformer, which provide faster training speed and less GPU demand than comparative methods. Finally, we take different settings of Transformer train several models and make a fusion of them.

System characteristics
Classifier CNN, Transformer
System complexity 1186755, 2367555 parameters
Acoustic features log-mel energies, spectrogram
Data augmentation mixup, spec augmentation
Decision making average
Subsystem count 6, 7
PDF

THE SMALL RICE CAMERA READY SUBMISSION TO THE DCASE2021

Xinyu Cai, Heinrich Dinkel, Zhiyong Yan, Yongqing Wang, Junbo Zhang, Zhiyong Wu, Yujun Wang
Technology Comittee, Beijing, China

Abstract

This paper describes our submission to the DCASE 2021 Task 2 challenge. The objective is identifying whether the sound emitted from a machine is normal or anomalous without having access to large amounts of anomalous samples. Our anomaly score calculator system is a combination of two models: i) AutoEncoder-based unsupervised training and ii) EfficientNet-based supervised model. To alleviate the problem of domain shift, we train the models with contrastive loss and hard example mining manner, which leads to a substantial improvement with regards to the main omega evaluation metric. Further we investigate the use of median-filtering, timemasking, time shifting and mixup augmentation for this task, which further boosts performance. Our best single model submission achieves an official omega score of 71.72, 70.05, 72.14, 67.26, 66.17, 71.97, 68.47 for Fan, Gearbox, Slider, Toy Train, Toy Car, Pump, Valve on the development dataset, respectively.

System characteristics
Classifier AE, CNN
System complexity 4282586 parameters
Acoustic features log-mel energies
Data augmentation mixup, median-filtering, timeshift, timemask
Decision making average, maximum
Subsystem count 5
External data usage pre-trained model
PDF

AN ENSEMBLE APPROACH FOR ABNORMAL SOUND DETECTION WITH DATA AUGMENTATION

Bo Cheng Chan, and Chung Li Lu
National Taipei University, Taipei, Taiwan and National Taiwan University, Taipei, Taiwan

Abstract

In this paper, we present the task description and discuss the results of DCASE 2021 Challenge Task 2: Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions. The task is identifying whether the sound emitted from a machine is normal or anomalous in test dataset. The training dataset does not contain any abnormal machine sounds. Our approached is based on MobileNetV2 and ResNetV2-50 with data augmentation mix up to identify abnormal sounds in each machine.

System characteristics
Classifier MobileNetV2, ResNet50V2
System complexity 23570947, 23642014, 24291004, 710067 parameters
Acoustic features log-mel energies
Data augmentation mixup
Decision making average
Subsystem count 2
PDF
Source code

Description and discussion on DCASE 2021 challenge task2: unsupervised anomalous sound detection for machine condition monitoring under domain shifted conditions

Yohei Kawaguchi, Keisuke Imoto, Yuma Koizumi, Noboru Harada, Daisuke Niizumi, Kota Dohi, Ryo Tanabe, Harsh Purohit, and Takashi Endo
Research and Development Group, Hitachi, Ltd., Tokyo, Japan and Doshisha University, Kyoto, Japan and Google LLC, Tokyo, Japan

Abstract

We present the task description and discussion on the results of the DCASE 2021 Challenge Task 2. Last year, we organized unsupervised anomalous sound detection (ASD) task; identifying whether the given sound is normal or anomalous without anomalous training data. In this year, we organize an advanced unsupervised ASD task extit{under domain-shift conditions} which focuses on the inevitable problem for the practical use of ASD systems. The main challenge of this task is to detect unknown anomalous sounds where the acoustic characteristics of the training and testing samples are different, i.e. domain-shifted. This problem is frequently occurs due to changes in seasons, manufactured products, and/or environmental noise. After the challenge submission deadline, we will add challenge results and analysis of the submissions.

