Trash Can! An AI system for automatic classification of waste
Abhay Chopde, Niranjan Bharate, Sajal Bhattar, Ashish Kunvar, Shreya Bhadwal · 2022 International Conference on Computational Intelligence and Sustainable Engineering Solutions (CISES) · 2022
Countries around the world struggle with the problem of waste recycling and management. This problem, if not addressed appropriately, poses serious consequences to the health and development of people globally. In this work an approach for a computer vision-based system which detects and classifies waste into various recyclable and non-recyclable categories is presented. It detects the objects present in an image frame and classifies them in their correct category- plastic, paper, metal, glass or trash (non-recyclable) with an accuracy of 67.4%. The system is designed as a highly portable system which can be used in almost all domestic or office settings. The detection is done through images captured by a Raspberry Pi 4 camera module with the help of a Raspberry Pi. The architectural model used for this task is the EfficientDet-Lite0 object detection model. This system can greatly help in increasing the amount of waste recycled and reducing the contamination of recycling appropriate materials.