this is the pytorch implementation for 'P3D'.
this project can be used as video understanding/recognition
paper : Learning Spatio-Temporal Representation with Pseudo-3D Residual,(2017-ICCV)
reference github : https://github.com/qijiezhao/pseudo-3d-pytorch
tensorboard check : tensorboard --logdir=./run --port=8008
refer result : - Action recognition(mean accuracy on UCF101):
| modality/model | RGB | Flow | Fusion |
|---|---|---|---|
| P3D199 (Sports-1M) | 88.5% | - | - |
| P3D199 (Kinetics) | 91.2% | 92.4% | 98.3% |
- Action localization(mAP on Thumos14):
| Step | perframe | localization |
|---|---|---|
| P3D199(Sports-1M | 0.451 | 0.25 |
| P3D199(Kinetics) | 0.569(fused) | 0.307 |