Pose Estimation on COCO
78.2mAPHAP
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| HAPBackbone=ViT-B2026.01 | 78.2 | — | — | — | |
| CLASP (HRFormer)Backbone=Swin-B2026.01 | 77.8 | 82.2 | — | — | |
| HRFormer2023.03 | 77.2 | 82 | — | — | |
| HRFormerBackbone=HRFormer-B2026.01 | 77.2 | 82 | — | — | |
| SoTA †Note=multitask or stronger models2023.03 | 77.1 | — | — | — | |
| PATHBackbone=ViT-L, Evaluation Protocol=Partial FT2023.03 | 77.1 | — | — | — | |
| HulkBackbone=ViT-B2026.01 | 77 | — | — | — | |
| CLASP (HRFormer)Backbone=Swin-S2026.01 | 76.8 | 81.8 | — | — | |
| SOLIDERBackbone=Swin-B2023.03 | 76.6 | 81.5 | — | — | |
| SOLIDERBackbone=Swin-B2026.01 | 76.6 | 81.5 | — | — | |
| UniHCPBackbone=ViT-B2026.01 | 76.5 | — | — | — | |
| PATHBackbone=ViT-B, Evaluation Protocol=FT2023.03 | 76.3 | — | — | — | |
| HRNet2023.03 | 76.3 | 81.2 | — | — | |
| SOLIDERBackbone=Swin-S2023.03 | 76.3 | 81.3 | — | — | |
| HRNetBackbone=HRNet-W482026.01 | 76.3 | 81.2 | — | — | |
| PATHBackbone=ViT-B2026.01 | 76.3 | — | — | — | |
| PATHBackbone=ViT-B, Evaluation Protocol=Partial FT2023.03 | 76 | — | — | — | |
| SoTA2023.03 | 75.8 | — | — | — | |
| MAEBackbone=ViT-B2023.03 | 75.8 | — | — | — | |
| TokenPose2023.03 | 75.8 | 80.9 | — | — | |
| CLASP (HRFormer)Backbone=Swin-T2026.01 | 75.7 | 80.3 | — | — | |
| PATHBackbone=ViT-B, Evaluation Protocol=Head FT2023.03 | 75.2 | — | — | — | |
| PATHBackbone=ViT-B, Evaluation Protocol=w/o FT2023.03 | 75 | — | — | — | |
| PATHBackbone=ViT-L, Evaluation Protocol=w/o FT2023.03 | 74.7 | — | — | — | |
| CLIPBackbone=ViT-B2023.03 | 74.4 | — | — | — | |
| SOLIDERBackbone=Swin-T2023.03 | 74.4 | 79.6 | — | — | |
| SimpleBase2023.03 | 74.3 | 79.7 | — | — | |
| SimpleBaselineBackbone=ViTAE-S-Stage, InputSize=256x192, Params (M)=272021.06 | 73.7 | 79 | — | — | |
| SimpleBaselineBackbone=ResNet-50, InputSize=256x192, Params (M)=342021.06 | 71.8 | 77.3 | — | — | |
| CPN2023.03 | 68.6 | — | — | — | |
| YOLO26l-poseSize (px)=640, Speed CPU ONNX (ms)=275.4 ± 2.4, Speed T4 TRT10 (ms)=6.5 ± 0.1, Params (M)=25.9, FLOPs (B)=91.32025.09 | — | — | 70.4 | 90.5 | |
| YOLO26m-poseSize (px)=640, Speed CPU ONNX (ms)=218.0 ± 1.5, Speed T4 TRT10 (ms)=5.0 ± 0.1, Params (M)=21.5, FLOPs (B)=73.12025.09 | — | — | 68.8 | 89.6 | |
| YOLO26n-poseSize (px)=640, Speed CPU ONNX (ms)=40.3 ± 0.5, Speed T4 TRT10 (ms)=1.8 ± 0.0, Params (M)=2.9, FLOPs (B)=7.52025.09 | — | — | 57.2 | 83.3 | |
| YOLO26s-poseSize (px)=640, Speed CPU ONNX (ms)=85.3 ± 0.9, Speed T4 TRT10 (ms)=2.7 ± 0.0, Params (M)=10.4, FLOPs (B)=23.92025.09 | — | — | 63 | 86.6 | |
| YOLO26x-poseSize (px)=640, Speed CPU ONNX (ms)=565.4 ± 3.0, Speed T4 TRT10 (ms)=12.2 ± 0.2, Params (M)=57.6, FLOPs (B)=201.72025.09 | — | — | 71.6 | 91.6 |