What this repository contains:
- Report outlining our method and results
- Google colab notebook containing the neural network for body part segmentation and code for matching poses between images
- Modified matlab code from https://www.cs.stanford.edu/~roozbeh/pascal-parts/pascal-parts.html for getting all examples with peopple in them and creating segmentation masks with 7 classes for the desired limbs and background. The matlab code also creates an 80-20 split between training and test data.
What you need to run our project:
- Our weights can be downloaded from https://drive.google.com/open?id=1mqep4SfOxaEd9fujQ6fNxbQwqvZ9cuO9
- Our taining and test data can be downloaded from https://drive.google.com/open?id=1-wh4P5CXLz_YjTHWyTvfGVOYW1D_VBfi
Credit:
The code and methods were developed in colloboration with trazd66 (Tarry) and the report was written by myself (Shimi).