Image Matting On Composition 1K 1
Métriques
Conn
Grad
MSE
SAD
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | Conn | Grad | MSE | SAD |
|---|---|---|---|---|
| pp-matting-high-accuracy-natural-image | 45.4 | 22.69 | 5.0 | 46.22 |
| vitmatte-boosting-image-matting-with | 14.78 | 6.74 | 3.0 | 20.33 |
| 190909725 | 33.2 | 17.3 | 8.2 | 35.8 |
| deep-image-matting | 50.8 | 31.0 | 14.0 | - |
| trimap-guided-feature-mining-and-fusion | 17.6 | 6.7 | 3.6 | 22.1 |
| indices-matter-learning-to-index-for-deep | 43.7 | 25.9 | 13.0 | 45.8 |
| f-b-alpha-matting | 21.5 | 10.6 | 5.3 | 26.4 |
| improving-deep-image-matting-via-local | 21.5 | 9.25 | 5.4 | 25.9 |
| rethinking-context-aggregation-in-natural | 12.46 | 4.76 | 2.26 | 17.53 |
| semantic-image-matting | 24.8 | 10.8 | 5.8 | 28.0 |
| matteformer-transformer-based-image-matting | 18.9 | 8.7 | 4.0 | 23.8 |
| diffusion-for-natural-image-matting | 11.42 | 5.13 | 2.26 | 17.15 |
| long-range-feature-propagating-for-natural | 18.5 | 8.4 | 4.1 | 23.6 |