Object Detection on Real-to-Art Style Transfer (VOC -> Comic, Watercolor, Clipart) (test)
87.6mAP (VOC)Style Evolving along Chain-of-Thought (Ours)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Style Evolving along Chain-of-Thought (Ours)Backbone=Swin2025.03 | 87.6 | 36.9 | 60.7 | 42.5 | |
| Style Evolving along Chain-of-Thought (Ours)Backbone=ResNet-1012025.03 | 82.9 | 34.8 | 57.5 | 40.2 | |
| C-GapBackbone=ResNet-1012025.03 | 80.5 | 29.4 | 50.7 | 36.7 | |
| Faster R-CNNBackbone=ResNet-1012025.03 | 80.4 | 19.4 | 45.6 | 26.5 | |
| DIV2025.03 | 80.1 | 33.2 | 57.4 | 38.9 | |
| NPBackbone=ResNet-1012025.03 | 79.2 | 28.9 | 53.3 | 35.4 | |
| Style Evolving along Chain-of-Thought (Ours)Backbone=ResNet-502025.03 | 78.5 | 29.4 | 53.2 | 34.7 |