This repository tries to mimic the searching of faces like in Apple photos and Google photos. A supervised learning strategy is used as opposed to the semi-supervised learning strategy used by Apple and Google.
Make sure to setup an environment before doing anything with this repository for the sake of maintaining dependencies and not installing or uninstalling unwanted modules.
- Create the environment:
virtualenv face-detect
- Activate the environment:
source face-detect/bin/activate
- Install dependencies:
pip3 install -r requirements.txt
- Deactivate environment:
deactivate
All the faces in an image are detected (using yolov3-tiny), cropped and saved. All the copped face images are first classified by the classifier. Depending on whether the accuracy of the topmost prediction is greater than a certain threshold, the face image is made to be tagged by the user.
Details of the database can be found here.
Details of the classifier can be found here.