Real Life Implementation of Object Detection and Classification Using Deep Learning and Robotic Arm

Описание: Real Life Implementation of Object Detection and Classification Using Deep Learning and Robotic Arm”
Presented at the International Conference on Recent Advances in Interdisciplinary Trends in Engineering & Applications
Authors: Yogesh Kakde, Niketan Bothe, Aniket Paul

Abstract:

This project demonstrates the integration of deep learning concepts with Arduino programming, forming a comprehensive framework. Deep learning has become one of the most favorable domains in today's era of computer science. In this paper, we discuss the implementation of deep learning concepts using an Arduino Uno in conjunction with a robotic application. For object detection and classification, a robotic arm is employed to automatically identify and sort different objects—in this case, fruits. A camera captures images of the fruits, which are then processed by a model based on a Convolutional Neural Network (CNN). Upon detecting the object in the image, the trained model sends a specific signal to the robotic arm via the Arduino Uno, prompting it to place the detected object into the appropriate basket. Through this method, the project successfully recognizes and classifies two different fruits, placing them into separate baskets accordingly. This endeavor showcases the effective combination of deep learning and Arduino programming, offering a versatile framework applicable to various real-life problems. Post-implementation, the system achieved an object detection accuracy of up to 99.22%.

Keywords: Arduino; Convolutional Neural Network; Robot; Image Processing; Deep Learning; Feature Extraction