Smart Surveillance System for Illegal Garbage Dumping

Abstract

Listen

Translate article

Effective garbage management is a key challenge faced by urban areas across the world Garbage dumping in unauthorized places is a common problem in most cities, and it poses a serious threat to public health and the environment. To address this issue, we propose a real-world surveillance platform that detects and reports instances of garbage dumping in unauthorized areas using vision-based action recognition techniques. Our approach involves training a deep neural network on a dataset of garbage dumping actions to recognize and classify instances of this behaviour in real-time video footage. The system operates in three stages detection, tracking, and classification. In the detection phase, we use object detection techniques to locate potential dumping events in video frames. In the tracking phase, we track the detected object across subsequent frames to obtain a trajectory that helps in understanding the context of the action. In the classification phase, we classify the action as garbage dumping or non-dumping using a convolutional neural network. Our system achieves an accuracy of 96.2% on a test dataset, demonstrating its effectiveness in detecting garbage dumping events in real-world surveillance scenarios. The proposed system can be integrated into existing surveillance infrastructure to improve the efficiency of garbage management systems and enhance public health and environmental protection.