The increased production of waste highlights an environmental and operational quest with a need to develop efficient and automated systems for waste management. Deep learning and computer vision is used in this study to detect types of garbage, to support smarter and more sustainable disposal practices. The study is based on the publicly available Keremberke Garbage Object Detection dataset, which includes 10,464 labelled images across six common waste categories: plastic, cardboard, glass, metal, paper, and biodegradable materials. This study’s objective is to quantify how effectively current object detection models can recognize and categorize types of waste from an image. Using the Keremberke dataset, the study focuses on evaluating the performance of YOLO-based models in processing real-world garbage images. The study also supports broader goals like advancing automation in waste sorting, bettering recycling rates and monitoring the environmental degradation. The results offer an equitable view of how these models perform under businesslike conditions, revealing both their capabilities and their constraints. Conclusively, this work connects academic model development with hands-on implementation, contributing to the growing role of AI in continual waste management.