Object Detection on VisDrone (val)
63.1AP50FRFDet-X†
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| FRFDet-X†Input Size=1024×1024, Test-set=o + ca, Params (M)=38.9, FLOPs (G)=272.2, FPS=45.82026.07 | 63.1 | 41.5 | — | — | — | — | — | — | — | — | — | |
| FRFDet-XInput Size=640×640, Test-set=ca, Params (M)=38.9, FLOPs (G)=106.3, FPS=94.62026.07 | 63 | 40.9 | — | — | — | — | — | — | — | — | — | |
| UFPMP w/ DN-FPNCropping Strategy=MP2024.06 | 62.7 | 37.8 | 38.6 | — | — | — | — | — | — | — | — | |
| UFPMPCropping Strategy=MP2024.06 | 62.4 | 36.6 | 36.7 | — | — | — | — | — | — | — | — | |
| UFPMP-DetInput Size=1333×800, Test-set=o + ca, FLOPs (G)=150.2, FPS=21.72026.07 | 62.4 | 36.6 | — | — | — | — | — | — | — | — | — | |
| ESODInput Size=1333×800, Test-set=o + ca, FLOPs (G)=180.6, FPS=28.62026.07 | 62.3 | 37.9 | — | — | — | — | — | — | — | — | — | |
| RemDet-XInput Size=1024×640, Test-set=o + ca, Params (M)=74.1, FLOPs (G)=182.0, FPS=110.22026.07 | 61.9 | 40 | — | — | — | — | — | — | — | — | — | |
| SFFNet-XScale=Extra-large, Param(M)=38.5, FLOPs(G)=203.72026.04 | 59.3 | 36.8 | 38.5 | 13.6 | 39.5 | 51.4 | — | — | 29.6 | 47.4 | 45.4 | |
| SFFNet-LScale=Large, Param(M)=24.6, FLOPs(G)=131.02026.04 | 58.5 | 36.1 | 37.8 | 13.4 | 38.8 | 50.7 | — | — | 28.9 | 46.6 | 48.7 | |
| FRFDet-X†Input Size=1024×1024, Test-set=o, Params (M)=38.9, FLOPs (G)=272.2, FPS=45.82026.07 | 58.4 | 37.5 | — | — | — | — | — | — | — | — | — | |
| CZ Det.Detector=Faster R-CNN2023.03 | 58.3 | 33.22 | 33.16 | — | — | — | — | — | 26.06 | 42.58 | 43.36 | |
| CZDetCropping Strategy=Density2024.06 | 58.3 | 33.2 | 33.2 | — | — | — | — | — | — | — | — | |
| CZDetInput Size=1999×1200, Test-set=o + dc, FPS=8.42026.07 | 58.3 | 33.2 | — | — | — | — | — | — | — | — | — | |
| SFFNet-BScale=Basic, Param(M)=19.7, FLOPs(G)=105.62026.04 | 58.2 | 35.9 | 37.3 | 13.6 | 39 | 51 | — | — | 28.6 | 46.6 | 46.2 | |
| CascadeNetDetector=Faster R-CNN, Model Fusion=true2023.03 | 58.02 | 30.12 | 27.53 | — | — | — | — | — | — | — | — | |
| DNTRCropping Strategy=MP2024.06 | 57.9 | 34.4 | 35.3 | — | — | — | — | — | — | — | — | |
| SFFNet-MScale=Medium, Param(M)=14.2, FLOPs(G)=63.22026.04 | 57.5 | 35.3 | 36.6 | 13.2 | 38.5 | 50.4 | — | — | 28 | 46 | 42.9 | |
| YOLOv8-XScale=Extra-large, Param(M)=68.2, FLOPs(G)=257.8, reimplemented=true2026.04 | 57.3 | 35.6 | 37.2 | 13.5 | 38.9 | 50.1 | — | — | 27.2 | 47.6 | 44.5 | |
| YOLO11-XScale=Extra-large, Param(M)=56.9, FLOPs(G)=194.9, reimplemented=true2026.04 | 57.3 | 35.7 | 37.2 | 13.3 | 38.7 | 50.2 | — | — | 28 | 47.4 | 44.9 | |
