Action Recognition on Video Frames Inference Speed (test)
0.1Latency (ms)DMC-Net
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DMC-NetEvaluation Stage=Generator only, Batch size=1, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 0.1 | 9,433.9 | |
| DMC-NetEvaluation Stage=Generator + Classifier, Batch size=1, Classifier backbone=ResNet-18, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 0.3 | 3,333.3 | |
| PWC-NetEvaluation Stage=Generator only, Batch size=1, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 28.6 | 35 | |
| PWC-NetEvaluation Stage=Generator + Classifier, Batch size=1, Classifier backbone=ResNet-18, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 28.8 | 34.8 | |
| TVNetEvaluation Stage=Generator only, Batch size=1, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 83.3 | 12 | |
| TVNetEvaluation Stage=Generator + Classifier, Batch size=1, Classifier backbone=ResNet-18, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 83.5 | 12 | |
| Flownet2.0Evaluation Stage=Generator only, Batch size=1, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 220.8 | 4.5 | |
| Flownet2.0Evaluation Stage=Generator + Classifier, Batch size=1, Classifier backbone=ResNet-18, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 221 | 4.5 | |
| DeepflowEvaluation Stage=Generator only, Batch size=1, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 1,449.2 | 0.7 | |
| DeepflowEvaluation Stage=Generator + Classifier, Batch size=1, Classifier backbone=ResNet-18, Hardware=NVIDIA GeForce GTX 1080 Ti2019.01 | 1,449.5 | 0.7 |