Abstract: Face detection has grown in popularity as a challenge in image processing and computer vision. Convolutional architectures are being used in many new algorithms to make them as accurate as feasible. These convolutional designs have allowed even pixel information to be extracted. We want to create a binary face classifier that can detect any face in the frame, regardless of alignment. We show how to make accurate face segmentation masks from any input image of any size. For feature extraction, the approach starts with an RGB image of any size and employs Predefined Training Weights of VGG - 16 Architecture.
Keywords: HAAR Cascade, Facial recognition, Deep Learning, CNN.
Copyright Form