System characteristics
Classifier AE, MobileNetV2
System complexity 269992, 710067 parameters
Acoustic features log-mel energies
PDF
Source code Source code

AITHU SYSTEM FOR UNSUPERVISED ANOMALOUS SOUND DETECTION

Yufeng Deng, Jia Liu, Jitao Ma, Xuchu Chen, Cheng Lu, Ruhang Xu, and Wei-Qiang Zhang
Electronic Engineering, Tsinghua University, Beijing, China and Tsinghua AI Plus, Beijing, China and North China Electric Power University, Beijing, China

Abstract

This report describes the AITHU system for Task 2 of the DCASE 2021 challenge, Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions. The task aims to detect audio recordings containing anomalous machine sounds in a test set, when the training dataset itself does not contain any examples of anomalies. Moreover, the task is performed under the conditions that the acoustic characteristics of the training data and the test data are different (i.e., domain mismatch). We perform weighted mixing of data in different sections instead of to distinguish the data in the same part of different fields, and train a neural network to recognize mixed weights. The results of our approach are better than baseline systems for all machine types. In the development set, the official score of our approach is 67.12%.

System characteristics
Classifier CNN
System complexity 533000 parameters
Acoustic features log-mel energies
Data augmentation mixup
Decision making average
PDF

UNSUPERVISED DETECTION OF ANOMALOUS SOUND FOR MACHINE MONITORING UNDER DOMAIN SHIFTED CONDITION BASED ON GANS AND AUTOENCODERS

Amirhossein Hassankhani, Afshin Dini, and Konstantinos Drossos
Tampere University, Tampere, Finland

Abstract

This report presents an unsupervised method for detecting anomalous industrial machine sounds, taken under two different conditions and shifted domains, and submitted to DCASE 2021 Task 2. The method tries to map the distribution of data into a learned latent space, using a reconstructive autoencoder followed by an additional second encoder. Furthermore, the method employs a discriminator trying to differentiate between the input and the reconstructed audio to and from the autoencoder. All components are jointly optimized, using a sum of weighted losses and utilizing an adversarial setting between the autoencoder and the discriminator. Anomaly is detected through the distance between the output of the two encoders. Obtained results show that the method performs better than the provided baseline in some cases.

System characteristics
Classifier GAN
System complexity 15224193 parameters
Acoustic features log-mel energies
PDF
Source code

Several Approaches For Anomaly Detection From Sound

Yaoguang Wang,Yaohao Zheng, Yunxiang Zhang, Ying Hu, Minqiang Xu, and Liang He
Department of Electronic Engineering, Tsinghua University, Bejing, China and School of information science and engineering, Xinjiang University, Xinjiang, China

Abstract

The task2 of IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Events mainly research unsupervised anomalous sound detection for machine condition monitoring under domain shifted conditions, three methods are proposed to solve this problem: principal component analysis (PCA), outlier classifier and contarative learning. Firstly, PCA is used for anomaly detection with three components. Secondly, outlier classifier is used by selecting the normal sound of other section as outlier samples. At last, contrastive learning is used by taking normal samples of other sections as negative examples. We present results obtained by each kind of models separately,as well as,a results of an esemble obtained by averaging anomaly scores computed by individual models.

System characteristics
Classifier CNN, PCA
System complexity 2000000, 220000, 2791000, 571000 parameters
Acoustic features log-mel energies
Subsystem count 3
PDF

DCASE Challenge 2021: Unsupervised Anomalous Sound Detection of Machinery with LeNet Architecture

Lam Pham, Anahid Jalali, Olivia Dinica, and Alexander Schindler
Data Science and Artificial Intelligence, AIT., Vienna, Austria

Abstract

In this study, we present an unsupervised anomalous sound detection framework trained on the DCASE2021 audio dataset. We use LeNet architecture to classify the machine IDs and use the classification loss as a threshold for detecting the anomalies in an unsupervised manner. We train our classifier on log-mel-bands and use the Mixup approach to augment our training set. Our framework outperforms both DCASE2021 benchmarks: the dense autoencoder and the MobileNet. The dense autoencoder has a harmonic mean of AUC of 61.92 and pAUC of 53.26 and the MobileNet has a harmonic mean of AUC of 59.72 and pAUC of 56.37. Our framework achieved the harmonic mean AUC of 66.72 and pAUC of 60.59, over all the machines, which shows an improved performance of 7.75% and 13.76%, AUC- and pAUC-harmonic-mean respectively from the dense autoencoder. The improved performance of our approach from the Mobilenet baseline is 11.72% and 7.48%, AUC- and pAUC-harmonic-mean respectively.

