Face Detection and Recognition
View in GitHubA face recognition system
How this works
- First, we use OpenCV to find faces.
- Crop out the faces.
- Use an algorithm to give encoding for each face in the form of 128 element array.
- If the person in two pictures is the same, their encodings will be similar otherwise the difference in encodings of the two pictures will be large.
This algorithm is inspired by FaceNet and DeepFace Try reading these papers, FaceNet is pretty simple. You know, FaceNet is developed by Google.
There is a readme and Google Colab notebook in the project that extensively describes the project pipeline.