Design and implementation of object image detection interface system based on PyQt5 and improved SSD algorithm

Abstract

With the explosion of image and video information, more and more image information needs to be processed. Image-based object recognition is widely used in various scenes. The system realizes object detection and recognition based on the improved SSD algorithm, through the interactive interface system designed by PyQt5, through the way of local images, videos and cameras, it can quickly identify 85 kinds of objects such as general objects and scenes, animals and plants, fruits and vegetables, and display key information such as object positioning. Aiming at the high requirement of object recognition accuracy, an improved target detection algorithm based on SSD is proposed in this paper, based on the SSD algorithm, replace the backbone network VGG16 of SSD with ResNet50, and increase the preselected number of output layers, it solves the problem that SSD algorithm has insufficient ability to extract small target features, thus, the defect of poor detection effect of small targets is improved.