System characteristics
Classifier LeNet
System complexity 391302 parameters
Acoustic features log-mel energies
Data augmentation mixup
PDF

ANOMALOUS SOUND DETECTION WITH ENSEMBLE OF AUTOENCODER AND BINARY CLASSIFICATION APPROACHES

Ibuki Kuroyanagi, Tomoki Hayashi, Yusuke Adachi, Takenori Yoshimura, Kazuya Takeda, and Tomoki Toda
Nagoya University and Human Dataware Lab. Co., Ltd., Nagoya, Japan and Human Dataware Lab. Co., Ltd. and Nagoya University, Nagoya, Japan and Human Dataware Lab. Co., Ltd., Nagoya, Japan and Nagoya University, Nagoya, Japan

Abstract

This paper describes a solution with the ensemble of two unsupervised anomalous sound detection (ASD) methods for the DCASE2021 Challenge Task 2. The first ASD method is based on a sequence-level autoencoder with section ID regression and a self-attention architecture. We introduce the data augmentation techniques such as SpecAugment to boost up the performance and combine the simple scorer module for each section and each domain to address the domain shift problem. The second ASD method is based on a binary classification model using metric learning, which utilizing task-irrelevant outliers as pseudo-anomalous data and considering the centroid of normal and outlier data in the feature space. As a countermeasure against the domain shift problem, we perform data augmentation using Mixup with data from the target domain, resulting in a stable performance for each section. On the development set, our method achieves a harmonic mean of 76.59% harmonically averaged over of area under the curve (AUC) and partial AUC (p = 0.1) of all machines, sections, and domains.

System characteristics
Classifier AE, ArcFace, CNN, Conformer, GMM, ID regression, binary classification, ensemble
System complexity 400000000, 471127950, 71127950 parameters
Acoustic features log-mel energies
Data augmentation SpecAugment, SpecAugment, mixup, gaussian noise, volume perturbation, mixup, gaussian noise, volume perturbation
Decision making average, maximum, median, raking
System embeddings ResNet34, ResNeXt50, efficientnet-b3
Subsystem count 10, 18, 6
External data usage pre-trained model
PDF

Unsupervised Adversarial domain adaptive abnormal sound detection for machine condition monitoring under Domain Shift Conditions

Renjie Li, Xiaohua Gu, Fei Lu, Hongfei Song, and Jutao Pan
College of Electrical Engineering, Chongqing University of Science and Technology, Chongqing, China and Chongqing University of Science and Technology, Chongqing, China

Abstract

In the industrial field, it is very important to detect unknown anomalies based on normal production data. Facing the actual production situation, it is also of great significance to study the abnormal detection of the machine under the condition of constantly changing operating conditions. In dcase2021 task 2, we propose to use an unsupervised abnormal sound detection method based on adversarial domain adaptation. This method proposes a framework of adding domain discriminator and one-class classifier on the basis of auto-encoder, and achieves good results on the development dataset provided by the contest.

