Reaction Rumble : A Facial Expression Recognition Application using CNN

However, CNNs and SSD MobileNet serve a slightly different plane of purpose. Their main adaptability lies in the differentiation of features of face landmarks at various levels, such as eyes, nose, or mouth, particularly necessary for understanding facial structure and expressions. Due to such hierarchical feature extraction, CNNs can be used in very precise landmark detection. The SSD Mobile Net is a compact, efficient model for face detection, designed for mobile and real-time applications. SSD-a Single Shot Multibox Detector employs the single pass of feature extraction where faces are detected by predicting bounding boxes. The use of Mobile Net simplifies computationally without leaving behind the performance. The latent representation obtained using CNNs helps further to do real-time facial landmarking and expression detection on devices with little resource.