| YOLOv10-XScale=Extra-large, Param(M)=29.5, FLOPs(G)=160.4, reimplemented=true2026.04 | 57.1 | 35.1 | 36.5 | 12.9 | 38.1 | 49.9 | — | — | 27.5 | 45.9 | 43.4 | |
| YOLOv10-LScale=Large, Param(M)=24.4, FLOPs(G)=120.3, reimplemented=true2026.04 | 56 | 34.4 | 35.9 | 13 | 37.6 | 49.3 | — | — | 26.8 | 45.1 | 42.2 | |
| GLSANCropping Strategy=Density2024.06 | 55.8 | 32.5 | 33 | — | — | — | — | — | — | — | — | |
| YOLOv8-LScale=Large, Param(M)=43.7, FLOPs(G)=165.2, reimplemented=true2026.04 | 55.7 | 34.6 | 35.8 | 13.3 | 38.2 | 49.3 | — | — | 26.5 | 46 | 45.7 | |
| YOLO11-LScale=Large, Param(M)=25.3, FLOPs(G)=86.9, reimplemented=true2026.04 | 55.6 | 34.5 | 35.7 | 13.2 | 38 | 49.2 | — | — | 26.6 | 45.5 | 45.8 | |
| GLSANDetector=Faster R-CNN2023.03 | 55.4 | 30.7 | 30 | — | — | — | — | — | — | — | — | |
| YOLOv10-BScale=Basic, Param(M)=19.1, FLOPs(G)=92.0, reimplemented=true2026.04 | 55.4 | 33.8 | 34.9 | 12.7 | 37.3 | 49.1 | — | — | 26.3 | 44.2 | 38.5 | |
| GLSANInput Size=1333×800, Test-set=o + ca2026.07 | 55.4 | 30.7 | — | — | — | — | — | — | — | — | — | |
| EFSI-DETR*Image Size=800, Latency (ms)=5.3, Params (M)=27.32026.01 | 55.2 | 35 | 36.9 | — | — | — | — | — | 27.3 | 45.3 | 43.3 | |
| YOLCInput Size=1024×640, Test-set=o + ca, FLOPs (G)=151.0, FPS=2.32026.07 | 55 | 31.8 | — | — | — | — | — | — | — | — | — | |
| YOLO11-MScale=Medium, Param(M)=20.1, FLOPs(G)=68.0, reimplemented=true2026.04 | 54.4 | 33.5 | 34.6 | 12.9 | 37.1 | 48.2 | — | — | 25.8 | 44.1 | 47.5 | |
| DNTRCropping Strategy=None2024.06 | 53.8 | 33.1 | 34.8 | — | — | — | — | — | — | — | — | |
| FMC-DETR-TInput size=640×640, Param=13.8M, FLOPs=146.8G2025.09 | 53.6 | 33.7 | — | — | — | — | — | — | — | — | — | |
| YOLOv8-MScale=Medium, Param(M)=25.9, FLOPs(G)=78.9, reimplemented=true2026.04 | 53.3 | 32.9 | 33.9 | 12.8 | 36.5 | 47.3 | — | — | 25 | 43.6 | 43.4 | |
| SFFNet-SScale=Small, Param(M)=6.3, FLOPs(G)=24.02026.04 | 52.8 | 31.6 | 32.3 | 12.1 | 35.7 | 47.4 | — | — | 25.3 | 40.8 | 35.7 | |
| EFSI-DETRImage Size=640, Latency (ms)=5.3, Params (M)=27.32026.01 | 52.7 | 33.1 | 34.7 | — | — | — | — | — | 24.8 | 44 | 44 | |
| AMRNetCropping Strategy=Adaptive2024.06 | 52.7 | 31.7 | 33.1 | — | — | — | — | — | — | — | — | |
| AMRNetInput Size=1500×800, Test-set=o + aug2026.07 | 52.7 | 31.7 | — | — | — | — | — | — | — | — | — | |
| YOLOv10-MScale=Medium, Param(M)=15.4, FLOPs(G)=59.1, reimplemented=true2026.04 | 52.6 | 31.9 | 32.8 | 12.3 | 35.8 | 47.4 | — | — | 24.7 | 42 | 40.7 | |
| SDPCropping Strategy=None2024.06 | 52.5 | 30.2 | 30.6 | — | — | — | — | — | — | — | — | |