System characteristics
Classifier AE, ELM, GAN
System complexity 894781 parameters
Acoustic features log-mel energies
PDF

UNSUPERVISED ANOMALOUS SOUND DETECTION VIA SEMI-SUPERVISED GANOMALY ADVERSARIAL TRAINING

Wenbin Zhu, Jie Ou, Ying Zeng, Yi Zhou, and Hongqing Liu
School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing, China and Chongqing University of Posts and Telecommunications, Chongqing, China

Abstract

This technical report describes the submission from our team for Task 2 of the DCASE2021 challenge Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions. In this work, we adopt a GANomaly semi-supervised anomaly detection method via adversarial training to perform anomalous sound detection. By using the conditional generation of the confrontation network, the generator network effectively fits the data distribution of the normal samples during training, and calculates the reconstruction error of the anomaly score of the test samples.

System characteristics
Classifier GAN
System complexity 1166568 parameters
Acoustic features log-mel energies
PDF
Source code

Ensemble of Complementary Anomaly Detectors Under Domain Shifted Conditions

Jose A. Lopez, Georg Stemmer, Paulo Lopez-Meyer, Pradyumna S. Singh Juan Del Hoyo Ontiveros, and Hector Courdourier
Intel Labs, Santa Clara, CA and Intel Labs, Neubiberg, Germany and Intel Labs, Zapopan, Mexico

Abstract

We present our submission to the DCASE2021 Challenge Task 2, which aims to promote research in anomalous sound detection. We found that blending the predictions of various anomaly detectors, rather than relying on well-known domain adaptation techniques alone, gave us the best performance under domain shifted conditions. Our submission is composed of two self-supervised classifier models, a probabilistic model we call NF-CDEE, and an ensemble of the three.

System characteristics
Classifier classifier, ensemble, normalizing flow
System complexity 11805773, 1889899, 3341026, 6574848 parameters
Acoustic features STFT, log-mel energies, raw waveform
Front end system Teager-Kaiser
PDF

Anomalous Sounds Detection Using Autoencoder and Classification Methods

Haisheng Lu, Yujie Fu, Huajing Qin, Shijin Huang, Yihan Wang, Chen Deng, Tianchu Yao, Huitian Jiang, Haifeng Wen, and Chuang Shi
School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, China

Abstract

This report described our contribution to Unsupervised Detection of Anomalous Sounds on DCASE 2020 challenge (Task2). Previous research results show that AE and outlier detection is a very effective solution to abnormal sound detection (ASD). This design based on previous research, using IDNN, FREAK and MobileFaceNets to implement unsupervised ASD.

System characteristics
Classifier AE, CNN, VAE
System complexity 1675971, 185496, 186528, 989692 parameters
Acoustic features STFT, log-mel energies
Decision making average
Subsystem count 2
Front end system STFT cut
PDF

ANOMALOUS SOUND DETECTION USING CNN-BASED FEATURES BY SELF SUPERVISED LEARNING

Kazuki Morita, Tomohiko Yano, and Khai Q. Tran
Intelligent Systems Laboratory, SECOM CO.,LTD., Tokyo, Japan

Abstract

We propose a detection method for the anomalous sound detection task of DCASE2021 task2 in this report. This is the task of anomalous sound detection for machine condition monitoring, and it is required to detect unknown anomalous sound only from normal sound data. We use the normal sound of the machine and its section index to train the Convolutional Neural Network (CNN) in a self-supervised learning manner. Then, we detect anomalous sound by using feature vectors extracted from CNN. As a result, for the development dataset we show the detection performance of 78.05% in Area Under Curve (AUC) and 68.09% in partial AUC (pAUC).

System characteristics
Classifier CNN, LOF, k-NN
System complexity 994048 parameters
Acoustic features spectrogram
PDF

Unsupervised Anomalous Sound Detection Using Intermediate Representation of Trained Models and Metric Learning Based Variational Autoencoder

Hiroki Narita and Akira Tamamori
Aichi Institute of Technology, Aichi, Japan

Abstract

This paper is a technical report of DCASE Challenge2021 Task2. The objective of the DCASE Challenge2021 Task2 is unsupervised anomalous sound detection under domain shift. Our method consists of feature extraction using a pretrained model and Center-Loss VAE (CL-VAE) based on Center-Loss and Variational AutoEncoder (VAE). In feature extraction with pre-trained models, ResNet38 trained on acoustic data is used as a feature extractor to obtain intermediate representations. The CL-VAE is trained with the intermediate representations as input and is trained to minimize the Center-Loss of the section labels and the loss function of the VAE. As a result of validation on the development dataset, we confirmed that the performance of CL-VAE is superior to that of Conditional VAE (CVAE) using baseline models and section labels.

