Object Detection on VisDrone (test)
65.8AP50DroneFINE-L
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DroneFINE-LBackbone=Swin-L2026.07 | 65.8 | — | 42.3 | — | — | — | — | — | — | — | — | — | |
| RT-OVADBackbone=ResNet-502026.07 | 64.6 | — | — | — | — | — | — | — | — | — | — | — | |
| +Both (ZoomDet + DG.)Zooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=329.42026.02 | 62.6 | — | 38.1 | — | — | — | 31.4 | 48.6 | 53.5 | — | — | — | |
| UFPMP-DetBackbone=ResNet-502026.07 | 62.4 | — | 36.6 | — | — | — | — | — | — | — | — | — | |
| ESODBackbone=YOLOv52026.07 | 62.3 | — | 37.9 | — | — | — | — | — | — | — | — | — | |
| RemDet-XBackbone=YOLOv8X2026.07 | 61.9 | — | 40 | — | — | — | — | — | — | — | — | — | |
| DroneFINE-BBackbone=Swin-B2026.07 | 61.9 | — | 38.8 | — | — | — | — | — | — | — | — | — | |
| DG.Zooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=305.92026.02 | 61.7 | — | 37.4 | — | — | — | 30.2 | 47.3 | 52 | — | — | — | |
| DroneFINE-TBackbone=Swin-T2026.07 | 61.2 | — | 38.4 | — | — | — | — | — | — | — | — | — | |
| Dome-DETRBackbone=HGNetv2L2026.07 | 61.1 | — | 39 | — | — | — | — | — | — | — | — | — | |
| +Both (ZoomDet + ESOD)Zooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=141.62026.02 | 60.3 | — | 36.8 | — | — | — | 29.3 | 46.9 | 44.9 | — | — | — | |
| ZoomDetZooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=210.22026.02 | 60.2 | — | 35.1 | — | — | — | 28.1 | 44.7 | 44.3 | — | — | — | |
| ZoomDetZooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=158.72026.02 | 60.1 | — | 36.3 | — | — | — | 28.4 | 46.9 | 52.8 | — | — | — | |
| ESODZooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=135.42026.02 | 59.7 | — | 36 | — | — | — | 28.3 | 45.9 | 45 | — | — | — | |
| BaseZooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=153.02026.02 | 58.8 | — | 34.8 | — | — | — | 26.4 | 46.3 | 53.2 | — | — | — | |
| BaseZooming Type=Patch-based zooming (Pat.Z.), F.P. (GFlops)=202.12026.02 | 57.7 | — | 33.9 | — | — | — | 25.5 | 42.7 | 43.7 | — | — | — | |
| LAE-DINOBackbone=Swin-T2026.07 | 56.4 | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO-WorldBackbone=YOLOv8L2026.07 | 55.3 | — | — | — | — | — | — | — | — | — | — | — | |
| YOLC(k=1)(Ours)Backbone=HRNet + MoonNet(1+tanh)2025.09 | 55 | — | 32.9 | 33.6 | — | — | — | — | — | — | — | — | |
| ClusDetBackbone=ResNeXt1012025.09 | 53.2 | — | 28.4 | 26.4 | — | — | — | — | — | — | — | — | |
| YOLC(k=1)(Ours)Backbone=HRNet + MoonNet(sigmoid)2025.09 | 53 | — | 31 | 31.1 | — | — | — | — | — | — | — | — | |
| YOLC(k=1)Backbone=HRNet2025.09 | 51.6 | — | 30.3 | 30.7 | — | — | — | — | — | — | — | — | |
| CEASCBase Detector=GFL V1 [21], Backbone=ResNet182023.03 | 50.7 | — | 28.7 | 28.4 | 150.18 | 21.55 | — | — | — | — | — | — | |
| CEASCBackbone=ResNet-502026.07 | 50.7 | — | 28.7 | — | — | — | — | — | — | — | — | — | |
| ClusDetBackbone=ResNet-502026.07 | 50.6 | — | 26.7 | — | — | — | — | — | — | — | — | — | |
| ClusDetBackbone=ResNet1012025.09 | 50.4 | — | 26.7 | 25.2 | — | — | — | — | — | — | — | — | |
| MobileNet V2Base Detector=GFL V1 [21], Backbone=MobileNet V22023.03 | 50.2 | — | 28.5 | 28.1 | 491.47 | 13.63 | — | — | — | — | — | — | |
| BaselineBase Detector=GFL V1 [21], Backbone=ResNet182023.03 | 50 | — | 28.4 | 27.8 | 524.95 | 13.46 | — | — | — | — | — | — | |
| +Both (ZoomDet + RAF.)Zooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=77.82026.02 | 49.3 | — | 30.9 | — | — | — | 21 | 45.3 | 57.9 | — | — | — | |
| DMNetBackbone=ResNext1012025.09 | 49.3 | — | 29.4 | 30.6 | — | — | — | — | — | — | — | — | |
| ZoomDetZooming Type=Uniform zooming (Uni.Z.), F.P. (GFlops)=21.22026.02 | 48.5 | — | 27.5 | — | — | — | 16.9 | 42.5 | 60.7 | — | — | — | |