| EFCInput Size=1333×800, Test-set=o, Params (M)=40.0, FLOPs (G)=477.62026.07 | 52.1 | 30.1 | — | — | — | — | — | — | — | — | — | |
| UAV-DETR-R50Image Size=640, Params (M)=422026.01 | 51.1 | 31.5 | — | — | — | — | — | — | — | — | — | |
| CEASCImage Size=1333, Latency (ms)=43.82026.01 | 50.7 | 28.7 | 28.4 | — | — | — | — | — | — | — | — | |
| CEASCCropping Strategy=None2024.06 | 50.7 | 28.7 | 28.4 | — | — | — | — | — | — | — | — | |
| CEASCInput Size=1333×800, Test-set=o, FLOPs (G)=150.2, FPS=21.72026.07 | 50.7 | 28.7 | — | — | — | — | — | — | — | — | — | |
| ClusterNetDetector=Faster R-CNN2023.03 | 50.63 | 26.72 | 24.7 | — | — | — | — | — | 17.61 | 38.92 | 51.4 | |
| ClusDetCropping Strategy=Cluster2024.06 | 50.6 | 26.7 | 24.7 | — | — | — | — | — | — | — | — | |
| ClusDetInput Size=1000×600, Test-set=o + ca, FLOPs (G)=436.0, FPS=6.52026.07 | 50.6 | 26.7 | — | — | — | — | — | — | — | — | — | |
| FSAFInput Size=1333×800, Test-set=o, FLOPs (G)=518.3, FPS=15.82026.07 | 50.3 | 26.3 | — | — | — | — | — | — | — | — | — | |
| GFLV1Image Size=1333, Latency (ms)=5252026.01 | 50 | 28.4 | 27.8 | — | — | — | — | — | — | — | — | |
| GFL V1Input Size=1333×800, Test-set=o, FLOPs (G)=525.0, FPS=13.52026.07 | 50 | 28.4 | — | — | — | — | — | — | — | — | — | |
| CDMNetDetector=Faster R-CNN2023.03 | 49.5 | 29.2 | 29.8 | — | — | — | — | — | 20.8 | 40.7 | 41.6 | |
| CDMNetInput Size=1000×600, Test-set=ca2026.07 | 49.5 | 29.2 | — | — | — | — | — | — | — | — | — | |
| HRDNetImage Size=13332026.01 | 49.3 | 28.3 | 28.2 | — | — | — | — | — | — | — | — | |
| DetectoRS*Cropping Strategy=MP2024.06 | 49.3 | 29.4 | 30.2 | — | — | — | — | — | — | — | — | |
| HRDNetInput Size=2666×1600, Test-set=o2026.07 | 49.3 | 28.3 | — | — | — | — | — | — | — | — | — | |
| UAV-DETR-R18Image Size=640, Params (M)=202026.01 | 48.8 | 29.8 | — | — | — | — | — | — | — | — | — | |
| FRFDet-XInput Size=640×640, Test-set=o, Params (M)=38.9, FLOPs (G)=106.3, FPS=94.62026.07 | 48.8 | 30.8 | — | — | — | — | — | — | — | — | — | |
| FBRT-YOLO-XPublication=AAAI2025, Input size=640×640, Param=22.8M, FLOPs=185.8G2025.09 | 48.4 | 30.1 | — | — | — | — | — | — | — | — | — | |
| FMC-DETR-BInput size=640×640, Param=17.4M, FLOPs=62.9G2025.09 | 48.4 | 29.4 | — | — | — | — | — | — | — | — | — | |
| RemDet-XInput Size=640×640, Test-set=o, Params (M)=74.1, FLOPs (G)=112.0, FPS=110.22026.07 | 48.3 | 29.9 | — | — | — | — | — | — | — | — | — | |
| QueryDetDetector=Faster R-CNN2023.03 | 48.14 | 28.32 | 28.75 | — | — | — | — | — | — | — | — | |
| QueryDetCropping Strategy=Evenly2024.06 | 48.1 | 28.3 | 28.8 | — | — | — | — | — | — | — | — | |
