Object Detection on MS COCO (val)
0.603mAPDyHead+
Evaluation results
| Method | Links | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DyHead+Backbone=Swin-L, Iteration=2x, Extra data=true2021.06 | 0.603 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DyHeadBackbone=Swin-L, Iteration=2x, Extra data=false2021.06 | 0.584 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HTC++Backbone=Swin-L, Iteration=6x, Extra data=false2021.06 | 0.58 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CopyPasteBackbone=Efficient-B7, Iteration=8x, Extra data=true2021.06 | 0.57 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterNet2+Backbone=Res2Net-101-DCN, Iteration=8x, Extra data=false2021.06 | 0.561 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv10-LFLOPs(G)=120.3, Parameters (M)=24.42026.06 | 0.532 | — | 0.701 | — | 0.572 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO26Variant=l, CPU ONNX (ms)=286.2, T4 TensorRT (ms)=6.5, Params (M)=24.8, FLOPs (B)=86.42025.10 | 0.53 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv8-LFLOPs(G)=165.2, Parameters (M)=43.72026.06 | 0.53 | — | 0.698 | — | 0.577 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv9-CFLOPs(G)=102.1, Parameters (M)=25.32026.06 | 0.53 | — | 0.702 | — | 0.578 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-LFLOPs(G)=88.9, Parameters (M)=26.42026.06 | 0.53 | — | 0.7 | — | 0.579 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-MFLOPs=59.8 G, Latency=4.27 ms, Param.=19.6 M2025.05 | 0.526 | — | 0.695 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-LFLOPs(G)=86.9, Parameters (M)=25.32026.06 | 0.523 | — | 0.692 | — | 0.557 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11Variant=l, CPU ONNX (ms)=308.7, T4 TensorRT (ms)=4.8, Params (M)=24.8, FLOPs (B)=53.42025.10 | 0.522 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-SoftHGNN-MFLOPs=80.2 G, Latency=6.01 ms, Param.=26.3 M2025.05 | 0.522 | — | 0.692 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-SoftHGNN-M⋆FLOPs=73.6 G, Latency=5.09 ms, Param.=22.3 M2025.05 | 0.522 | — | 0.691 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hyper-YOLO-MFLOPs=103.3 G, Latency=9.0 ms, Param.=33.3 M2025.05 | 0.52 | — | 0.69 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-M†FLOPs=59.8 G, Latency=4.75 ms, Param.=19.6 M2025.05 | 0.519 | — | 0.686 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv6-3.0-LFLOPs(G)=150.7, Parameters (M)=59.62026.06 | 0.518 | — | 0.692 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gold-YOLO-LFLOPs(G)=151.7, Parameters (M)=75.12026.06 | 0.518 | — | 0.689 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO26Variant=m, CPU ONNX (ms)=220.0, T4 TensorRT (ms)=4.9, Params (M)=20.4, FLOPs (B)=68.2, Inference Mode=End-to-End (e2e)2025.10 | 0.517 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO26Variant=m, CPU ONNX (ms)=220.0, T4 TensorRT (ms)=4.9, Params (M)=20.4, FLOPs (B)=68.2, Inference Mode=Standard (NMS)2025.10 | 0.515 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-MFLOPs=68.0 G, Latency=4.70 ms, Param.=20.1 M2025.05 | 0.515 | — | 0.685 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv9-MFLOPs=76.3 G, Latency=6.18 ms, Param.=20.0 M2025.05 | 0.514 | — | 0.681 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv10-MFLOPs=59.1 G, Latency=4.74 ms, Param.=15.4 M2025.05 | 0.511 | — | 0.681 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-SoftHGNN-MFLOPs=85.0 G, Latency=6.07 ms, Param.=25.1 M2025.05 | 0.511 | — | 0.681 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-SoftHGNN-M⋆FLOPs=81.7 G, Latency=5.82 ms, Param.=22.8 M2025.05 | 0.51 | — | 0.68 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv8Variant=m, CPU ONNX (ms)=197.5, T4 TensorRT (ms)=3.30, Params (M)=25.9, FLOPs (B)=63.32025.10 | 0.505 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-M†FLOPs=68.0 G, Latency=5.04 ms, Param.