Still a work in progress

• Based on easy_ViTPose
• Objective: easier to use and more flexible interface for ViTPose
• This repo is not for training, only for inference (for now, at least)
• CoreML model added respect to the original repo

TODOs

• Add support for video, webcam inference
• Turn this model into a package and add distribute it to PyPi
• YOLO like apis for inference
• Check if TensorRT backend is better without the use of pycuda

Installation

git clone https://github.com/lum1t4/vitpose_extra.git