Light Weight Pose Network

This repo contains a quick and easy to understand re-implementation of the LightWeight Pose Network architecture from corresponding paper https://arxiv.org/pdf/1911.10346.pdf. This was a work I did for understanding the core architecture of the LPN model. Main reference for completing this model were from original corresponding implementation from the said paper by authors, https://github.com/Daniil-Osokin/lightweight-human-pose-estimation.pytorch.

Details of the uploads are:

1. img.jpeg = Sample test image
2. LPN.ipynb = Jupyter notebook having lpn50 implementation, LightWeight 2D pose network for extracting 2d poses from images
3. conf.yaml = lpn configurations
4. UntrainLPN_Results = Results of untrained LPN50

Update:

The repo contains just the LPN model architecture, and said untrained model's performance on pose estimation for a sample image. Model described here is nto trained and trained weights for the model are not provided either.