=20.1 M2025.05 | 0.505 | — | 0.673 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11Variant=m, CPU ONNX (ms)=171.0, T4 TensorRT (ms)=3.0, Params (M)=20.0, FLOPs (B)=44.82025.10 | 0.503 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv8-MFLOPs=78.9 G, Latency=5.09 ms, Param.=25.9 M2025.05 | 0.503 | — | 0.672 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gold-YOLO-MFLOPs=87.5 G, Latency=6.38 ms, Param.=41.3 M2025.05 | 0.498 | — | 0.67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| REGO-Deformable DETR**Backbone=X101, Epochs=50, GFLOPs=434, #Params (M)=119, iterative box refinement and two-stage processing=true2021.12 | 0.491 | — | 0.675 | — | 0.531 | 0.3 | 0.526 | 0.65 | — | — | — | — | — | — | — | — | — | — | — | |
| REGO-Deformable DETR**Backbone=R101, Epochs=50, GFLOPs=257, #Params (M)=73, iterative box refinement and two-stage processing=true2021.12 | 0.485 | — | 0.67 | — | 0.524 | 0.295 | 0.52 | 0.644 | — | — | — | — | — | — | — | — | — | — | — | |
| Cascade-RCNNBackbone=ResNeSt1012020.04 | 0.483 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hyper-YOLO-SFLOPs=39.0 G, Latency=4.7 ms, Param.=14.8 M2025.05 | 0.48 | — | 0.651 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-SoftHGNN-SFLOPs=25.2 G, Latency=3.17 ms, Param.=11.4 M2025.05 | 0.48 | — | 0.648 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| KAConvNet-L2026.04 | 0.48 | — | 0.657 | — | 0.519 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| IPT-T2026.04 | 0.479 | — | 0.655 | — | 0.518 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HVT Base2026.04 | 0.478 | — | 0.656 | — | 0.515 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Deformable DETR**Backbone=X101, Epochs=50, GFLOPs=417, #Params (M)=105, iterative box refinement and two-stage processing=true, reproduced=true2021.12 | 0.477 | — | — | — | 0.514 | 0.293 | 0.512 | 0.628 | — | — | — | — | — | — | — | — | — | — | — | |
| REGO-Deformable DETR**Backbone=R50, Epochs=50, GFLOPs=190, #Params (M)=54, iterative box refinement and two-stage processing=true2021.12 | 0.476 | — | 0.668 | — | 0.516 | 0.296 | 0.506 | 0.623 | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO26Variant=s, CPU ONNX (ms)=87.2, T4 TensorRT (ms)=2.7, Params (M)=9.5, FLOPs (B)=20.7, Inference Mode=End-to-End (e2e)2025.10 | 0.476 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-SFLOPs=19.4 G, Latency=2.42 ms, Param.=9.1 M2025.05 | 0.476 | — | 0.641 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EfficientNet-B22026.04 | 0.476 | — | 0.653 | — | 0.514 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MobileOne-S42026.04 | 0.475 | — | 0.651 | — | 0.514 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Deformable DETR**Backbone=R101, Epochs=50, GFLOPs=240, #Params (M)=59, iterative box refinement and two-stage processing=true, reproduced=true2021.12 | 0.472 | — | 0.666 | — | 0.511 | 0.285 | 0.509 | 0.624 | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO26Variant=s, CPU ONNX (ms)=87.2, T4 TensorRT (ms)=2.7, Params (M)=9.5, FLOPs (B)=20.7, Inference Mode=Standard (NMS)2025.10 | 0.472 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-SoftHGNN-SFLOPs=27.4 G, Latency=3.16 ms, Param.=12.1 M2025.05 | 0.471 | — | 0.638 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-SFLOPs(G)=21.4, Parameters (M)=9.32026.06 | 0.471 | — | 0.642 | — | 0.51 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11Variant=s, CPU ONNX (ms)=90.0, T4 TensorRT (ms)=1.9, Params (M)=9.2, FLOPs (B)=16.72025.10 | 0.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv12-S†FLOPs=19.4 G, Latency=2.49 ms, Param.