Object Detection on CVTCropDet Ours (test)
49.17AP50CropVLM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CropVLMType=Open Vocabulary, Backbone=ResNet50 + ViT, Protocol=zero-shot2026.05 | 49.17 | 38.44 | |
| GDINOType=Open Vocabulary, Backbone=ViT, Protocol=zero-shot2026.05 | 48.58 | 42.04 | |
| DetProType=Open Vocabulary, Backbone=ResNet50, Protocol=zero-shot2026.05 | 39.67 | 30.75 | |
| CFMType=Open Vocabulary, Backbone=ResNet50 + ViT, Protocol=zero-shot2026.05 | 28.69 | 22.11 | |
| OV-DETRType=Open Vocabulary, Backbone=ResNet50 + ViT, Protocol=zero-shot2026.05 | 20.96 | 17.07 | |
| DETRType=Closed Vocabulary, Backbone=ResNet101, Protocol=zero-shot2026.05 | 19.67 | 16.42 | |
| RNCDLType=Open Vocabulary, Backbone=ResNet50, Protocol=zero-shot2026.05 | 17.91 | 14.33 | |
| YOLOv9Type=Closed Vocabulary, Backbone=CSPDarknet-X, Protocol=zero-shot2026.05 | 17.68 | 16.06 | |
| RCNNType=Closed Vocabulary, Backbone=ResNet50, Protocol=zero-shot2026.05 | 16.97 | 11.23 | |
| YOLOv8Type=Closed Vocabulary, Backbone=CSPDarknet53, Protocol=zero-shot2026.05 | 16.27 | 16.8 |