System characteristics
Classifier CenterLoss, VAE
System complexity 128043983, 182304719 parameters
Acoustic features log-mel energies
Decision making average
System embeddings PANNs ResNet38
Subsystem count 2
External data usage pre-trained model
Front end system Resampling(16kHz to 32kHz)
PDF

CP-JKU Submission to DCASE'21: Improving Out-of-Distribution Detectors for Machine Condition Monitoring with Proxy Outliers & Domain Adaptation via Semantic Alignment

Paul Primus, Martin Zwifl, and Gerhard Widmer
Institute of Computational Perception, Johannes Kepler University Linz, Linz, Austria and Johannes Kepler University Linz, Linz, Austria

Abstract

This technical report contains a detailed summary of our submissions to the Unsupervised Anomalous Sound Detection under Domain Shifted Conditions Task for Machine Condition Monitoring of the IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Events 2021 (DCASE). Our goal was to learn out-of-distribution (OOD) detectors without access to OOD data, i.e., we trained only on recordings of undamaged machines. To this end, we employed a range of popular unsupervised anomaly detection methods based on auxiliary classification, density estimation, and reconstruction error. OOD detectors were trained for each of the seven machine categories included in the development dataset. We then showed that the OOD detectors’ performance was enhanced by utilizing metadata labels and other machines’ regular sounds as proxy outliers. To further improve detection performance under domain-shifted conditions, we fine-tuned the auxiliary classifiers to semantically align the hidden representations of source and target domain, using the limited target domain data. In addition to this technical description, we release our complete source code to make our submission fully reproducible.

System characteristics
Classifier Ensemble, MADE, MAF, ResNet
System complexity 144900000, 228200000, 421372000, 46200000 parameters
Acoustic features log-mel energies
Decision making average
Subsystem count 3
PDF
Source code

COMBINE MAHALANOBIS DISTANCE, INTERPOLATION AUTO ENCODER AND CLASSIFICATION APPROACH FOR ANOMALY DETECTION

Yuya Sakamoto and Naoya Miyamoto
Fixstars Corporation, Tokyo, Japan

Abstract

This paper is a technical report of the method we submitted to DCASE 2021 Challenge Task 2. In our method, one sample is converted into a time-series log-mel-spectrogram similar to the Autoencoder-based baseline. For the feature vector obtained from this log-mel-spectrogram, 3 types of anomaly detection models, section ID classification, interpolation deep neural network and mahalanobis distance are constructed, and the final degree of anomaly is calculated as an ensemble of 3 models. In this task, it is necessary to deal with the domain shift problem, which has different characteristics between training data and test data. We addressed this problem by absorbing the difference in the mean of log-mel-spectrogram features between domains.

System characteristics
Classifier IDNN, Mahalanobis distance, Section ID classification
System complexity 184320, 301315 parameters
Acoustic features log-mel energies
Subsystem count 2, 3
PDF

Anomaly Sound Detection Using Essemble of Autoencoders

Ee-Leng Tan, Santi Peksi, and Nguyen Duy Hai
EEE, Nanyang Technological University, Singapore, Singapore

Abstract

This technical report outlines our solution to task 2 of the detection and classification of acoustic scenes and events (DCASE) 2021 challenge. The objective of this task is to identify anomalous sounds using an anomaly detector trained with normal sound only and to avoid identifying normal sounds that deviate from the operating condition of the normal sounds in the training dataset as anomalous sounds. Our approach is based on an assemble of autoencoders with different network architectures targeted to different machine types.

