Command Palette
Search for a command to run...
Christoph Feichtenhofer; Haoqi Fan; Jitendra Malik; Kaiming He
Abstract
We present SlowFast networks for video recognition. Our model involves (i) a Slow pathway, operating at low frame rate, to capture spatial semantics, and (ii) a Fast pathway, operating at high frame rate, to capture motion at fine temporal resolution. The Fast pathway can be made very lightweight by reducing its channel capacity, yet can learn useful temporal information for video recognition. Our models achieve strong performance for both action classification and detection in video, and large improvements are pin-pointed as contributions by our SlowFast concept. We report state-of-the-art accuracy on major video recognition benchmarks, Kinetics, Charades and AVA. Code has been made available at: https://github.com/facebookresearch/SlowFast
Code Repositories
Benchmarks
| Benchmark | Methodology | Metrics |
|---|---|---|
| action-classification-on-charades | SlowFast (Kinetics-600 pretraining, NL) | MAP: 45.2 |
| action-classification-on-charades | SlowFast (Kinetics-600 pretraining) | MAP: 42.1 |
| action-classification-on-charades | SlowFast (Kinetics-400 pretraining, NL) | MAP: 42.5 |
| action-classification-on-kinetics-400 | SlowFast 16x8 (ResNet-101) | Acc@1: 78.9 Acc@5: 93.5 |
| action-classification-on-kinetics-400 | SlowFast 16x8 (ResNet-101 + NL) | Acc@5: 93.9 |
| action-classification-on-kinetics-400 | SlowFast 4x16 (ResNet-50) | Acc@1: 75.6 Acc@5: 92.1 |
| action-classification-on-kinetics-400 | SlowFast 8x8 (ResNet-101) | Acc@1: 77.9 Acc@5: 93.2 |
| action-classification-on-kinetics-400 | SlowFast 16x8 (ResNet-101 + NL) | Acc@1: 79.8 |
| action-classification-on-kinetics-400 | SlowFast 8x8 (ResNet-50) | Acc@1: 77 Acc@5: 92.6 |
| action-classification-on-kinetics-600 | SlowFast 8x8 (ResNet-50) | Top-1 Accuracy: 79.9 Top-5 Accuracy: 94.5 |
| action-classification-on-kinetics-600 | SlowFast 16x8 (ResNet-101 + NL) | Top-1 Accuracy: 81.8 Top-5 Accuracy: 95.1 |
| action-classification-on-kinetics-600 | SlowFast 16x8 (ResNet-101) | Top-1 Accuracy: 81.1 Top-5 Accuracy: 95.1 |
| action-classification-on-kinetics-600 | SlowFast 8x8 (ResNet-101) | Top-1 Accuracy: 80.4 Top-5 Accuracy: 94.8 |
| action-classification-on-kinetics-600 | SlowFast 4x16 (ResNet-50) | Top-1 Accuracy: 78.8 Top-5 Accuracy: 94 |
| action-recognition-in-videos-on-ava-v21 | SlowFast (Kinetics-400 pretraining) | mAP (Val): 26.3 |
| action-recognition-in-videos-on-ava-v21 | SlowFast++ (Kinetics-600 pretraining, NL) | mAP (Val): 28.3 |
| action-recognition-in-videos-on-ava-v21 | SlowFast (Kinetics-600 pretraining, NL) | mAP (Val): 27.3 |
| action-recognition-in-videos-on-ava-v21 | SlowFast (Kinetics-600 pretraining) | mAP (Val): 26.8 |
| action-recognition-in-videos-on-something | SlowFast | Top-1 Accuracy: 61.7 |
| action-recognition-on-ava-v2-2 | SlowFast, 4x16, R50 (Kinetics-400 pretraining) | mAP: 21.9 |
| action-recognition-on-ava-v2-2 | SlowFast, 8x8, R101 (Kinetics-400 pretraining) | mAP: 23.8 |
| action-recognition-on-ava-v2-2 | SlowFast, 16x8 R101+NL (Kinetics-600 pretraining) | mAP: 27.5 |
| action-recognition-on-ava-v2-2 | SlowFast, 8x8 R101+NL (Kinetics-600 pretraining) | mAP: 27.1 |
| action-recognition-on-diving-48 | SlowFast | Accuracy: 77.6 |
| action-recognition-on-h2o-2-hands-and-objects | SlowFast | Actions Top-1: 77.69 Hand Pose: No Object Label: No Object Pose: No RGB: Yes |
Build AI with AI
From idea to launch — accelerate your AI development with free AI co-coding, out-of-the-box environment and best price of GPUs.