| QueryDetInput Size=1333×800, Test-set=o, FLOPs (G)=888.4, FPS=11.32026.07 | 48.1 | 28.3 | — | — | — | — | — | — | — | — | — | |
| CenterNetImage Size=800, Latency (ms)=95.22026.01 | 47.9 | 27.8 | 27.6 | — | — | — | — | — | 21.3 | 42.1 | 49.8 | |
| DEIM-RT-DETRv2-R50Image Size=640, Latency (ms)=1.9, Params (M)=422026.01 | 47.7 | 28.2 | 28.1 | — | — | — | — | — | 19 | 39.2 | 47.9 | |
| FBRT-YOLO-LPublication=AAAI2025, Input size=640×640, Param=14.6M, FLOPs=119.2G2025.09 | 47.7 | 29.7 | — | — | — | — | — | — | — | — | — | |
| DensityMapDetector=Faster R-CNN2023.03 | 47.62 | 28.21 | 28.9 | — | — | — | — | — | 19.9 | 39.61 | 55.81 | |
| DMNetInput Size=1000×600, Test-set=o + dc, FLOPs (G)=471.4, FPS=5.92026.07 | 47.6 | 28.2 | — | — | — | — | — | — | — | — | — | |
| RemDet-LImage Size=640, Latency (ms)=7.1, Params (M)=35.32026.01 | 47.4 | 29.3 | 30.3 | — | — | — | — | — | 18.7 | 43.4 | 55.8 | |
| RemDet-LInput Size=640×640, Test-set=o, Params (M)=35.3, FLOPs (G)=66.7, FPS=138.42026.07 | 47.4 | 29.3 | — | — | — | — | — | — | — | — | — | |
| FRFDet-LInput Size=640×640, Test-set=o, Params (M)=22.6, FLOPs (G)=57.2, FPS=78.42026.07 | 47.3 | 29.8 | — | — | — | — | — | — | — | — | — | |
| CascadeNetDetector=Faster R-CNN, Model Fusion=false2023.03 | 47.1 | 28.8 | 29.3 | — | — | — | — | — | — | — | — | |
| RT-DETR-R50Publication=CVPR2024, Input size=640×640, Param=42.0M, FLOPs=136.0G2025.09 | 47 | 28.4 | — | — | — | — | — | — | — | — | — | |
| YOLOv8-XPublication=Ultralytics, Input size=640×640, Param=68.2M, FLOPs=257.8G2025.09 | 46.8 | 28.9 | — | — | — | — | — | — | — | — | — | |
| FRFDet-MInput Size=640×640, Test-set=o, Params (M)=17.4, FLOPs (G)=47.7, FPS=98.82026.07 | 46.7 | 29.2 | — | — | — | — | — | — | — | — | — | |
| YOLOv8-SScale=Small, Param(M)=11.2, FLOPs(G)=28.6, reimplemented=true2026.04 | 46.6 | 27.8 | 28.2 | 11.1 | 32.2 | 42.6 | — | — | 20.6 | 37 | 35.8 | |
| YOLO11-SScale=Small, Param(M)=9.4, FLOPs(G)=21.5, reimplemented=true2026.04 | 46.5 | 27.8 | 28.1 | 11 | 32.4 | 42.8 | — | — | 20.5 | 37.3 | 34.9 | |
| YOLOv10-SScale=Small, Param(M)=7.2, FLOPs(G)=21.6, reimplemented=true2026.04 | 46.2 | 27.2 | 27.5 | 10.7 | 31.9 | 43.2 | — | — | 20.8 | 35.6 | 31.9 | |
| RemDet-MImage Size=640, Params (M)=23.32026.01 | 46.1 | 28.2 | 28.9 | — | — | — | — | — | 18.2 | 41.7 | 51 | |
| YOLOv10-XPublication=NeurIPS2024, Input size=640×640, Param=29.5M, FLOPs=160.4G2025.09 | 46.1 | 28.7 | — | — | — | — | — | — | — | — | — | |
| RemDet-MInput Size=640×640, Test-set=o, Params (M)=23.3, FLOPs (G)=34.42026.07 | 46.1 | 28.2 | — | — | — | — | — | — | — | — | — | |
