/ECE387_Facial_Recognition

This project uses web scraping and modified version of Ageitgeys' "face_recognition" software to identify Jane and John Does.

Primary LanguagePython

ECE387_Facial_Recognition

Version 1.2.1

This repository includes detailed information about a facial recognition system using the PYNQ-Z1 board. For this project, two indipendent projects are working in conjuction:

  1. Web Scrapping
    We are obtaining the name and profile picture of each member of a private Facebook group. Those profile pictures are then saved, the name being the name on the Facebook page, into a file to be read later.

  2. Facial Recognition
    The images saved by the web scraper are read into the facial recognition system. The system then compares the saved image to the live video feed from the webcam. The system will then output the live feed to a computer monitor. The displayed feed will show a box around earch persons face, if the person is recognized by the system that persons name will appear under the box. If the person is not recognized, the word "unknown" will appear under the box. "Unknown" simply means that the person is not a member of the Facebook group.

Getting Started

  1. PYNQ-Z1 Board

    • See this guide for details on connecting the PYNQ board to your computer or router.

  2. Web Scrapping

    • We used beauiful soup for our web srapping needs.
      • You will need to download beautiful soup to your PYNQ board.
    • This guide has a video tutorial and static examples on how to log into and scrape Facebook using beautiful soup.

  3. Saving the Scrapped Information

    • We need to be able to acccess this data later so we want to save it to a specific folder. This page will help you to do just that.

  4. Facial Recognition

    • In order to use the facial recongination libary, you must first download the libary onto the PYNQ board. The board must have access to the internet to complete this download. See the PYNQ guide above for details on connecting your board to the internet.
    • !!! THIS DOWNLOAD WILL TAKE 5+ HOURS !!! Consider letting it download over night.
    • These instructions will help you to download the libary.
    • We used the glob API to access and read the saved files inside the folder created by the web scraper.
    • How to use video via the base overlay can be found here.
    • The facial reconition specific topics, such as comparing faces and creating a box around a persons' face can be found on Ageitgeys' Github.

  5. Crontab

    • This document describes Crontab and how Crontab is impliminted to make a program/script run on a set schedule.

Contributors

Austin Dulaney, Jake Iarve, Kyle Nichols and Nicky Nocerino.