Single Person 2d Joint Position Estimation Project:

Human pose estimation is the task of localizing key joint locations given a picture containing a person. This pose estimation model utilizes a basic Visual Transformer backbone and decoder for single person human pose estimation tasks implemented in PyTorch. The model was trained on the LSP, LSP extended and MPII human pose datasets. This model is not pretrained on any masking tasks as we were trying to test the feasability of performing this task without a pre-trained backbone.