Action Recognition on ActivityNet 1.3 (val)
95.9Top-1 AccuracyInternVideo2
Evaluation Results
| Method | Links | |
|---|---|---|
| InternVideo2TFLOPS=36.3, Backbone=InternVL-6B, #F=162025.03 | 95.9 | |
| CASTTFLOPS=1.2, Backbone=VideoMAE-B&CLIP-B, #F=482025.03 | 91.3 | |
| NSNetBackbone=Swin-L, #F=1002025.03 | 90.2 | |
| TSQNetBackbone=Swin-L, #F=502025.03 | 88.7 | |
| MARLBackbone=SEResNeXt-152, #F=1202025.03 | 85.7 | |
| SMART#F=242025.03 | 84.4 | |
| DSNBackbone=R(2+1)D-34, #F=802025.03 | 82.6 |