Computer vision system for waste recognition

The goal of the project "Computer Vision System for Garbage Recognition (Garbage Eye)" was to create a computer vision system for automatic recognition of garbage in photos, videos, and in real-time video streams. The system is designed for integration into "smart city" infrastructure and can operate both locally and in the cloud.

As part of the implementation:
- Analyzed existing analogs and selected effective neural network models;
- Prepared a dataset and trained the model;
- Created a graphical interface based on PyQt5;
- Implemented functionality for recognizing garbage in images, videos, and in real-time;
- Configured automatic email notifications to the official email of the communal services of Kremenchuk upon detection of a critical amount of garbage.

Technologies used:
Python, PyTorch, OpenCV, YOLO, PyQt5, Roboflow.

Result:
The Garbage Eye system successfully performs household waste recognition, can be integrated into city monitoring and response services. Automatic notifications to communal services facilitate quick response to pollution incidents in the urban environment.