README.md

This is the solution and implementation for project 2 face detection in COMP 5212.

I have finished the following basic tasks:

1. Use the training images to create positive and and negative training HoG features.
2. Train a linear SVM classifier.
3. Create a multi-scale, sliding window object detector.
4. Write up with design decisions and evaluation. (html/index.html)
5. Implement hard negative mining and evaluate the corresponding improvement. (Extra credit)

Usage:

1. If you want to test the extra test scenes, please change "EXTRA_TEST_FLAG = 0" to be "EXTRA_TEST_FLAG = 1" in line 45 in proj4.m