Overview

This repository contains the Video2RF project, an innovative venture combining Signal Processing and Deep Learning. The project focuses on converting video data into Simulated RADAR signals and enhancing these signals using a Generative Adversarial Network (GAN) for CNN-based Human Activity Recognition (HAR).

Features

• GAN Implementation: For refining simulated RADAR signals.
• HAR System: Utilizing CNN for activity recognition with enhanced RADAR signals.

Results

• Achieved a 16% enhancement in signal fidelity using GAN.
• Attained preliminary 20% accuracy in HAR system using GAN-improved signals.

Project Pipeline

Areas of Improvement and Approach