Computer Vision: Facial-Identification program

Hoang Le Duc

ML Engineer
Software Engineer
Python
PyTorch

A Facial Identification Jupyter notebook using Facenet framework by Google Researchers on Pytorch

Program can run on both photo files and a live webcam
Program is tested on photo database from VKIST in Hanoi,Vietnam that are collected during their events
Photos are taken at different angles to improve the accuracy overall
Photo database can be found here Database
Photo files are seperated to 4 files (poliface_orign,poliface_test, poliface_run and poliface_run_max) with different functions
poliface_origin is used to first identified the face
poliface_test is the files used to test the AI on identifying face with more unknown subject
poliface_run is the result photo files seperated with the name of the subject as the folder name
poliface_run_max is the result photos including with frame over detected photos and name that program is identifying them with
To test run, download the photos database from above link into the root directory
Run the notebook inside IDE and result should be as above

To use live webcam feed to test

Run the last cell to test, press ESC to close the webcam window
Press the spacebar to save photo
Partner With Hoang
View Services

More Projects by Hoang