Respiratory Sound Classification on ICBHI 2017 (test)
81.66SpecificityBae&Kim et al. (MixCL)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Bae&Kim et al. (MixCL)Backbone=AST, Number of parameters=86M2024.04 | 81.66 | 43.07 | 62.37 | |
| M2D-XBackbone=M2D ViT, Masking ratio=0.7, eta (background noise ratio)=0.3, Patch size=16x4, T (time duration)=2.0s, Pre-training=further pre-trained on ICBHI2017, Number of parameters=86M2024.04 | 81.51 | 45.08 | 63.29 | |
| M2DBackbone=M2D ViT, Masking ratio=0.7, eta (background noise ratio)=0.3, Patch size=16x4, T (time duration)=2.0s, Pre-training=further pre-trained on ICBHI2017, Number of parameters=86M2024.04 | 80.7 | 44.76 | 62.73 | |
| M2DBackbone=M2D ViT, Masking ratio=0.7, Patch size=16x4, T (time duration)=2.0s, Pre-training=AudioSet only, Number of parameters=86M2024.04 | 79.48 | 44.38 | 61.93 | |
| M2DBackbone=M2D ViT, Masking ratio=0.7, Patch size=16x16, T (time duration)=6.08s, Pre-training=AudioSet only, Number of parameters=86M2024.04 | 79.47 | 42.93 | 61.2 | |
| M2DBackbone=M2D ViT, Masking ratio=0.7, eta (background noise ratio)=0, Patch size=16x4, T (time duration)=2.0s, Pre-training=further pre-trained without BG noise, Number of parameters=86M2024.04 | 78.66 | 43.67 | 61.17 | |
| M2D-XBackbone=M2D ViT, Masking ratio=0.7, eta (background noise ratio)=0, Patch size=16x4, T (time duration)=2.0s, Pre-training=further pre-trained without BG noise, Number of parameters=86M2024.04 | 77.39 | 46.04 | 61.71 | |
| Bae&Kim et al. (CE)Backbone=AST, Number of parameters=86M2024.04 | 77.14 | 41.97 | 59.55 | |
| Moummad et al. (SCL)Backbone=CNN6, Number of parameters=4.3M2024.04 | 75.95 | 39.15 | 57.55 | |
| Moummad et al. (CE)Backbone=CNN6, Number of parameters=4.3M2024.04 | 70.09 | 40.39 | 55.24 |