paper-with-me

Pose Estimation 벤치마크

Pose Estimation on COCO test-dev

141개 결과 · ⬇ CSV · JSON

AP

61.8 66.62 71.45 76.27 81.1 2016-11 2026-09 CMU-Pose — 61.8 (2016-11-24) CMU-Pose — 61.8 (2016-11-24) CMU-Pose — 61.8 (2016-11-24) RMPE++ — 72.3 (2016-12-01) RMPE — 61.8 (2016-12-01) RMPE++ — 72.3 (2016-12-01) RMPE — 61.8 (2016-12-01) RMPE++ — 72.3 (2016-12-01) RMPE — 61.8 (2016-12-01) G-RMI — 64.9 (2017-01-06) G-RMI — 64.9 (2017-01-06) G-RMI — 64.9 (2017-01-06) Mask-RCNN — 63.1 (2017-03-20) Mask-RCNN — 63.1 (2017-03-20) Mask-RCNN — 63.1 (2017-03-20) Faster R-CNN (ImageNet+300M) — 64.4 (2017-07-10) Faster R-CNN (ImageNet+300M) — 64.4 (2017-07-10) Faster R-CNN (ImageNet+300M) — 64.4 (2017-07-10) CPN+ [6, 9] — 73.0 (2017-11-20) CPN — 72.1 (2017-11-20) CPN+ [6, 9] — 73.0 (2017-11-20) CPN — 72.1 (2017-11-20) CPN+ [6, 9] — 73.0 (2017-11-20) CPN — 72.1 (2017-11-20) Flow-based (ResNet-152) — 73.7 (2018-04-17) Flow-based (ResNet-152) — 73.7 (2018-04-17) Flow-based (ResNet-152) — 73.7 (2018-04-17) PoseFix — 74.7 (2018-12-10) PoseFix — 74.7 (2018-12-10) PoseFix — 74.7 (2018-12-10) OpenPose — 64.2 (2018-12-18) OpenPose — 64.2 (2018-12-18) OpenPose — 64.2 (2018-12-18) MSPN — 76.1 (2019-01-01) MSPN — 76.1 (2019-01-01) MSPN — 76.1 (2019-01-01) HRNet-W48 + extra data — 77.0 (2019-02-25) HRNet-W48 + extra data — 77.0 (2019-02-25) HRNet-W48 + extra data — 77.0 (2019-02-25) HRNet-W48+DARK — 77.4 (2019-10-14) HRNet-W48+DARK — 77.4 (2019-10-14) HRNet-W48+DARK — 77.4 (2019-10-14) HRNet-W48+UDP — 76.5 (2019-11-18) DirectPose (ResNet-101) — 63.3 (2019-11-18) HRNet-W48+UDP — 76.5 (2019-11-18) DirectPose (ResNet-101) — 63.3 (2019-11-18) HRNet-W48+UDP — 76.5 (2019-11-18) DirectPose (ResNet-101) — 63.3 (2019-11-18) Simple Pose — 68.1 (2019-11-24) Simple Pose — 68.1 (2019-11-24) Simple Pose — 68.1 (2019-11-24) CCM+ — 78.9 (2020-02-03) CCM+ — 78.9 (2020-02-03) CCM+ — 78.9 (2020-02-03) 4xRSN-50 (ensemble) — 79.2 (2020-03-09) 4xRSN-50 — 78.6 (2020-03-09) 4xRSN-50 (ensemble) — 79.2 (2020-03-09) 4xRSN-50 — 78.6 (2020-03-09) 4xRSN-50 (ensemble) — 79.2 (2020-03-09) 4xRSN-50 — 78.6 (2020-03-09) EvoPose2D-L — 76.8 (2020-11-17) EvoPose2D-L — 76.8 (2020-11-17) EvoPose2D-L — 76.8 (2020-11-17) TransPose-H-A6 — 75.0 (2020-12-28) TransPose-H-A6 — 75.0 (2020-12-28) TransPose-H-A6 — 75.0 (2020-12-28) MIPNet — 75.7 (2021-01-27) MIPNet — 75.7 (2021-01-27) MIPNet — 75.7 (2021-01-27) OmniPose (WASPv2) — 76.4 (2021-03-18) OmniPose (WASPv2) — 76.4 (2021-03-18) OmniPose (WASPv2) — 76.4 (2021-03-18) TFPose (ND=6 ResNet-50) — 72.2 (2021-03-29) TFPose (ND=6 ResNet-50) — 72.2 (2021-03-29) TFPose (ND=6 ResNet-50) — 72.2 (2021-03-29) Lite-HRNet-30 — 69.7 (2021-04-13) Lite-HRNet-18 — 66.9 (2021-04-13) Lite-HRNet-30 — 69.7 (2021-04-13) Lite-HRNet-18 — 66.9 (2021-04-13) Lite-HRNet-30 — 69.7 (2021-04-13) Lite-HRNet-18 — 66.9 (2021-04-13) S-ViPNAS-HRNetW32 — 73.9 (2021-05-21) S-ViPNAS-Res50 — 70.3 (2021-05-21) S-ViPNAS-HRNetW32 — 73.9 (2021-05-21) S-ViPNAS-Res50 — 70.3 (2021-05-21) S-ViPNAS-HRNetW32 — 73.9 (2021-05-21) S-ViPNAS-Res50 — 70.3 (2021-05-21) UDP-Pose-PSA(384x288) — 79.5 (2021-07-02) UDP-Pose-PSA(256x192) — 78.9 (2021-07-02) UDP-Pose-PSA(384x288) — 79.5 (2021-07-02) UDP-Pose-PSA(256x192) — 78.9 (2021-07-02) UDP-Pose-PSA(384x288) — 79.5 (2021-07-02) UDP-Pose-PSA(256x192) — 78.9 (2021-07-02) HRFormer-B — 76.2 (2021-10-18) HRFormer-B — 76.2 (2021-10-18) HRFormer-B — 76.2 (2021-10-18) KAPAO-L — 70.3 (2021-11-16) KAPAO-M — 68.8 (2021-11-16) KAPAO-S — 63.8 (2021-11-16) KAPAO-L — 70.3 (2021-11-16) KAPAO-M — 68.8 (2021-11-16) KAPAO-S — 63.8 (2021-11-16) KAPAO-L — 70.3 (2021-11-16) KAPAO-M — 68.8 (2021-11-16) KAPAO-S — 63.8 (2021-11-16) Dite-HRNet-30 — 70.6 (2022-04-22) Dite-HRNet-30 — 70.6 (2022-04-22) Dite-HRNet-30 — 70.6 (2022-04-22) ViTPose (ViTAE-G, ensemble) — 81.1 (2022-04-26) ViTPose (ViTAE-G) — 80.9 (2022-04-26) ViTPose (ViTAE-G, ensemble) — 81.1 (2022-04-26) ViTPose (ViTAE-G) — 80.9 (2022-04-26) ViTPose (ViTAE-G, ensemble) — 81.1 (2022-04-26) ViTPose (ViTAE-G) — 80.9 (2022-04-26) SwinV2-L 1K-MIM — 77.2 (2022-05-26) SwinV2-B 1K-MIM — 76.7 (2022-05-26) SwinV2-L 1K-MIM — 77.2 (2022-05-26) SwinV2-B 1K-MIM — 76.7 (2022-05-26) SwinV2-L 1K-MIM — 77.2 (2022-05-26) SwinV2-B 1K-MIM — 76.7 (2022-05-26) PPE (ResNeXt-101) — 75.7 (2022-06-15) PPE (ResNeXt-101) — 75.7 (2022-06-15) PPE (ResNeXt-101) — 75.7 (2022-06-15) DPIT-L — 74.6 (2022-09-02) DPIT-L — 74.6 (2022-09-02) DPIT-L — 74.6 (2022-09-02) PCT (256x256) — 78.3 (2023-03-21) PCT (256x256) — 78.3 (2023-03-21) PCT (256x256) — 78.3 (2023-03-21) PoseBH-H — 79.5 (2025-05-23) PoseBH-H — 79.5 (2025-05-23) PoseBH-H — 79.5 (2025-05-23) CMU-Pose — 61.8 (2016-11-24) RMPE++ — 72.3 (2016-12-01) CPN+ [6, 9] — 73.0 (2017-11-20) Flow-based (ResNet-152) — 73.7 (2018-04-17) PoseFix — 74.7 (2018-12-10) MSPN — 76.1 (2019-01-01) HRNet-W48 + extra data — 77.0 (2019-02-25) HRNet-W48+DARK — 77.4 (2019-10-14) CCM+ — 78.9 (2020-02-03) 4xRSN-50 (ensemble) — 79.2 (2020-03-09) UDP-Pose-PSA(384x288) — 79.5 (2021-07-02) ViTPose (ViTAE-G, ensemble) — 81.1 (2022-04-26)
RankModel APAP50AP75APLAPMAR Extra Training Data PaperCodeYear
1 ViTPose (ViTAE-G, ensemble) 81.195.088.286.077.885.6 ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation huggingface/transformers · vitae-transformer/vitpose · vitae-transformer/qformer · +3 2022
2 ViTPose (ViTAE-G) 80.994.888.185.977.585.4 ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation huggingface/transformers · vitae-transformer/vitpose · vitae-transformer/qformer · +3 2022
