Visual Reasoning 벤치마크
Visual Reasoning on VSR
accuracy
- 2022-04-30 — LXMERT: accuracy 70.1
| Rank | Model | accuracy | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | LXMERT | 70.1 | Visual Spatial Reasoning | cambridgeltl/visual-spatial-reasoning · ziyan-xiaoyu/spatialmqa · sohojoe/clip_visual-spatial-reasoning · +1 | 2022 |
| 2 | ViLT | 69.3 | Visual Spatial Reasoning | cambridgeltl/visual-spatial-reasoning · ziyan-xiaoyu/spatialmqa · sohojoe/clip_visual-spatial-reasoning · +1 | 2022 |
| 3 | CLIP (finetuned) | 65.1 | Visual Spatial Reasoning | cambridgeltl/visual-spatial-reasoning · ziyan-xiaoyu/spatialmqa · sohojoe/clip_visual-spatial-reasoning · +1 | 2022 |
| 4 | CLIP (frozen) | 56.0 | Visual Spatial Reasoning | cambridgeltl/visual-spatial-reasoning · ziyan-xiaoyu/spatialmqa · sohojoe/clip_visual-spatial-reasoning · +1 | 2022 |
| 5 | VisualBERT | 55.2 | Visual Spatial Reasoning | cambridgeltl/visual-spatial-reasoning · ziyan-xiaoyu/spatialmqa · sohojoe/clip_visual-spatial-reasoning · +1 | 2022 |