IDEA-Research/DWPose

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2807 stars · 176 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DWPose is the official implementation of 'Effective Whole-body Pose Estimation with Two-stages Distillation' (ICCV 2023), providing whole-body human pose estimation models from tiny to large sizes. It can replace Openpose as a ControlNet preprocessor in Stable Diffusion workflows and is built on MMPose.

Use cases

• estimate whole-body human pose from images
• use pose estimation as ControlNet preprocessor for Stable Diffusion image generation
• run pose estimation with ONNX models without installing mmcv
• animate characters with pose-guided image-to-video synthesis
• choose pose models of different sizes for speed vs accuracy tradeoffs

When to choose

• you need accurate whole-body (body, face, hands) pose estimation
• you want a better pose preprocessor than Openpose for ControlNet
• you need lightweight ONNX pose models for inference

When to avoid

• you need 3D pose estimation
• you need pose tracking across video frames rather than per-frame estimation
• you need a non-Python or non-GPU environment without ONNX support

Facets

library · maturity maintenance

computer-vision machine-learning image-processing stable-diffusion computer-vision image-processing machine-learning artificial-intelligence python cross-platform pose-estimation whole-body-pose knowledge-distillation controlnet onnx mmpose stable-diffusion-webui human-pose gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="IDEA-Research/DWPose")