This repository contains the code for a perception stack designed and developed for a human tracking robot. The perception stack utilizes deep learning-based systems for person detection and re-identification.

The following dependencies are required to run the code:

• PyTorch
• OpenCV
• NumPy

Create a virtual environment in the root directory:

pip install virtualenv

virtualenv <env name>

To activate the environment:

source <env name>/bin/activate

Clone the repository:

git clone https://github.com/Ashwij3/Human_following_robot.git

To install the dependencies, run the following command:

pip install -r requirements.txt

To use the perception stack, run the following command:

Copy code

python3 scripts/main.py

The perception stack utilizes the following algorithms:

• YOLOv5 for human detection and instance segmentation
• SuperPoint for feature extraction and matching

The implemented perception stack achieves accurate identification of the target individual, allowing for effective human tracking by the robot.

The code in this repository is based on the following research papers:

• YOLOv5: A Universal Object Detector
• SuperPoint: Self-Supervised Interest Point Detection and Description