=9.1 M2025.05 | 0.47 | — | 0.638 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-SFLOPs=21.5 G, Latency=2.50 ms, Param.=9.4 M2025.05 | 0.469 | — | 0.639 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CPNBackbone=HG-104, Input Size=ori., Flip=true2020.07 | 0.468 | — | — | 7.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv9-SFLOPs=26.4 G, Latency=3.46 ms, Param.=7.1 M2025.05 | 0.468 | — | 0.634 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv9-SFLOPs(G)=26.4, Parameters (M)=7.12026.06 | 0.468 | — | 0.634 | — | 0.507 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Cascade Mask RCNNBackbone=X101, Epochs=36, GFLOPs=627, #Params (M)=1352021.12 | 0.466 | — | 0.651 | — | 0.506 | 0.293 | 0.505 | 0.601 | — | — | — | — | — | — | — | — | — | — | — | |
| AN (w/ BN)Architecture=AOGNet40M, Normalization=Attentive Normalization with BN, Framework=Cascade Mask R-CNN, Head #Params=92.58M2019.08 | 0.465 | — | 0.65 | — | 0.508 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Deformable DETR**Backbone=R50, Epochs=50, GFLOPs=173, #Params (M)=40, iterative box refinement and two-stage processing=true, reproduced=true2021.12 | 0.464 | — | 0.653 | — | 0.506 | 0.3 | 0.498 | 0.614 | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv10-SFLOPs=21.6 G, Latency=2.49 ms, Param.=7.2 M2025.05 | 0.463 | — | 0.63 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv10-SFLOPs(G)=21.6, Parameters (M)=7.22026.06 | 0.463 | — | 0.63 | — | 0.504 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Cascade-RCNNBackbone=ResNeSt502020.04 | 0.4619 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| REGO-Deformable DETRBackbone=R50, Epochs=50, GFLOPs=190, #Params (M)=542021.12 | 0.459 | — | 0.652 | — | 0.497 | 0.276 | 0.489 | 0.615 | — | — | — | — | — | — | — | — | — | — | — | |
| AN (w/ BN)Architecture=ResNet101, Normalization=Attentive Normalization with BN, Framework=Cascade Mask R-CNN, Head #Params=96.77M2019.08 | 0.458 | — | 0.643 | — | 0.498 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| KAConvNet-B2026.04 | 0.458 | — | 0.631 | — | 0.496 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-SFLOPs(G)=21.5, Parameters (M)=9.42026.06 | 0.458 | — | 0.626 | — | 0.498 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Mask-RCNNBackbone=ResNeSt1012020.04 | 0.4575 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MobileVit-S2026.04 | 0.457 | — | 0.63 | — | 0.495 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PVT-Tiny2026.04 | 0.457 | — | 0.628 | — | 0.495 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BNArchitecture=AOGNet40M, Normalization=Batch Normalization, Framework=Cascade Mask R-CNN, Head #Params=92.35M2019.08 | 0.456 | — | 0.639 | — | 0.497 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLO11-S†FLOPs=21.5 G, Latency=2.57 ms, Param.=9.4 M2025.05 | 0.456 | — | 0.622 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gold-YOLO-SFLOPs=46.0 G, Latency=3.82 ms, Param.=21.5 M2025.05 | 0.454 | — | 0.625 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gold-YOLO-SFLOPs(G)=46, Parameters (M)=21.52026.06 | 0.454 | — | 0.625 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TSP-RCNNBackbone=R50, Epochs=96, GFLOPs=188, #Params (M)=null2021.12 | 0.45 | — | 0.645 | — | 0.496 | 0.297 | 0.477 | 0.58 | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv8-SFLOPs=28.6 G, Latency=2.33 ms, Param.=11.2 M2025.05 | 0.45 | — | 0.618 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv8-SFLOPs(G)=28.6, Parameters (M)=11.22026.06 | 0.45 | — | 0.618 | — | 0.487 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv8Variant=s, CPU ONNX (ms)=128.4, T4 TensorRT (ms)=1.90, Params (M)=11.2, FLOPs (B)=25.92025.10 | 0.449 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterNetBackbone=HG-104, Input Size=ori., Flip=true2020.07 | 0.448 | — | — | 5.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Cascade-RCNNBackbone=ResNet1012020.04 | 0.448 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| REGO-Deformable DETRBackbone=R50, Epochs=36, GFLOPs=190, #Params (M)=542021.12 | 0.448 | — | 0.638 | — | 0.487 | 0.27 | 0.48 | 0.602 | — | — | — | — | — | — | — | — | — | — | — | |
