README.md

---
license: mit
---

# Face Recognition Model
"CSSE416-final-project/faceRecogModel" is the Hugging Face (HF) repository we deployed the final, fully trained facial recognition model to. Checkout the HF Space "CSSE416-final-project/faceRecogModel" to easily test and play around with the model.


## Files
- modelWeights101.bin
  - Weights of model we created by training on Gebru servers
- app.py, model.py, and config.json
  - Standard files used by Hugging Face to setup the architecture of the model in conjunction with the weights
- updateHFtemplate
  - Used this file to update the above files in the Hugging Face repository
  - The weights file did not update properly using standard Git, so we used Hugging Face Hub with this file to properly update the repo
    - Could also use Git LFS instead.
  - You will need a personal access token to use the file. Mine has been removed for security purposes.
- testModelWeights.ipynb
  - Converts the weights file into the model.
  - Test the model on a dataset

## Dataset
  - testData
    - Folder of images with names of the subfolders representing classes that can easily be used
    - Change the contents of this folder, or specify a new filepath, to run your custom dataset
    

## Neccessary Package
  - Since the model all runs on Hugging Face, you don't need to install any packages.
  - However, if you do want to run the model locally, you will need to install these following packages:
      - torch
      - numpy
      - torchvision
      - huggingface_hub

## Running
  - You can run this model using the Hugging Face Space: CSSE416-final-project/faceRecogModel.
    - You can upload images through the Space UI.
  - Alternatively, if you want to test the model locally, you can download the repo and run "testModelWeights.ipynb."
    - You can run the notebook on CoLab as well, if you change the dataset reference to be a filepath on your computer.