equinox013/smart-waste-classifier

Smart Waste Classifier

A deep learning image classifier that identifies the waste category of a household item from a photo and provides correct disposal guidance.

What It Does

Upload any photo of a household waste item. The model classifies it into one of ten categories and tells you exactly how to dispose of it — recycling bin, compost, hazardous waste drop-off, or general landfill.

Classes

battery · biological · brown-glass · cardboard · green-glass · metal · paper · plastic · trash · white-glass

Model Details

• —Architecture: MobileNetV2 (pre-trained on ImageNet) + additional Dense layers (256 → 128), fine-tuned end-to-end
• —Transfer Learning: Base frozen during initial training; dense head trained for classification task
• —Input: 224×224 RGB images, normalised with MobileNetV2's preprocessing pipeline
• —Performance: ~92% test accuracy on held-out evaluation set
• —Framework: TensorFlow / Keras

Files Required in This Space

Portfolio

Part of the bdatasci_portfolio — SIT319 Deep Learning.