Facial Feature Extraction (Viola-Jones Algorithm)

This project implements a basic version of the Viola-Jones object detection algorithm for facial feature extraction. It reads a PGM (Portable Graymap) image, processes it, and outputs a modified PGM image.

Prerequisites

Before you can build and run the code, you need the following:

1. A C++ Compiler (GNU C++ Compiler): You'll need a C++ compiler and build tools.

2. make Utility: Verify that it's installed and in your PATH by opening a command prompt or PowerShell and typing make --version. You should see version information.

3. IrfanView (or another PGM viewer): To view the PGM images, it's recommended to install IrfanView. Download it from https://www.irfanview.com/. This program can also be used to convert between JPEG/PNG and PGM formats.

4. Python Virtual Environment with certain modules The modules required are in requirements.txt

Building the Code

Build with make: Use the provided Makefile to build the project:

$ make

This will create an executable file named facial_extract_cpu and facial_extract_gpu. Since the Makefile is automated, the testing of our dataset across both executables are logged in logs/cpu and logs/gpu. The detected (drawn rectangles on candidates) images are located in Detected_Images/. The data analysis script generates plots in logs/plots

Reference

• Viola-Jones-cpp

License

Please refer to the MIT license attached to this repository.