Target detection and visual positioning method based on YOLO series

Abstract

The invention discloses a target detection and visual positioning method based on a YOLO series. The method comprises the steps: firstly, obtaining an RGB color image and a depth image of a to-be-captured target; inputting the acquired RGB color image into a YOLO series target detection model built based on a darknet framework to obtain coordinate information, category and confidence; and combining the coordinate information with the depth image information to solve the spatial three-dimensional coordinates of the to-be-detected target. According to the positioning method based on the target detection algorithm provided by the invention, two-dimensional positioning can be carried out on the RGB color image acquired by the depth camera by using a YOLO series target detection algorithm deployed on a darknet framework, and three-dimensional positioning is realized by combining depth information acquired at a special position of the camera. Compared with other target detection algorithms, the YOLO series target detection algorithm has the advantages that the detection speed is higher, but the detection precision is not high, the target detection precision can be ensured, and the detection speed is increased.

Description

technical field [0001] The invention belongs to the fields of machine vision, visual positioning, target detection and deep learning, in particular to a method for target detection and visual positioning based on the YOLO series. Background technique [0002] In recent years, as the technology of deep learning has become more and more mature, the update speed of the target detection model is also accelerating. The existing target detection models can be divided into two categories, one-stage detection algorithm, such as SSD, YOLO, etc. ;Because it does not require the region proposal stage, it can directly generate the category probability and position coordinate value of the object, and the final detection result can be directly obtained after a single detection, which is characterized by faster detection speed; the other type is two-stage Detection algorithms, such as Faster-RCNN, Fast-RCNN, etc.; this type of detection algorithm divides the detection problem into two stag...

AI Technical Summary

Problems solved by technology

[0002] In recent years, as the technology of deep learning has become more and more mature, the update speed of the target detection model is also accelerating. The existing target detection models can be divided into two categories, one-stage detection algorithm, such as SSD, YOLO, etc. ;Because it does not require the region proposal stage, it can directly generate the category probability and position coordinate value of the object, and the final detection result can be directly obtained after a single detection, which is characterized by faster detection speed; the other type is two-stage Detection algorithms, such as Faster-RCNN, Fast-RCNN, etc.; this type of detection algorithm divides the detection problem into two stages, first generating candidate areas, and then classifying the candidate areas, which is characterized by low error rate and low missed detection rate, but the speed relatively slow

[0003] Existing target detection and visual positioning systems based on deep learning, such as the Chinese patent application "Target Detection and Positioning Method Based on YOLOv3 and OpenCV (CN111563458A)", can only use the YOLOv3 algorithm of the YOLO series for target detection, and the applicability is low

Images