Target detection method and storage medium based on depth map

Abstract

The invention discloses a specific target detection method based on a depth image and a storage medium, and the method comprises the steps: defining the size of a real candidate box through the size of a to-be-detected target object, and calculating a candidate box window traversal interval; Traversing the depth image to obtain a central point pixel coordinate of the candidate box; Acquiring a depth value of a central point of each candidate box, and performing screening to obtain an effective candidate box; Calculating the actual required frame length of the remaining valid candidate frames;And setting a filtering threshold, filtering out points with too large depth difference from the central point in the effective candidate frame, and carrying out deep learning preprocessing and deep learning subsequently. According to the invention, the step length of image traversal can be increased as much as possible; some invalid candidate boxes can be filtered out; the edge length of the candidate box is calculated according to the depth value of the central point and the real size of the target object, the generation of multi-scale candidate boxes at the same position can be prevented, alarge amount of calculation amount is saved, and the method provides good convenience for rapid target detection.

Description

technical field [0001] The present invention relates to the field of image detection, in particular to a target detection method based on a depth image, which can greatly reduce the initial number of candidate frames on the depth image, thereby greatly reducing the number of model calculations and improving detection efficiency. Background technique [0002] Image classification, detection and segmentation are the three major tasks in the field of computer vision. As the most basic part, the current main target detection image objects are RGB images, and with the development of structured light technology and TOF technology, depth images have gradually become a new data source. [0003] In recent years, with the rapid development of deep learning technology, the speed and accuracy of specific target detection in images have been greatly improved. But it is far from being able to achieve the real-time detection effect of video images. But compared with the traditional targe...

AI Technical Summary

Problems solved by technology

[0008] (3) There may be a large amount of overlap between the candidate boxes, resulting in an increase in the number of candidate boxes;

[0009] (4) The initial convolution image is too large and the amount of calculation is large

Images