Arnold.
Back to projects
Data & AI

Risaikuru

Deep Learning Trash Classification Web App

PythonTensorFlowFlaskDeep LearningComputer VisionHTML/CSS
Risaikuru

About this project

Risaikuru (from Japanese リサイクル — recycle) is a web-based application that uses deep learning to classify trash into recyclable categories. Users upload or capture an image of their waste, and the model instantly predicts the correct disposal category.

The backend is powered by a custom-trained convolutional neural network served via a Flask API, while the frontend provides a clean, responsive interface. The model was trained and evaluated on a curated dataset with strong accuracy across organic, plastic, metal, glass, and paper categories.

Key highlights

  • Custom CNN trained on multi-class waste image dataset
  • Flask REST API serving real-time image classification
  • Responsive web frontend with drag-and-drop image upload
  • Model achieves high accuracy across 5 waste categories

Gallery

Risaikuru screenshot 2
Risaikuru screenshot 3