Abstract:
This project leverages computer vision for material classification, promoting
sustainable practices and incentivizing user participation. Using YOLOvl 1, a state-of-
the-art object detection model with 75% accuracy in identifying materials like metal
and plastic, it ensures reliability through a well-annotated Roboflow dataset. The
system features a Flutter-based mobile app and a MERN stack web app, with
MongoDB as the database backbone. Users can capture images via the mobile app,
which transmits them to the classification model for inference, displaying results like
material details and unit breakdowns. The web app provides analytics, including
classification percentages and user contributions. A unique point-based reward system
encourages user engagement, offering redeemable points for recycling activities. By
integrating machine learning, cross-platform development, and incentives, this
scalable solution addresses waste management challenges and promotes
environmental awareness in daily life.