| RAF.Zooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=72.02026.02 | 48.5 | — | 30 | — | — | — | 19.5 | 44.1 | 58.6 | — | — | — | |
| DMNetBackbone=ResNet1012025.09 | 48.1 | — | 28.5 | 29.4 | — | — | — | — | — | — | — | — | |
| QueryDetBackbone=ResNet-502026.07 | 48.1 | — | 28.3 | — | — | — | — | — | — | — | — | — | |
| Fe.-S.Zooming Type=Uniform zooming (Uni.Z.), F.P. (GFlops)=73.72026.02 | 47.8 | — | 27 | — | — | — | 16.6 | 41.8 | 59.8 | — | — | — | |
| +Both (ZoomDet + Rem.)Zooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=39.62026.02 | 47.5 | — | 29.2 | — | — | — | 19.4 | 43.7 | 55.8 | — | — | — | |
| ZoomDetZooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=74.12026.02 | 47 | — | 28.9 | — | — | — | 18.6 | 43 | 57.1 | — | — | — | |
| ShuffleNet V2Base Detector=GFL V1 [21], Backbone=ShuffleNet V22023.03 | 46.6 | — | 26.2 | 25.7 | 488.94 | 13.92 | — | — | — | — | — | — | |
| Rem.Zooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=34.42026.02 | 46.1 | — | 28.2 | — | — | — | 18.2 | 41.7 | 51 | — | — | — | |
| BaseZooming Type=Uniform zooming (Uni.Z.), F.P. (GFlops)=16.42026.02 | 44.9 | — | 25.5 | — | — | — | 14.4 | 39.9 | 57.2 | — | — | — | |
| ZoomDetZooming Type=Uniform zooming (Uni.Z.), F.P. (GFlops)=128.92026.02 | 44.6 | — | 26.1 | — | — | — | 15.5 | 40.2 | 57 | — | — | — | |
| ZoomDetBase Detector=YOLOv8, FLOPs=48.22026.02 | 44.4 | — | 25.9 | — | — | — | 15.7 | 39.1 | 57.8 | 1.24 | 1.18 | 1.05 | |
| SR4I.Zooming Type=Uniform zooming (Uni.Z.), F.P. (GFlops)=126.62026.02 | 44.2 | — | 25.8 | — | — | — | 15.5 | 39.7 | 57.1 | — | — | — | |
| ZoomDetZooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=45.32026.02 | 43.8 | — | 26.8 | — | — | — | 17 | 39.9 | 50.2 | — | — | — | |
| BaseZooming Type=Uniform zooming (Uni.Z.), F.P. (GFlops)=123.12026.02 | 43.7 | — | 25 | — | — | — | 13.9 | 39 | 56.7 | — | — | — | |
| BaseZooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=68.22026.02 | 43.5 | — | 26.6 | — | — | — | 15.9 | 39.2 | 48.6 | — | — | — | |
| SaliencyBase Detector=YOLOv8, Strategy=saliency parametrization, FLOPs=48.12026.02 | 43.4 | — | 25.1 | — | — | — | 14.2 | 38.5 | 56.9 | 1.2 | 1.17 | 1.11 | |
| FOVEABase Detector=YOLOv8, FLOPs=48.62026.02 | 43.3 | — | 25.4 | — | — | — | 14.8 | 38.8 | 57.9 | 1.18 | 1.13 | 1.07 | |
| LZUBase Detector=YOLOv8, FLOPs=49.02026.02 | 42.8 | — | 24.9 | — | — | — | 14.1 | 38.5 | 58.5 | — | — | — | |
| UniformBase Detector=YOLOv8, Strategy=uniform downsampling baseline, FLOPs=43.72026.02 | 42.6 | — | 24.6 | — | — | — | 13.7 | 38.3 | 58.2 | 1 | 1 | 1 | |
| BaseZooming Type=Implicit zooming (Imp.Z.), F.P. (GFlops)=39.52026.02 | 40.5 | — | 24.4 | — | — | — | 14.5 | 36.7 | 46 | — | — | — | |
| BaselineBase Detector=RetinaNet [22], Backbone=ResNet502023.03 | 36.9 | — | 20.2 | 19.5 | 586.77 | 10.27 | — | — | — | — | — | — | |
| QueryDetBase Detector=RetinaNet [22], Backbone=ResNet502023.03 | 35.7 | — | 19.6 | 19 | — | 10.65 | — | — | — | — | — | — | |
| QueryDet-CSQBase Detector=RetinaNet [22], Backbone=ResNet502023.03 | 35 | — | 19.3 | 18.9 | — | 11.71 | — | — | — | — | — | — | |
| CEASCBase Detector=RetinaNet [22], Backbone=ResNet502023.03 | 35 | — | 20.8 | 27.7 | 201.96 | 14.27 | — | — | — | — | — | — | |
| RS-MPODPrompt=visual-322026.02 | 5.5 | — | — | — | — | — | — | — | — | — | — | — | |
| GroundingDINOPrompt=text2026.02 | 4.7 | — | — | — | — | — | — | — | — | — | — | — | |
| RS-MPODPrompt=multimodal2026.02 | 4.6 | — | — | — | — | — | — | — | — | — | — | — | |
| RS-MPODPrompt=text2026.02 | 3.8 | — | — | — | — | — | — | — | — | — | — | — | |
| LAE-DINOPrompt=text2026.02 | 2.3 | — | — | — | — | — | — | — | — | — | — | — |