| YOLOv8-LPublication=Ultralytics, Input size=640×640, Param=43.7M, FLOPs=165.2G2025.09 | 45.9 | 28.4 | — | — | — | — | — | — | — | — | — | |
| FBRT-YOLO-MPublication=AAAI2025, Input size=640×640, Param=7.2M, FLOPs=58.7G2025.09 | 45.9 | 28.4 | — | — | — | — | — | — | — | — | — | |
| YOLOv12-XImage Size=640, Latency (ms)=6.5, Params (M)=59.12026.01 | 45.7 | 28.1 | 29.1 | — | — | — | — | — | 17.9 | 41.5 | 46.5 | |
| DEIM-RT-DETRv2-R18Image Size=640, Latency (ms)=1.2, Params (M)=19.92026.01 | 45.7 | 27.1 | 27.1 | — | — | — | — | — | 18.8 | 37.1 | 46.8 | |
| SFFNet-NScale=Nano, Param(M)=1.7, FLOPs(G)=7.22026.04 | 45.4 | 26.7 | 26.8 | 10.4 | 31.5 | 42.6 | — | — | 19.8 | 35.3 | 35 | |
| YOLO11-XImage Size=640, Latency (ms)=5.8, Params (M)=56.82026.01 | 45.1 | 27.7 | 28.4 | — | — | — | — | — | 18.1 | 40.4 | 47.6 | |
| RetinaNetCropping Strategy=Evenly2024.06 | 44.9 | 26.2 | 27.1 | — | — | — | — | — | — | — | — | |
| RT-DETR-R50Image Size=640, Latency (ms)=5.1, Params (M)=422026.01 | 44.8 | 26.8 | 26.9 | — | — | — | — | — | 18.4 | 37.1 | 43.9 | |
| YOLOv8-XImage Size=640, Latency (ms)=4.9, Params (M)=68.22026.01 | 44.6 | 27.2 | 27.7 | — | — | — | — | — | 17.3 | 39.8 | 46.3 | |
| YOLOv10-LPublication=NeurIPS2024, Input size=640×640, Param=24.4M, FLOPs=120.3G2025.09 | 44.6 | 27.6 | — | — | — | — | — | — | — | — | — | |
| RT-DETR-R18Publication=CVPR2024, Input size=640×640, Param=20.0M, FLOPs=60.0G2025.09 | 44.6 | 26.7 | — | — | — | — | — | — | — | — | — | |
| YOLOv10-XImage Size=640, Latency (ms)=4.5, Params (M)=29.42026.01 | 44.5 | 27.2 | 28 | — | — | — | — | — | 17.4 | 40.3 | 48.3 | |
| YOLO11-MInput Size=640×640, Test-set=o, Params (M)=20.0, FLOPs (G)=33.8, FPS=144.02026.07 | 44.5 | 27.3 | — | — | — | — | — | — | — | — | — | |
| Yolov5Cropping Strategy=None2024.06 | 44.1 | 24.1 | 24.7 | — | — | — | — | — | — | — | — | |
| YOLO11-LImage Size=640, Latency (ms)=4.2, Params (M)=25.32026.01 | 44 | 27 | 27.6 | — | — | — | — | — | 16.5 | 40.4 | 51.6 | |
| YOLOv11-XPublication=Ultralytics, Input size=640×640, Param=56.9M, FLOPs=194.9G2025.09 | 43.8 | 26.6 | — | — | — | — | — | — | — | — | — | |
| SAHIInput Size=1333×800, Test-set=o, Params (M)=26.12026.07 | 43.5 | 25.6 | — | — | — | — | — | — | — | — | — | |
| YOLOv8-LImage Size=640, Latency (ms)=4.2, Params (M)=43.62026.01 | 43.4 | 26.5 | 27 | — | — | — | — | — | 16.1 | 39.5 | 46.1 | |
| RT-DETR-R18Image Size=640, Latency (ms)=3.6, Params (M)=19.92026.01 | 43.3 | 25.7 | 25.9 | — | — | — | — | — | 17.5 | 35.8 | 39.4 |