Multi-task deep learning pipeline for automated waste sorting — classifies incoming waste by material type, recyclability, liquid content, & component count using four CNN models. Achieved 92.4% accuracy on real-world external data. Built in collaboration with Nestlé & Dawar; deployed in a physical smart bin prototype. — AUC Senior Thesis 2026
data-science sustainability computer-vision deep-learning tensorflow cnn image-classification transfer-learning smart-bin waste-classification recyclability
-
Updated
May 30, 2026 - Jupyter Notebook