Note: This is a modification of the code written by Akshay Chandra Lagandula: https://github.com/akshaychandra21/Selfie_Filters_OpenCV. His repository contains additional files which create a neural network for facial keypoint detection and train it. This repository simply loads the resulting model and uses it for keypoint detection.
Chad Bloxham. Done as the final project for a graduate course in neural networks. Group partners: Mehul Kothari, Srishti Tomar, Akshit Goyal.
Uses voice activation to change between facial filters. Includes several hat and mustache filters in addition to sunglasses.
Requires: an internet connection, Python 3.6.x (won't work for 3.7.x), Keras, Tensorflow, opencv-python, numpy, speech_recognition, threading.
Run filters.py. Filter 1 will be displayed. To change to another filter, say into microphone: "1", "2", "3", "4", or "5". There is a short lag between saying the filter and it being displayed. Any other audio input will cause the program to terminate, as will being too close to the webcam.
