Method and device for assisting target detection through wireless positioning

Abstract

The invention discloses a method for assisting target detection by using wireless positioning. The method comprises the following steps of: acquiring two-dimensional matrix information of an image andpositioning information obtained by an antenna array; calculating coordinates of each positioning result in the image through an imaging principle; obtaining a positioning area corresponding to eachpositioning result in the image according to the calculated coordinates and an obtained distance; outputting a detection result which is not subjected to non-maximum suppression through a deep learning detection method; calculating the overlap ratio gamma of the final detection result and the positioning area; modifying the confidence value of each positioning result; and carrying out improved non-maximum suppression on the detection result, so as to finish target detection assisted by wireless positioning. By combining two technologies of target detection and wireless positioning, the methodof the invention can assist in solving the problem that the detection number of final results of target detection is too large or too small, and the problem that images such as ghosting, reflection and the like are detected as scenes of real objects.

Description

technical field [0001] The invention relates to the fields of wireless positioning and target detection, in particular to a method and device for assisting target detection by using wireless positioning. Background technique [0002] Wireless positioning is an important technology in the Internet of Things, and with the popularity of various mobile devices and wireless tags (RFID), wireless positioning also has a broad application prospect. Through the content information contained in the wireless signal, the positioning antenna can first obtain the unique identifier of each object. Then, through the frequency and phase contained in each signal, the angle and distance from the target object to the antenna array can be inferred by calculation. Through the principle of horizontal angle, vertical angle and distance imaging, the center of the target and the approximate size in the image can be solved to provide additional information based on the image for the technology in com...

AI Technical Summary

Problems solved by technology

[0004] However, the existing deep learning-based target detection methods also have deficiencies: first, there may be multiple detections or missed detections; second, the generalization of models trained in different environments is not enough; finally, deep learning target detection requires a confidence value Threshold, but the threshold in different environments needs to be adjusted

Images