Image preprocessing without pose estimation and skeleton rendering (Pop!_OS 22.04 LTS)

[Andrei630]

Issue Summary

I have converted the openpose body_25 model to tensorflow's SavedModel format as I want to develop a tensorflow module for openpose. I thus would need some advice on how OpenPose preprocesses images, and if it is possible to use the python/c++ APIs just for image preprocessing (saving the processed image without actually performing pose estimation / rendering skeletons).

From what I understand the "emplaceAndPop" function of the Wrapper class does all the preprocessing, pose estimation, and skeleton rendering all at once. If that's not the case could anyone explain how it actually works? Is there any similar function that can be used just for the preprocessing or skeleton rendering? It would be very helpful if anyone could point out the exact functions and files that are used for the preprocessing of the image.

Type of Issue

• Question

Your System Configuration

OpenPose version: 1.7.0

General configuration:
- Installation mode: Cmake on the terminal
- Operating system (lsb_release -a in Ubuntu): Pop!_OS codename jammy
- Operating system version: 22.04 LTS x86_64 (Based on Ubuntu)
- Release or Debug mode? (by default: release): release
- Compiler (gcc --version in Ubuntu or VS version in Windows): GCC 11.4.0-1ubuntu1 22.04, G++ Ubuntu 11.4.0-1ubuntu1 22.04 11.4.0

3rd-party software:
- Caffe version: Default from OpenPose
- CMake version (cmake --version in Ubuntu): 3.19.6
- OpenCV version: 4.11.0 built from source