Skip to content

README.md

Latest commit

 

History

History
93 lines (71 loc) · 2.95 KB

README.md

File metadata and controls

93 lines (71 loc) · 2.95 KB
Raw

Computer Vision

1. Gaze Detection and Tracking :-

Overview

This project implements gaze detection and tracking using computer vision techniques. It processes real-time video input to identify eye contours, apply masking, reshape NumPy arrays, and run inference.

Included Files

  • Contours.py: Extracts contours of the eyes for tracking
  • EyeMasking.py: Applies masking techniques to isolate relevant eye regions
  • NumPyReshaper.py: Handles reshaping of NumPy arrays for processing
  • run.py: Main script to execute gaze detection and tracking

Requirements

Libraries

Ensure the following dependencies are installed before running the project:

  • dlib
  • cv2 (OpenCV)
  • numpy

You can install the required libraries using:

pip install opencv-python numpy dlib

Setup and Usage

  1. Ensure you have Python installed (recommended version 3.7 or later).
  2. Install the necessary libraries as listed above.
  3. Run run.py to start gaze detection and tracking.
python run.py
  1. Follow on-screen instructions or refer to output for further details.

Precautions for Using dlib

  • dlib requires CMake and Boost installed for compilation if building from source.
  • Windows users may need to install Visual Studio Build Tools to resolve dependencies.
  • Linux users should ensure build-essential and cmake are installed.
  • If encountering installation issues, consider using pre-compiled binaries where available.

2. Generative Adversarial Network (GAN) Project

This repository contains two implementations of a Generative Adversarial Network (GAN) for generating synthetic MNIST digits.

Overview

This project implements a GAN to generate handwritten digit images using the MNIST dataset.

  • The PyTorch implementation is fully functional and well-structured for ease of use.
  • The TensorFlow implementation is currently incomplete and contains known bugs.

Features

✅ PyTorch implementation works flawlessly
✅ Generates high-quality synthetic MNIST digits
✅ Modular code for better maintainability
✅ Saves model checkpoints every 5 epochs
❌ TensorFlow version is not working due to bugs

Installation

Before running the project, install the required dependencies:

pip install torch torchvision numpy matplotlib

For the TensorFlow implementation (if fixed in the future):

pip install tensorflow numpy matplotlib

Usage

Training the GAN (PyTorch) To train the model:

cd GAN_Pytorch
python train.py

Checkpoints are saved every 5 epochs in the checkpoints/ directory.

Generating Images

Once trained, generate images using:

python generate.py

Generated images will be stored in the output/ directory.

Notes

Use the PyTorch implementation as it is stable and fully functional. The TensorFlow version currently contains issues and may not work correctly. If modifying or improving the TensorFlow version, ensure proper debugging before use.