Trash-Scan: Smart Waste Segregation

Abstract

In today’s world, the amount of waste generated is increasing rapidly due to urbanization, industrial growth, and daily human activities. One of the major problems in waste management is the improper segregation of dry and wet waste at the source. When these types of waste are mixed, it becomes difficult to recycle or treat them properly, leading to environmental pollution and health risks. To address this issue, we propose Trash-Scan, a smart and affordable waste segregation system that automatically classifies waste as dry or wet using a machine learning model and sorts it using Arduino-based hardware. The system uses a pre-trained image classification model developed in Python, which can accurately predict the category of waste from an input image. This prediction is then communicated to an Arduino Uno through serial communication. The Arduino controls a servo motor that rotates a flap to direct the waste into the appropriate bin—either dry or wet. A 16x2 LCD display is also included to show the classification result in real-time to the user. The entire system is cost-effective, portable, and easy to use. Unlike other complex or expensive models that require high-end processors like Raspberry Pi or internet connectivity, Trash-Scan works offline and is suitable for schools, homes, and public spaces. The project has been tested under real-world conditions with high accuracy and fast response. By making waste segregation simple and automated, Trash-Scan helps promote better environmental practices and supports cleaner and smarter waste management solutions for the future.

Introduction

Trash-Scan is a low-cost, automated waste segregation system designed to address the widespread problem of improper waste sorting, which leads to recyclable and biodegradable materials being mixed and harder to process. Traditional methods rely on manual sorting or expensive industrial systems, while existing smart bins often use costly sensors or complex image-processing units. Trash-Scan combines an affordable Arduino-based hardware setup with a machine learning model that classifies waste as dry or wet using a simple Python interface.

Research shows that many countries struggle with waste segregation due to low public awareness, limited guidance, and small or poor-quality datasets. While CNN-based waste-classification models, sensor-based systems, and IoT smart bins have shown promise, most are costly, lack user-friendliness, or fail to provide educational support. Trash-Scan builds on this knowledge by offering an accessible system that merges image classification with mechanical sorting.

The system architecture includes two parts:

• Software: A Python ML model classifies waste images as dry or wet.

• Hardware: An Arduino Uno receives the prediction, rotates a servo motor to direct waste into the correct bin, and displays the result on an LCD.

A Gradio interface allows users to upload or capture an image, making the system intuitive and offline-friendly. The system completes classification and sorting within seconds, performs well under different lighting conditions, and provides real-time feedback.

Testing on 100+ waste samples showed 91% classification accuracy, fast servo response, and consistent LCD feedback. Users found the system reliable and easy to use. Compared to traditional solutions, Trash-Scan is cheaper, more accurate, portable, and more interactive.

Future enhancements include expanding classification to more waste types (glass, metal, e-waste), adding wireless connectivity, integrating mobile apps and solar power, and including features like UV disinfection, voice feedback, and smart-city data integration. Overall, Trash-Scan presents an affordable, scalable, and user-friendly approach to improving waste management and promoting environmentally responsible behavior.

Conclusion

The Trash-Scan project provides a practical solution for smart waste segregation using a combination of machine learning and affordable electronics. By accurately classifying waste as dry or wet and automating the sorting process using Arduino hardware, the system addresses key challenges in daily waste handling. The model is simple, low-cost, and efficient, making it a viable alternative to more expensive and complex systems currently available. It promotes better hygiene, reduces manual labour, and encourages eco-friendly practices. The flexibility of the design allows for further development and real-world deployment in a variety of settings. Trash-Scan proves that even small-scale, student-led innovations can make a big difference in solving real-world environmental issues.

References

[1] AI-Powered Waste Classification Using Convolutional Neural Networks (CNNs) Chan Jia Yi, Chong Fong Kim Faculty of Data Science and Information Technology, INTI International University, Nilai, Malaysia. (IJACSA) International Journal of Advanced Computer Science and Applications, Vol. 15, No. 10, 2024 [2] DeepWaste: Applying Deep Learning to Waste. Classification for a Sustainable Planet . Yash Narayan The Nueva School yasnara@nuevaschool.org . arXiv:2101.05960v1 [cs.LG] 15 Jan 2021

Copyright

Copyright © 2025 Madhuri Barhate, Samarth Bidkar, Riddhi Bhutada, Ritesh Biradar, Bhumi Chitte, Bhuvan Uke, Rohan Bhusare. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.