This app helps identify what fruit or vegetable it is. It has web interface and performs all tasks asynchronously.
Images:
web: flask web app
worker: celery
model: flask restful web service, trained model
redis: message broker
Container orchestration tool is used for managing containers.
Model: ResNet50, loss: CrossEntropyLoss, metric: Accuracy
Trained model is already included into model image. To generate a new model:
- run
model/train.ipynb - rebuild docker images.
Fruits and vegetables were planted in the shaft of a low-speed motor (3 rpm) and a short movie of 20 seconds was recorded.
A Logitech C920 camera was used for filming the fruits. This is one of the best webcams available.
Behind the fruits, we placed a white sheet of paper as a background.
https://www.kaggle.com/datasets/moltean/fruits
Run the app:
$ sudo docker-compose up --buildOpen browser to view web gui http://localhost:5000
Get prediction:
- upload new fruit/vegetable image (or use already uploaded)
- click 'Check'
And get fruit/vegetable name!