<- back to projects
Garbage Classification Website
Our project during the university course, we have deployed a webapp to classify garbage images. We have used ResNet50 model to classify the images. The model has been trained on the dataset of 6 classes: cardboard, glass, metal, paper, plastic, trash. The model has achieved 90% accuracy on the test set. You can see the proof at our report. But because of the cost, we have to stop the service.
Overview
An end-to-end MLOps-focused web app that classifies waste images to support better sorting and recycling.
Highlights
- TensorFlow ResNet50 model for image classification
- FastAPI backend for model inference APIs
- React frontend for user interaction
- Dockerized deployment and GitHub Actions CI/CD
- Hosted and integrated on AWS services
Why This Project Matters
This project combines data science and production engineering, from model serving to web delivery.