Object Detection on COCO Generalization Benchmark
88.2VOC mAPDiff. Detector (DINO*)
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Diff. Detector (DINO*)Training Scale=100% COCO, Framework=SD-1.5, Inf. time (ms)=2322025.06 | 88.2 | 51.2 | 51.1 | 45.9 | 61.7 | 50 | 63.6 | 42 | 39.3 | 22 | 40.2 | 50.5 | |
| Diff. Detector (SD-1.5)Training Scale=100% COCO, Detector=Faster R-CNN, Inf. time (ms)=1642025.06 | 86.6 | 54.1 | 51.2 | 45.8 | 64.6 | 51.9 | 66.6 | 41.7 | 39.2 | 21.7 | 40.7 | 51.3 | |
| ViT-BaseTraining Scale=100% COCO, Detector=Faster R-CNN, Inf. time (ms)=782025.06 | 86.3 | 35.1 | 40.4 | 22.9 | 38.3 | 24.1 | 48.9 | 31.2 | 29.1 | 12.4 | 26.3 | 35.9 | |
| Diff. Detector (FCOS*)Training Scale=100% COCO, Framework=SD-1.5, Inf. time (ms)=1712025.06 | 85 | 50.3 | 48.3 | 42 | 62.9 | 52.1 | 65.9 | 39.6 | 36.4 | 19.7 | 39 | 49.2 | |
| ResNet-R50Training Scale=100% COCO, Detector=Faster R-CNN, Inf. time (ms)=272025.06 | 84 | 49.5 | 45.8 | 35.2 | 32.6 | 23.7 | 40.8 | 33.3 | 24.8 | 10.7 | 30.3 | 37.3 | |
| ConvNeXt-BaseTraining Scale=100% COCO, Detector=Faster R-CNN, Inf. time (ms)=542025.06 | 83.4 | 53.4 | 50.9 | 43.1 | 43.5 | 35.5 | 47.9 | 40.5 | 34.4 | 17.6 | 36 | 44.2 | |
| GLIP (Swin-Tiny)Training Scale=100% COCO, Detector=Faster R-CNN, Inf. time (ms)=312025.06 | 79.9 | 54 | 48.6 | 41.4 | 38.9 | 29.6 | 40.3 | 37.9 | 31.5 | 15 | 34.9 | 41.1 | |
| Swin-BaseTraining Scale=100% COCO, Detector=Faster R-CNN, Inf. time (ms)=552025.06 | 79.6 | 52.9 | 48 | 38.9 | 37.8 | 29.4 | 41.7 | 37.6 | 32.1 | 14.4 | 33 | 40.5 | |
| Diff. Detector (SD-1.5)Training Scale=1% COCO, Detector=Faster R-CNN2025.06 | 70.4 | 36.5 | 27.9 | 31 | 44 | 37.8 | 50.1 | 26.6 | 19.8 | 10.7 | 20.8 | 34.1 | |
| ConvNeXt-BaseTraining Scale=1% COCO, Detector=Faster R-CNN2025.06 | 53.9 | 34.9 | 28.2 | 13.1 | 29.8 | 18.8 | 26.9 | 25.6 | 16.2 | 5.9 | 14.9 | 24.9 | |
| GLIP (Swin-Tiny)Training Scale=1% COCO, Detector=Faster R-CNN2025.06 | 53.9 | 36.1 | 27.5 | 19.6 | 18.6 | 13 | 16.1 | 20.1 | 14.5 | 4.1 | 13.7 | 21.6 | |
| ViT-BaseTraining Scale=1% COCO, Detector=Faster R-CNN2025.06 | 53.3 | 18 | 18 | 8.8 | 11.1 | 6.1 | 11.4 | 13.8 | 11.1 | 2.6 | 6.9 | 14.6 | |
| Swin-BaseTraining Scale=1% COCO, Detector=Faster R-CNN2025.06 | 46 | 29 | 23.1 | 16.3 | 12.9 | 8.1 | 19.8 | 16.8 | 10.3 | 2.1 | 9 | 17.6 | |
| ResNet-R50Training Scale=1% COCO, Detector=Faster R-CNN2025.06 | 36.4 | 26.6 | 18.6 | 8.2 | 10.1 | 6.6 | 14.3 | 10.6 | 5.9 | 1.1 | 6.5 | 13.2 |