ComfyUI-ultimate-openpose-estimator

<div align="center">

ComfyUI ultimate openpose(dwpose) estimator

</div> <p align="center"> <img src="assets/estimator_example_1.png" /> </p>

This is an enhancement of the project ComfyUI Dwpose TensorRT by giving control and output options. Check original project for the super fast performance.

Note: This is for tensorrt only, which is only working if you have cuda based Nvidia card

This project following the original project's license which is CC BY-NC-SA, everyone is FREE to access, use, modify and redistribute with the same license.

If you use this project for commercial purposes, please contact at [email protected] (the original project) and cc to me [email protected]

If you like the project, please give me a star! ⭐

---

Features

• Giving the detecting threshold option <p align="center"> <img src="assets/estimator_example_2.png" /> </p>
• Giving output canvas resolution adjustment
  • It will keep original picture's ratio
  • Minimum requirement for resolution_x >= 64px, for value < 64, the output pose image will be 512px
• Giving the plot control options for body pose, face and hands
  • The default marker size value is optimized for 1024px pictures, for smaller size picture, lower the marker size value, otherwise increase the marker size value
• Giving output options of pose keypoints and json string formats for further using, check my ultimate-openpose-editor if you need editting the pose

Installation

• Navigate to the ComfyUI /custom_nodes directory

git clone https://github.com/westNeighbor/ComfyUI-ultimate-openpose-estimator
cd ./ComfyUI-ultimate-openpose-estimator
pip install -r requirements.txt # if you use portable version, see below

if you use portable version, install requirement accordingly, for example, I have portable in my E: disk

E:/ComfyUI_windows_portable/python_embeded/python.exe -m pip install -r requirements.txt

• Restart ComfyUI

Building Tensorrt Engine

1. Download the following onnx models:

   • dw-ll_ucoco_384.onnx
   • yolox_l.onnx

2. Build tensorrt engines for both of these models by running:

   • python export_trt.py

3. Place the exported engines (yolox_l.engine and dw-ll_ucoco_384.engine) inside ComfyUI /models/tensorrt/dwpose directory

Usage

• Insert node by Right Click -> ultimate-openpose -> Opnepose Estimator Node

Credits

• https://github.com/yuvraj108c/ComfyUI-Dwpose-Tensorrt
• https://github.com/IDEA-Research/DWPose
• https://github.com/legraphista/dwpose-video

License

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)