muhammadasgharaliqureshi/Depth-Sensing-Module-With-Face-AntiSpoofing-Feature

This prototype uses Depth as a feature in Neural network and then train it self to do discriminate between Depth face(i.e. Real Face) and Spoof Attack (i.e. Image, video). The concept used is by first creating a depth sensing stereo camera based module. Stereo Camera module is designed by using two low cost cameras and then calibrate them in stereo configuration. After calibration and rectification Depth can be precepted by calculating Disparity between the two R.G.B images obtained( By stereo cameras). The obtained depth was tested on real faces and fake faces(i.e. image & video of a person face) and Data set was collected. after collecting Data set I trained a model (which I named Face Anti spoofeer) and used that model with a pretrained face Recognition model to validate a recognized persona as a real person or a spoof attack. Further more this project also includes controlling a security Door wirelessly via Arduino with the support of GUI that is running on Server and of course Raspberry pi was used as a client because it itself is a stereo module.

Python