lecture: Real-time Face Detection and Emotion/Gender classification with Convolutional Neural Networks
In this work we present a real-time system for face detection and emotion/gender classification using Convolutional Neural Networks and Haar-like features.
State-of-the-art methods in object detection and image classification are all based in Convolutional Neural Networks (CNNs). These applications require architectures constituted of millions of weights such as: VGG16, ResNet-101 and GoogLeNet. Consequently, real-time systems become unfeasible when using the state-of-the-art architectures. In this work we present the design and implementation of a CNN model for real-time face detection and gender/emotion classification. We report accuracies of 96% in the IMDB gender dataset and 66% on the fer2013 emotion dataset. Moreover, we argue that the careful implementation of CNN models and the use of the current regularization techniques can help us reduce the gap between slow performances and real-time systems. The latest implementation is freely available in the following link: https://github.com/oarriaga/face_classification