Gender recognition is consider as one of the interesting visual task for an extremely social animal like us humans, many social interactions critically depend on the correct gender perceptions. In contrast, as the field of artifcial life emerged, researchers began applying principles such as gender recognition to achieve socially interactive robot behavior. Toward to implement a social competence in a humanoid robot, first task has to be detecting human faces, then recognize and identify more information about detected faces such as gender. Gender recognition can be regarded as classifcation problem of detected faces into two classes (male and female). In the literature there are three popular approaches to detection of features for gender classifcation: Principal Components Analysis (Eigenfaces), Linear Discriminant Analysis(Fisherfaces) and Local Binary Pattern Histogram (LBP), where previous work in machine learning focused on different types of classiffers such as Nearest-Neighbor Classifers(NN) versus Radial Basis Function networks (RBF), Adaboost-Based Classifers or Support Vector Machine (SVM).
This project addresses the problem of gender classifcation using frontal facial images as training images, where the goal of this project is to automatically detect faces on images or video, quickly and reliably classify the gender of the detected faces. Our gender classiffer has been trained using two different datasets; more details are in the evaluation section. I did this project for Machine Learning course. Please read the Nadine Report of Machine Learning Project.pdf for more information. Please watch video for the results: https://vimeo.com/90117910.
library used in the porject:
- Boost
- OpenCV Lib
- Qt
- and Threading Building Blocks Intel TBB.
