CSE 802 Pattern Recognition Final Project: Unconstrained Face Recognition Challenge

This repo are the source code and documents for the CSE802 Pattern Recognition final project.

Basics

• Use convolutional neural network (CNN) to extract the face features. Validate the extracted features with BLUFR protocol. Compare the results with:
  • PCA (Principal Component Analysis)
  • LDA (Linear Discreminant Analysis)
  • LBP (Local Binary Patterns)
  • LE (Learning-baed Descriptors)
• Training database: CASIA WEBFace Database, 494,414 pictures of 10,575 subjects.
• Test database: LFW (Labeled Face in the Wild) database, 13,233 pictures of 5,749 subjects.

Results

Check the final report and the slides in the "docs" folder.

Detailed Requirement

Check the file "project.pdf" for detail.

Where Are The Data?

When we worked on this project, the image data are stored on HPCC (High Performance Computation Center) server. If you are interested in this project, check the reference [2] and [3] in the "project.pdf"

Others

• BLUFR