This code implements a Gender Classification algorithm using LBP features in C++ using OpenCV 2.4.9. The code was originally developed as a homework for a Computer Vision course in 2014 so it possesses no optimizations whatsoever and focused solely on working as intended.

• The code is intended for use with the images inside the data folder.
• The LBP algorithm used is LBP(8,1,u2) and is described in the references.

After compiling, the program has two ways of executing:

./gender_classifier

./gender_classifier (int from 1 to 4)

./gender_classifier (int from 1 to 4) (0 or 1)

• The first one uses a default kernel (LINEAR) and no-hair faces for the SVM.
• The second one allows the use of other kernels (1 for LINEAR, 2 for POLYNOMIAL, 3 for RADIAL BASIS FUNCTION and 4 for SIGMOIDAL). Parameters for each kernel can be modified in src/main.cpp for tuning up purposes.
• The third one lets you choose of the two face folders which to use (0 for images with no-hair, 1 for images with hair)

• Ahonen, T.; Hadid, A.; Pietikainen, M., "Face Description with Local Binary Patterns: Application to Face Recognition," Pattern Analysis and Machine Intelligence, IEEE Transactions on , vol.28, no.12, pp.2037,2041, Dec. 2006

Cristobal Silva

crsilva at ing dot uchile dot cl