PlasticDetectionModel is a machine learning model used for detecting marine debris, specifically plastic, from Sentinel-2 L2A satellite images. The model is customized from the marinedebrisdetector to work in memory on a serverless cloud infrastructure. This repository contains the code and resources necessary for deploying the model on a Runpod serverless GPU. This repository is part of a larger pipeline aimed at identifying marine debris on a schedule: PlasticDetectionService. Ultimately, the predictions are displayed on our mapping application, deployed here: https://oceanecowatch.org/en
Add a .env file with the following keys
RUNPOD_API_KEY=<your_runpod_api_key>
https://hub.docker.com/repository/docker/oceanecowatch/plasticdetectionmodel/general
docker pull oceanecowatch/plasticdetectionmodel:latest
docker run -d -p 8080:8080 --name plastic_detection_model oceanecowatch/plasticdetectionmodel:latest
To deploy this Dockerized version on a Runpod serverless instance, follow these steps:
- Log in to your Runpod account and create a new serverless instance.
- In the deployment settings, specify the Docker image to use: oceanecowatch/marinext:latest.
- Deploy the instance and monitor the logs to ensure everything is running smoothly. For detailed instructions, refer to the Runpod documentation.
