This is the ML implementation of the features preented in our official Repo of Hackfest
- The repo contains two models:
- People Counter
- Face Recognition (API)
People Counter :
This models uses a pretrained mobilenet to classify objects as humans and non humans and gets the number of people entering the door or leaving the door.
The model is fed by a live camrea video stream (either from a local camera or a camrea publishing to a specific link (IOT) )
The live stream is breaken up into frames and the frames are used to clssify the objects as humans o and hence help to count the number of people inside the building.
Tech Stack : OpenCV , Cafee , PIL , Python , Esp32 library
Face Recognition :
This modeule is a fully functing API deployed on heroku.
This module uses cascade detector of Open-CV to generate encodings of the face.
The input image is converted to its encodings and are comparerd against the known encodings once a sucessful match is obtained the details of the student are returned to the calling application.
API Documentation:
Methods :
Type | Function | Usage
| |
Get | reset | To reset the whole database and generate new encodings
Post |update | To add encoding of new user to the database
Post |predict | To recognize a face
Api Link : https://facereco23.herokuapp.com/
Teck Stack : Heroku , Open-Cv , Python , PIL , Flask