Combines knowledge of computer vision and deep learning (CNNs) to build dectection system for facial keypoints including points around the eyes, nose, and mouth. It has several applications including facial tracking, facial pose recognition, facial filters, and emotion recognition. The completed code is able to look at any image, detect faces, and predict the locations of facial keypoints on each face; examples of these keypoints are displayed below.
The project is broken up into a few main parts in four Python notebooks:
Notebook 1 : Loading and Visualizing the Facial Keypoint Data
Notebook 2 : Defining and Training a Convolutional Neural Network (CNN) to Predict Facial Keypoints
Notebook 3 : Facial Keypoint Detection Using Haar Cascades and the Trained CNN
Notebook 4 : Fun Filters and Keypoint Uses
Notebook Documentation and Starter Code are part of project files provided by Udacity in the Computer Vision Nanodegree.