Action Semantic Segmentation on A2D (test)
49.4mIoUfully-convolutional model
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| fully-convolutional modelInput type=actor and action pairs as text2018.03 | 49.4 | 69.3 | 92.5 | |
| Kalogeiton et al. [10]2018.03 | 42.2 | 60.5 | 89.3 | |
| Xu et al. [28]2018.03 | 32 | 60.5 | 85.3 | |
| Few-shot transformerShot count=five-shot2021.04 | 0.525 | — | — | |
| Few-shot transformerShot count=one-shot2021.04 | 0.506 | — | — | |
| Siam et al.Shot count=five-shot2021.04 | 0.448 | — | — | |
| Siam et al.Shot count=one-shot2021.04 | 0.433 | — | — | |
| Xu et al. [29]2018.03 | — | 47 | 74.6 |