EcoShreds: A 5-Class Image Dataset for Edge-AI Smart Waste Segregation and MobileNetV3 Optimization

This dataset contains 2,057 high-resolution images categorized into five distinct classes tailored for source-level municipal solid waste (MSW) segregation using low-power Edge-AI systems. The dataset was curated to train, validate, and deploy lightweight deep learning models, specifically optimized for the MobileNetV3 Small architecture, running on resource-constrained hardware such as the ESP32-CAM module. The primary objective of this dataset is to facilitate automated real-time waste classification at the point of disposal within smart bin infrastructure.