A hackathon project created for McHacks 2022. The application is a scanless checkout system who's use case is to provide users a faster and seamless way to scan items from their cart.
- Youtube Video: 2-minute video summarizing the project and demoing the use case
- DevPost submission
- Create a conda environment along with the necessary python dependencies:
conda create --name mchacks -f environment.yml - Activate the conda environment:
conda activate mchacks - Install the pip dependencies:
pip install -r requirements.txt
NOTE: Make sure you're using the appropriate conda environment. Don't know what that is? Then follow the installation guide first.
- Run
python main.py.
This will load the machine learning model, connect to the camera, and start the web server. This is the application's intended use case: as a scanless checkout system.
- Run
python camera.py
This is mainly for testing the webcam, image difference detector and object class inference. It's a helpful tool for debugging. This will open a window that streams video feed from the webcam.
Press the ESC key the exit the program.
Press the spacebar key to compute an image difference and cropping of the largest connected-component from the image difference mask.
- Run
python model.py
You need to modify code in model.py to make any changes to training variables such as the number of epochs, criterion function, etc.