3 UDP-Pose-PSA(384x288) 79.593.685.984.376.381.9 Polarized Self-Attention: Towards High-quality Pixel-wise Regression PaddlePaddle/PaddleSeg · xmu-xiaoma666/External-Attention-pytorch · sithu31296/semantic-segmentation · +3 2021
3 PoseBH-H 79.591.985.886.575.984.5 ✓ PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation uyoung-jeong/PoseBH 2025
5 4xRSN-50 (ensemble) 79.294.487.176.183.884.1 Learning Delicate Local Representations for Multi-Person Pose Estimation open-mmlab/mmpose · chenyilun95/tf-cpn · caiyuanhao1998/RSN · +1 2020
6 CCM+ 78.993.88684.57583.6 Towards High Performance Human Keypoint Detection chaimi2013/CCM 2020
6 UDP-Pose-PSA(256x192) 78.993.685.883.676.181.4 Polarized Self-Attention: Towards High-quality Pixel-wise Regression PaddlePaddle/PaddleSeg · xmu-xiaoma666/External-Attention-pytorch · sithu31296/semantic-segmentation · +3 2021
8 4xRSN-50 78.694.386.675.583.383.8 Learning Delicate Local Representations for Multi-Person Pose Estimation open-mmlab/mmpose · chenyilun95/tf-cpn · caiyuanhao1998/RSN · +1 2020
9 PCT (256x256) 78.3985.9––– Human Pose as Compositional Tokens gengzigang/pct 2023
10 HRNet-W48+DARK 77.492.684.683.773.682.3 Distribution-Aware Coordinate Representation for Human Pose Estimation PaddlePaddle/PaddleDetection · open-mmlab/mmpose · leoxiaobin/deep-high-resolution-net.pytorch · +3 2019
11 SwinV2-L 1K-MIM 77.2––––– Revealing the Dark Secrets of Masked Image Modeling SwinTransformer/MIM-Depth-Estimation 2022
12 HRNet-W48 + extra data 7792.784.583.173.482 Deep High-Resolution Representation Learning for Human Pose Estimation open-mmlab/mmdetection · PaddlePaddle/PaddleDetection · open-mmlab/mmpose · +36 2019
13 EvoPose2D-L 76.892.584.382.573.581.7 EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation using Accelerated Neuroevolution with Weight Transfer wmcnally/evopose2d 2020
14 SwinV2-B 1K-MIM 76.7––––– Revealing the Dark Secrets of Masked Image Modeling SwinTransformer/MIM-Depth-Estimation 2022
15 HRNet-W48+UDP 76.592.78473.082.481.6 The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimation open-mmlab/mmpose · mindspore-lab/mindone · HuangJunJie2017/UDP-Pose 2019
16 OmniPose (WASPv2) 76.492.683.782.672.681.2 OmniPose: A Multi-Scale Framework for Multi-Person Pose Estimation bmartacho/OmniPose 2021
17 HRFormer-B 76.292.783.882.372.581.2 HRFormer: High-Resolution Transformer for Dense Prediction HRNet/HRFormer 2021
18 MSPN 76.193.483.881.572.381.6 Rethinking on Multi-Stage Networks for Human Pose Estimation open-mmlab/mmpose · chenyilun95/tf-cpn · megvii-detection/MSPN · +4 2019
19 MIPNet 75.792.483.381.271.480.5 Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation rawalkhirodkar/MIPNet 2021
19 PPE (ResNeXt-101) 75.790.376.379.580.7– Deep Multi-Task Networks For Occluded Pedestrian Pose Estimation 2022
1–20 / 141 다음 → 페이지당 10 20 50 100