The contemporary world grapples with a critical issue—the effective management of waste. The surge in population and industrial activities has caused a substantial rise in waste generation, contributing to environmental degradation, resource depletion, and various sustainability challenges. In addressing this dilemma, the practice of garbage classification has emerged as a crucial solution. It plays a significant role in mitigating the adverse impacts of waste on the environment and fostering a more sustainable approach to waste management.

Our project addresses the critical issue of garbage classification by leveraging the YOLOv7 real-time object detection framework. The first step involves assembling a comprehensive dataset of garbage items and categorizing them into several distinct groups. To ensure precise categorization of the images, we adapt YOLOv7, a powerful tool for real-time object detection. This project encompasses various stages, including data collection, preparation, and labeling, with a particular emphasis on employing the most effective methods for data labeling—an essential step in the project.

Additionally, the process involves data preprocessing, model training, evaluation, and real-time inference. Via these comprehensive steps, our project aims to contribute to the advancement of garbage classification methodologies, ultimately promoting a more sustainable and efficient approach to waste management.

Furthermore, it is worth noting that some of the achieved values closely align with the performance of YOLOv4, a more advanced iteration of YOLOv3.