Human-Activity-Recognition

This repository contains code files and reports for the Human Activity Recognition Project. The project uses the MPII Dataset to train two deep learning networks: a regression network to estimate 2D human body poses followed by a classification network to predict the major activity in the image. Feature Extraction was done using the ResNet 50-v2 architecture. Project code uses standard Python 3.x libraries such as Pandas and the deep learning frameworks TensorFlow 2.0 and Keras.