Research on real-time image processing system combining convolutional neural network and edge computing

With the rise of the Internet of Things (IoT) and edge computing technologies, traditional cloud-dependent convolutional neural network (CNN) image processing methods are facing the challenges of latency and bandwidth bottlenecks. In this study, we propose a real-time image processing system that combines CNN and edge computing, which enables the CNN model to run efficiently on resource-constrained edge devices through model pruning and quantisation optimisation. Experimental results show that the system significantly outperforms traditional methods in terms of inference speed, processing frame rate, and energy-efficiency ratio, especially in application scenarios requiring high real-time performance, such as intelligent surveillance, autonomous driving, and industrial inspection. In addition, the system shows good robustness and stability under different network conditions. This study provides an efficient and low-latency solution for image processing in edge computing environments and lays the foundation for future intelligent applications.