Paper Title
A Reliable and Robust Deep Learning Model for Effective Recyclable Waste Classification Using IoT

Abstract
Waste management is a growing challenge, especially when it comes to sorting recyclable materials accurately. In this project, we’ve developed a reliable and efficient deep learning model that can automatically classify waste into five categories: paper, plastic, glass, metal, and no object. Using Python and its libraries like TensorFlow, NumPy and neural network like CNN. The model was trained on image data to recognize different types of recyclable waste with high accuracy compared to other models available in the market. To make this system operational, we developed an IoT-based hardware platform based on a Raspberry Pi, along with a camera module to take pictures, a proximity sensor to sense approaching waste, a buck converter for power conditioning, and an LCD display to display the output and predefined messages like “capturing image, processing, dumping the waste”, A servo motor is utilized to mimic the sorting the waste according to the classification. Through the combination of deep learning and real-time hardware integration, this project presents an intelligent, affordable, and automated means of waste classification. Where the outputs shows the accuracy and potential of the model. The aim is to facilitate cleaner recycling and move toward a greener, more sustainable tomorrow where human’s interaction can be minimized in waste industry. Keywords - Waste Classification, Deep Learning, Convolutional Neural Network (CNN), Raspberry Pi.