Object Detection Based On Multi-Process

Object detection technology has been widely used. Traditionally, it usually uses a single process. However, a single process model can no longer meet the needs. With the complexity of object detection application scenarios, the entire system is divided into sub-modules, and data transmission between sub-modules is efficient. The multi-process architecture is adopted, which increases the cohesion of software modules and reduces coupling. Moreover, we need to adopt an efficient process communication and synchronization method, process communication to solve data transmission, and process synchronization to solve the problem of read and write conflicts. This article has made an innovation that introduces the concept of the shared memory frame buffer that uses multiple processes for object detection. It uses shared memory technology and process synchronization technology to complete inter-process communication and synchronization. Different processes use shared memory for efficient data transmission and use mutual exclusion locks to complete inter-process synchronization and control message communication. The server process is responsible for loading images and some general image processing algorithms. The client process only needs to implement a specific object detection algorithm, mount the shared memory and perform object detection on the image, and then send the detection results to the server. On the one hand, this solution reduces the time complexity. It only needs the main process to load the image into the shared memory, and other processes jointly use it to improve execution efficiency. If it is a very large image such as a GB level, the effect is more significant. On the other hand, sub-modules perform different functions respectively, which reduces code coupling and improves cohesion.