CNN waste classification

Personal project — cnn-waste

A convolutional neural network project for classifying waste images. Training and experiments live in a Jupyter notebook; the best weights, class labels, and a small Streamlit app let you run inference in the browser on Streamlit Community Cloud.

Overview

The repository includes a trained model checkpoint (best_waste_model.pth), class_names.json, loader utilities, and streamlit_app.py for uploading an image and viewing predictions. A companion notebook (cnn_experiments.ipynb) documents the modeling workflow.

Tech & Tools

• Python, PyTorch
• CNN image classification
• Streamlit (requirements-streamlit.txt for cloud deploy)
• Jupyter for experiments

Links

• Live app (Streamlit Community Cloud)
• GitHub repository

See the repo’s STREAMLIT.md and verify_streamlit_ready.py for deployment checks. Treat predictions as demonstrations; real waste-sorting systems need validation for your environment and regulations.