| Cascade-RCNNBackbone=ResNet101 [60]2020.04 | 0.4479 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Mask RCNNBackbone=X101, Epochs=36, GFLOPs=457, #Params (M)=1022021.12 | 0.445 | — | 0.649 | — | 0.487 | 0.276 | 0.483 | 0.577 | — | — | — | — | — | — | — | — | — | — | — | |
| Sparse RCNNBackbone=R50, Epochs=36, GFLOPs=null, #Params (M)=null2021.12 | 0.445 | — | 0.634 | — | 0.482 | 0.269 | 0.472 | 0.595 | — | — | — | — | — | — | — | — | — | — | — | |
| BNArchitecture=ResNet101, Normalization=Batch Normalization, Framework=Cascade Mask R-CNN, Head #Params=96.32M2019.08 | 0.444 | — | 0.625 | — | 0.484 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SMCABackbone=R101, Epochs=50, GFLOPs=218, #Params (M)=582021.12 | 0.444 | — | 0.652 | — | 0.48 | 0.243 | 0.485 | 0.61 | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv6-3.0-SFLOPs(G)=45.3, Parameters (M)=18.52026.06 | 0.443 | — | 0.612 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Anchor DETRFeature=DC5, Backbone=ResNet-50, Training Epochs=50, Batch size=1, GFLOPS=172, FPS=192021.09 | 0.442 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Efficient DETRBackbone=R50, Epochs=36, GFLOPs=159, #Params (M)=352021.12 | 0.442 | — | 0.622 | — | 0.48 | 0.284 | 0.475 | 0.566 | — | — | — | — | — | — | — | — | — | — | — | |
| Anchor DETRBackbone=R50, Epochs=50, GFLOPs=151, #Params (M)=null2021.12 | 0.442 | — | 0.647 | — | 0.475 | 0.247 | 0.482 | 0.606 | — | — | — | — | — | — | — | — | — | — | — | |
| REGO-DETR*-DC5Backbone=R50, Epochs=50, GFLOPs=213, #Params (M)=58, 300 queries, reference points, and focal loss=true2021.12 | 0.44 | — | 0.626 | — | 0.478 | 0.265 | 0.452 | 0.629 | — | — | — | — | — | — | — | — | — | — | — | |
| CPNBackbone=HG-52, Input Size=ori., Flip=true2020.07 | 0.438 | — | — | 9.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Deformable DETRFeature=multi-level, Backbone=ResNet-50, Training Epochs=50, Batch size=1, GFLOPS=173, FPS=152021.09 | 0.438 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Conditional DETRFeature=DC5, Backbone=ResNet-50, Training Epochs=50, Batch size=1, GFLOPS=195, FPS=102021.09 | 0.438 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Deformable DETRBackbone=R50, Epochs=50, GFLOPs=173, #Params (M)=402021.12 | 0.438 | — | 0.626 | — | 0.477 | 0.264 | 0.471 | 0.58 | — | — | — | — | — | — | — | — | — | — | — | |
| SMCAFeature=multi-level, Backbone=ResNet-50, Training Epochs=50, Batch size=1, GFLOPS=152, FPS=102021.09 | 0.437 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SMCABackbone=R50, Epochs=50, GFLOPs=152, #Params (M)=402021.12 | 0.437 | — | 0.636 | — | 0.472 | 0.242 | 0.47 | 0.604 | — | — | — | — | — | — | — | — | — | — | — | |
| DETRFeature=DC5, Backbone=ResNet-50, Training Epochs=500, Batch size=1, GFLOPS=187, FPS=122021.09 | 0.433 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DETR-DC5Backbone=R50, Epochs=500, GFLOPs=187, #Params (M)=412021.12 | 0.433 | — | 0.631 | — | 0.459 | 0.225 | 0.473 | 0.611 | — | — | — | — | — | — | — | — | — | — | — | |
| KAConvNet-S2026.04 | 0.433 | — | 0.606 | — | 0.471 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HTCBackbone=ResNet502020.04 | 0.432 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MobileOne-S02026.04 | 0.431 | — | 0.604 | — | 0.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — |