FishFinder is a computer vision project based on images from Monterey Bay Aquarium's live-streaming camera.
To run the model via Docker-Compose, run the following commands:
cd model_serving
docker-compose build
docker-compose up
To interact with the model-serving API, send HTTP requests according to the API Reference.
The model used for object detection is YoloV5, by Ultralytics. Feel free to check out their GitHub repository to learn more about this model.
The model has been trained on a custom dataset, composed of images from the Aquarium, which have been manually annotated.