AI Waste Bin using Edge IoT and Deep Learning for Restricting Animal Access and Allowing Humans

Nowadays, waste management is crucial for protecting nature, public health and the environment by reducing pollution, conserving resources, and promoting sustainable practices. This study introduces a combination of an IoT-based system and Deep Learning to control access to a waste container via the recognition of faces. An ESP32-CAM module captures real-time images and runs a machine learning model on the device itself. If a human face is detected, an actuator activates to open the lid briefly, whereas if there is any animal approach towards the bin, then the system will remain shut. The system processes data locally using edge computing, which ensures: 1) Fast response, 2) Low power consumption, and 3) No reliance on cloud services, as this system enables running deep learning directly on edge devices. The proposed framework achieved a good amount of accuracy over the other computer vision models, as well as over the other physical mechanisms.