Design and implementation of esp32-cam and yolov8-based intelligent camera systems for the detection of littering behavior

Abstract

This research aims to design and implement an intelligent camera based on ESP32-CAM and YOLOv8 algorithm to detect objects of people and garbage in the river area. The system is built by integrating a PIR sensor as a motion trigger, an LM2596 step-down module as a voltage regulator, and a PHP/MySQL backend for storing detected data. The ESP32-CAM is programmed using an Arduino IDE to be able to capture images when the PIR sensor is active and send them to the server. Furthermore, YOLOv8 running with Python (FastAPI) is used to analyze the images and classify the detected objects. The test results showed that the system worked as designed, with an average response time of ±10.27 seconds from motion detection to recording results to the database. YOLOv8's detection accuracy reached 96% for people objects and 90% for junk objects, with an average overall accuracy of 93%. Although this system has not been able to detect waste disposal behavior directly, this research has succeeded in providing an initial foundation towards the development of environmental monitoring systems based on behavior detection in the future.