Automated Surveillance Systems: Design, Monitoring, and Privacy Preservation

Abstract

The importance of automated surveillance technologies in terms of surveillance of traffic intelligence has been significantly increased within smart cities, as the technologies allow for traffic analysis, traffic congestion management, accident detection, and efficient traffic control. Nevertheless, contemporary traffic surveillance methods usually encounter problems such as inefficient traffic detection, lack of robustness in changing traffic environment, and inadequate data privacy protection, hindering their applications in practical cases in the area of large-scale smart cities. With regard to the problems mentioned above, a model of privacy-preserving intelligent traffic surveillance method utilizing Hunger Games Search-optimized Faster Region-based Convolutional Neural Networks (HGS-Faster R-CNN) is presented in the paper. It is suggested that the HGS-Faster R-CNN model can be able to enhance privacy-preserving traffic surveillance by determining the best hyperparameters of Faster R-CNN utilizing the principles of HGSO. The evaluation process can be performed based on a traffic surveillance privacy dataset. The pre-processing phase makes use of Gabor filters to improve the quality of the images obtained and highlight edges and texture information. Following this, feature extraction by the Histogram of Oriented Gradients (HOG) technique is done to extract motion and structural features of the objects while protecting user privacy. From the experimental outcomes, the developed model shows high accuracy levels of 98.3%, high recall rate of 98%, and F1-score of 98.29%. This model is superior to other models since it can detect traffic objects efficiently at a lower detection latency period. The implementation of the model involves the use of programming languages such as Python, TensorFlow, Keras, OpenCV, and CUDA. In general, the proposed model proves to be effective, scalable, and privacy-preserving, thus becoming quite appropriate for real-world smart city traffic surveillance applications.