Referring Image Matting Expression Based On
評価指標
MAD
MAD(E)
MSE
MSE(E)
SAD
SAD(E)
評価結果
このベンチマークにおける各モデルのパフォーマンス結果
モデル名 | MAD | MAD(E) | MSE | MSE(E) | SAD | SAD(E) | Paper Title | Repository |
|---|---|---|---|---|---|---|---|---|
| CLIPSeg (ViT-B/16) | 0.0394 | 0.0419 | 0.0358 | 0.0381 | 69.13 | 73.53 | Image Segmentation Using Text and Image Prompts | |
| CLIPMat (ViT-B/16) | 0.0273 | 0.0273 | 0.0245 | 0.0260 | 47.97 | 50.84 | Referring Image Matting | |
| CLIPMat (ViT-L/14) | 0.0238 | 0.0254 | 0.0212 | 0.0226 | 42.05 | 44.77 | Referring Image Matting | |
| MDETR (ResNet-101) | 0.0482 | 0.0515 | 0.0434 | 0.0463 | 84.70 | 90.45 | MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding |
0 of 4 row(s) selected.