A Review on Image and Video Processing with IoT-Enabled Supervised Learning for Intelligent Surveillance Systems

Abstract

Road traffic accidents are a major concern for
global safety, where the delay in detection results in a rise in
deaths and economic costs. Recent developments in Artificial
Intelligence (AI), Deep Learning (DL), and Internet of Things
(IoT) technology have made it possible to automatically detect
accidents using visual and sensor data. The research works
analyzed include image-based techniques using transfer learning
and convolutional neural networks, video-based techniques
using spatio-temporal and self-supervised anomaly detection, and
IoT-based intelligent traffic monitoring systems. Additionally,
resource-effective video compression and transmission techniques
applicable to real-time surveillance systems are also analyzed.
The comparison between the research works shows differences
in input types, computational complexity, real-time capabilities,
and applicability to smart city settings. The major issues that
arise from the analysis of the research works include the lack
of data for detecting rare accidents, the absence of temporal
information in image-based techniques, high computational complexity
of video-based techniques, and the lack of integration
between detection and response systems. The paper concludes
by providing recommendations for future research on efficient,
scalable, and integrated accident detection systems for intelligent
transportation systems.