Design and Fabrication of Automated Waste Segregation System

This paper introduces a design and implementation of an automated system of real-time object recognition and

sorting, tailored to waste sorting. The system combines a conveyor belt, color sensor, Raspberry Pi camera, and

microcontroller (Arduino) to identify objects and sort them into proper categories. OpenCV computer vision algorithms

can convert to grayscale, threshold, extract contours, and estimate size using perimeters. It also includes Object

recognition (via ORB-based feature matching) and optical character recognition (OCR) with Tesseract to read labels or

printed text on waste containers. A Python/Tkinter front-end allows real-time observability of the sorting process.

Experimental analysis on a varied sample group exhibits robust performance, with perimeter-based analysis showing

about 97% accuracy, feature matching at 95 percent, and OCR attaining 92 percent. These findings show that the system

can effectively detect, categorize, and group waste materials with accuracy that is worthy of use in real-life situations. The

next generation can involve more frame rate image capture, better lighting optimization, and hardware acceleration to

increase throughput and scalability. On the whole, this research shows that a low-cost computer- vision-based architecture

can be successfully used to automate waste segregation processes, providing a viable alternative to manual sorting

operations.