| --- |
| 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. |
| |
