The model was training on CIFAR10 and thus only supports the following classes:
["airplane", "car", "bird", "cat", "deer", "dog", "frog", "horse", "ship", "truck"]
You can download model weights, download the code, run docker-compose to build and run the backend and frontend with the following steps:
Place the model weights in folder image_classification_ml_app/state_dicts
.
Model weights to Google drive for weights of simple VGG model trained on CIFAR10 dataset.
git clone git@github.com:jonasdieker/image_classification_ml_app.git
cd image_classification_ml_app
docker-compose up -d --build
Then go to http://localhost:8501/ to interact with the web app and classify some images!