System characteristics
Classifier AE
System complexity 2,415,060 parameters
Acoustic features log-mel energies
Data augmentation Guassian noise, frequency shifting
Decision making Static classifier selection
Subsystem count 3
PDF

Unsupervised Anomalous Sound Detection by Siamese Network and Auto-Encoder

Jan Tozicka, Karel Durkota, and Michal Linda
NeuronSW SE, Prague, Czech Republic

Abstract

This paper describes our submission to the DCASE 2021 challenge Task 2 ”Unsupervised Detection of Anomalous Sounds for Machine Condition Monitoring under Domain Shifted Conditions.” Acoustic-based machine condition monitoring is a challenging task with a very unbalanced training dataset. In this submission, we propose two methods for anomaly detection and then their combination. The first method is based on feature extractor using Siamese Network with triplet loss and KNN for the actual anomaly detection. The second method uses very small auto-encoder on top of the OpenL3 embeddings. The combination of these two approaches selects the best performing method for each machine type. This is a novel approach and have not been used by NeuronSW SE so far.

System characteristics
Classifier AE, KNN
System complexity 10629898, 1576834, 9053064 parameters
Acoustic features log-mel energies
System embeddings OpenL3, Siamese Network, Siamese Network, OpenL3
PDF

DCASE 2021 TASK 2: ANOMALOUS SOUND DETECTION USING CONDITIONAL AUTOENCODER AND CONVOLUTIONAL RECURRENT NEURAL NETW

Wei-Lin Liao, Tsung-Han Wu, Shu-Yu Chen, Yun-Shing Wu, Chia-Yin Chen, Cai-Yu Yuan, Chung-Che Wang, and Jyh-Shing Roger Jang
Dept. of Mechanical Engineering, National Taiwan Univ., Taipei, Taiwan and Dept. of Computer Science and Information Engineering, National Taiwan Univ., Taipei, Taiwan and FinTech Center, National Taiwan Univ., Taipei, Taiwan

Abstract

This technical report describes our methods to Task 2 of the DCASE 2021 challenge: Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions. We use reconstruction error of a conditional autoencoder and 1 - classification confidence of a classifier as anomaly scores.

System characteristics
Classifier CAE, CRNN
System complexity 31989511, 663783, 7719648, 7916544 parameters
Acoustic features log-mel energies
PDF

DATA AUGMENTATION AND CLASS-BASED ENSEMBLED CNN-CONFORMER NETWORKS FOR SOUND EVENT LOCALIZATION AND DETECTION

Shuo Wang, Zihao Li, Yuxuan Zhang, Kejian Guo, Shijin Chen, and Yan Pang
Electronic, Electrical and Communication Engineering, Beijing, China

Abstract

In the industrial field, the anomaly detection of mechanical systems has played an important role. This technical report uses four modified autoencoders (AEs) to detect abnormal conditions of different machines in DCASE2021 Task 2. AE has been widely used in image reconstruction due to its excellent generalization ability. The reconstruction error can be used to evaluate the abnormal value of the machine condition when the development set only provide the normal mechanical sound signals. The performance of the anomaly detection system is evaluated by the area under the receiver operating characteristic curve (AUC) and partial-AUC (pAUC) scores. Finally, the experimental results show that the presented models can improve AUC and pAUC compared to the baseline system.

System characteristics
Classifier AE, IAE, IAVE, VAE
System complexity 1,854,376, 1,860,008, 3,714,384, 630,072 parameters
Acoustic features log-mel energies
Decision making average
PDF

Utilizing Sub-Cluster AdaCos for Anomalous Sound Detection under Domain Shifted Conditions

Kevin Wilkinghoff
Communication Systems, Fraunhofer Institute for Communication, Information Processing and Ergonomics FKIE, Wachtberg, Germany

Abstract

Anomalous sound detection systems based on sub-cluster AdaCos yield state-of-the-art performance on the DCASE 2020 dataset for anomalous sound detection. In contrast to the previous year, the dataset belonging to task 2 “Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions” of the DCASE challenge 2021 contains not only source domains with 1000 normal training samples for each machine but also so-called target domains with different acoustic conditions for which only 3 normal training samples are available. To address this additional problem, a novel anomalous sound detection system based on sub-cluster AdaCos for the DCASE challenge 2021 is presented. This system is trained to extract embeddings whose distributions are estimated in different ways for source and target domains, and utilize their negative log-likelihoods as anomaly scores. In experimental evaluations, it is shown that the presented system significantly outperforms both baseline systems on source and target domains of the development set.

