Target detection method and system

Abstract

The invention discloses a target detection method and system, and the method comprises the steps: carrying out the initialization size classification of a sample image, and determining the number of convolution layers in a network structure according to the classification number in a clustering result, so as to determine a feature extraction network; inputting a to-be-detected image into the feature extraction network, and determining feature maps of different sizes; respectively inputting each feature map into a region generation network to generate a corresponding candidate region feature map; determining a target region feature map according to each feature map and the corresponding candidate region feature map; and performing classification detection on the target region feature map to determine a target detection result. According to the method, the detection of the small target object is realized, the method adapts to the target objects of different sizes, the extraction precision of bottom-layer detail information is effectively improved, the classification and position regression of the multi-scale feature map are realized, the detection precision of the target objects of various sizes is effectively improved, and particularly, the problems of missing detection and false detection of the small target object are effectively relieved.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a target detection method and system. Background technique [0002] Object detection is a very important direction in the field of computer vision research, and is the basis of many computer vision applications. In recent years, with the rapid development of deep learning technology, object detection algorithms have also shifted from traditional algorithms based on manual features to deep neural networks. Network detection technology. Although the current object detection algorithm has greatly improved the accuracy of object detection, due to the predetermined size and number of anchor frames in the existing object detection algorithm, it usually cannot effectively cover target objects of different sizes, especially small-sized targets. As a result, missed detection and false detection of the target are caused. Contents of the invention [0003] In view of this, the embodiment ...

AI Technical Summary

Problems solved by technology

[0003] In view of this, the embodiment of the present invention provides a method and system for target detection, which solves the problem of missing and wrong detection of targets in the prior art because the size and number of anchor frames are predetermined and cannot cover target objects of different sizes. inspection problem

Images