Disruption Prediction on EAST multimodal generalization
94.51TPRMultimodal teacher network
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Multimodal teacher networkTrain Mode=Multimodal, Test Mode=Multimodal, Threshold=0.78, Inf. (ms)=8.09, FLOPs (G)=31.35, Params (M)=58.032026.07 | 94.51 | 10.93 | 82.88 | 94.82 | 859.11 | |
| Video-based modelTrain Mode=Video, Test Mode=Video, Threshold=0.75, Inf. (ms)=4.10, FLOPs (G)=21.60, Params (M)=30.022026.07 | 85.97 | 9.74 | 79.66 | 93.23 | 316.92 | |
| Time-series-based modelTrain Mode=Time series, Test Mode=Time series, Threshold=0.61, Inf. (ms)=3.75, FLOPs (G)=9.75, Params (M)=30.262026.07 | 78.04 | 21.47 | 64 | 85.24 | 1,716.22 | |
| Unimodal student networkTrain Mode=Multimodal, Test Mode=Time series, Threshold=0.63, Inf. (ms)=3.75, FLOPs (G)=9.75, Params (M)=30.262026.07 | 75 | 10.53 | 72.35 | 88.06 | 1,166.9 |