<![CDATA[Object detection is a method to recognize the class and location of objects in an image. The main challenge is integrating complex algorithms into lightweight and portable hardware, especially with expensive sensor and camera technologies. This research aims to develop an object detection system using the ESP-32 Cam for robotics monitoring and security. The focus is on utilizing the Yolov5 model transformed into TensorFlow Lite for integration with ESP32 AI CAMERA, expected to detect objects in real-time at a low cost. The methodology includes collecting 1710 datasets from 27 images, dividing the data into 70% training, 20% validation, and 10% testing, and labeling the dataset in Roboflow. The object detection model uses Yolov5, transformed into TensorFlow Lite, and implemented in ESP32 AI CAMERA with ESP-32 Cam as the microcontroller. Model evaluation shows high performance with mAP 95%, precision 97%, and recall 100%, indicating high accuracy. The research successfully develops an efficient and affordable object detection system with ESP-32 Cam and TensorFlow Lite from Yolov5. This integration enables the development of wheeled robots capable of real-time object detection, providing an effective solution for portable robotics monitoring and security.]]>
Copyrights © 2024
