Face Liveness Detection (CRMNet)

Description

CRMNet: A deep-learning pipeline capable of spotting fake vs legitimate faces and performing anti-face spoofing in face recognition systems. It is built with the help of Keras, Tensorflow, and OpenCV. A sample dataset is uploaded in the sample_dataset_folder.

Method

The problem of detecting fake faces vs real/legitimate faces is treated as a binary classification task. Basically, given an input image, we’ll train a Convolutional Neural Network capable of distinguishing real faces from fake/spoofed faces. There are 4 main steps involved in the task:

1. Build the image dataset itself.
2. Implement a CNN capable of performing liveness detector(Livenessnet).
3. Train the liveness detector network.
4. Create a Python + OpenCV script capable of taking our trained liveness detector model and apply it to real-time video.
5. Create a webplatform to access the liveness detection algorithm in an interactive manner.

Contents of this repository

1. sample_liveness_data : contains the sample dataset.
2. Face Liveness Detection -Saketh.pptx : A couple of slides that will give you information on th project and our motivation.
3. demo.py : Our demonstration script will fire up your webcam to grab frames to conduct face liveness detection in real-time.
4. deploy.prototxt : Support file for pretrained face detector.
5. le.pickle : Our class label encoder.
6. liveness.model : The liveness model file.
7. livenessnet.py : The python file containing the model.
8. res10_300x300_ssd_iter_140000.caffemodel: Pretrained face detector.
9. train_liveness.py: The python script to train the model.

Working flow

Further work

1. Gathering data having a larger set of ethnicity and different types of fake/spoofed photos.
2. Adding more heuristics to team up with deep-learning.

Disclaimer

This work was done during my internship at SimpleCRM, Nagpur.