WasteWatch: Intelligent Waste Level Detection

This paper presents a proof-of-concept system for waste level detection using an Oak-D Pro disparity camera integrated with a Raspberry Pi. The device captures and processes depth frames in real time to estimate the fill level of a paper waste bin in an industrial environment. A mean depth value is calculated from 500 consecutive frames captured every minute, serving as a proxy for bin fullness. Visual data collected alongside depth readings enables validation of the approach and identification of potential errors. Results demonstrate that the system can reliably distinguish between empty, full, and partially filled states, even under challenging conditions such as low lighting and variable ambient interference. This work highlights the potential for low-cost, adaptable vision-based solutions for automated waste monitoring.