A face recognizer for Princeton EE department trained with a single image per person. Greets recognized faces with a short message on top of the normal display slideshow. Currently in developmental stage. It runs two Python backend scripts: one for deploying the website on a local server through Django; and the second for capturing from a webcam and marking down which faces it detects. The webcam re-trains itself from a Dropbox sync-ed folder at midnight everyday. Please note: this currently uses a very hacky approach to asynchronous updating, by having the website parse a JSON updated by the webcam script. This is a very messy approach and will be updated in the future.
Instructions for deploying (full installation instructions)
- Clone this repo
- Install all necessary requirements (see requirements.txt, but cannot be installed all from pip, and also needs openCV)
- Clone the latest openface repo
- Start the face recognition backend and then Django server with:
./main.shand
cd eeslides
python manage.py runserver