Marine Debris Classifier

This repository contains a React web app designed to classify and segment marine debris using Roboflow's inference engine. The project highlights a modular, scalable approach to AI-powered object detection and citizen science, targeting the critical issue of ocean pollution.
Access the website here: https://new-one-eight-pearl.vercel.app

---

Technical Highlights

Frontend

• Framework: Built with React for a modern and responsive UI.
• Deployment: Hosted on Vercel for fast and reliable access.
• Features:
  • File upload interface for debris image classification.
  • Real-time interaction with the backend for inference.

Backend

• Framework: Developed with Flask, providing lightweight and efficient API endpoints.
• Deployment: Hosted on Heroku to ensure scalability and accessibility.
• Integration:
  • Communicates seamlessly with the frontend and the inference engine.

Inference Engine

• Powered By: Roboflow's API for object detection and segmentation.
• Dockerized: Runs in a Docker container for modularity and platform independence.
• Supported Classes:
  1. Fishing net
  2. Aluminum can
  3. Bottle
  4. Plastic bag
  5. Plastic waste
  6. Tire

---

Features

Dockerized Design

• The inference engine operates in a Docker container, ensuring platform independence and simplifying deployment.

Extensibility

• Easily swap or update models to incorporate new debris classes or refine current detection capabilities.

Citizen Science Integration

• Empowers users to contribute by uploading photos for debris classification, fostering community involvement in ocean health.

---

Deployment

Frontend

• Hosted on Vercel.

Backend

• Flask API hosted on Heroku.

Inference Engine

• Docker container running Roboflow's API for classification and segmentation.

---

Usage

1. Clone the repository:

   git clone https://github.com/yourusername/marine-debris-classifier.git
   cd marine-debris-classifier