System characteristics
Classifier CNN, GMM, ensemble
System complexity 1189557471, 1217299311, 609044474 parameters
Acoustic features log-mel energies
Data augmentation mixup
Decision making average, sum
Subsystem count 20, 40
PDF

VAE-based anomaly detection with domain adaptation

Jun’ya Yamashita, Hayato Mori, Satoshi Tamura, and Satoru Hayamizu
Graduate School of Natural Science and Technology, Gifu University, Gifu, Japan and Faculty of Engineering, Gifu University, Gifu, Japan

Abstract

This paper presents our anomaly detection scheme for DCASE 2021 Challenge, using a Variational AutoEncoder (VAE) with a framework of Interpolation Deep Neural Network (IDNN) and fine tuning as an adaptation method. VAE is built using normal training data for each machine, to predict a frame from its neighbor frames just like IDNN. In addition, we involve a kind of high-pass filter and a scheme to preserve particular frames or frequencies having larger errors. Finally an anomaly score is calculated based on reconstruction error in VAE. We further apply fine tuning to target data recorded in different settings, to adapt a model.

System characteristics
Classifier CNN, VAE
System complexity 1254167 parameters
Acoustic features log-mel energies
PDF

Unsupervised Anomalous Sound Detection Using Denoising-Detection System Under Domain Shifted Conditions

Chenxu Zhang, Yao Yao, Rui Qiu, Shengchen Li, and Xi Shao
NJUPT, Nanjing, China and XJTLU, Suzhou, China

Abstract

The DCASE2021 Challenge Task2 is to develop an unsupervised detection system of anomalous sounds for seven types of machines under domain shifted conditions. A common challenge in the detection of anomalous sounds for machine is to identify the diversity of malfunctioning sounds and the scarcity of malfunctioning sounds samples between normal and anomalous condition. In this paper, an unsupervised denoising-detection system is proposed to perform this task by: (1) removing noise in each recording to obtain signal that is more related to this task; (2) training an overfitting model by leveraging the information of sections in each machine type. The experimental evaluation demonstrates that the proposed system outperforms the provided baseline system across majority of machine types in both source domain and target domain.

System characteristics
Classifier AE, MobileNetV2
System complexity 2933855 parameters
Acoustic features log-mel energies
Data augmentation Audio denoising, Mixup
Subsystem count 2
External data usage train denoising model
PDF

ENSEMBLE OF ARCFACE BASED SYSTEMS FOR UNSUPERVISED ANOMALOUS SOUND DETECTION UNDER DOMAIN SHIFT CONDITIONS

Qiping Zhou
R&D department, PFU Shanghai Co., LTD, Shanghai, China

Abstract

In this report, we outline our ensemble of models solution for the DCASE 2021 challenge’s Task 2 (Unsupervised Anomalous Sound Detection for Machine Condition Monitoring under Domain Shifted Conditions) [1]. The basic approach follows our DCASE2020 Task 2 system [2]. In 2021 we diversify our CNN backbone architecture and input size. The final submissions are the ensemble of 6 models for each machine type. Models are trained on source domain data and fine-tuned on target domain data to improve the performance on the domain shifted data.

System characteristics
Classifier ArcFace, CNN, ensemble
System complexity 58316344 parameters
Acoustic features spectrogram
Data augmentation mixup
Decision making average, maximum
Subsystem count 6
PDF