Anchor point generation method based on geometric attributes

Abstract

The invention discloses an anchor point generation method based on geometric attributes, and belongs to the field of deep learning. The anchor point generation method comprises the steps: 1, extracting the coordinates of all marked bounding boxes in an image, and carrying out the normalization of the widths and heights of the marked bounding boxes; 2, clustering the aspect ratio of the marked bounding box through the Euclidean distance; 3, clustering the size of the marked bounding box through the SIoU distance; and 4, generating anchor points of the dimension and aspect ratio vector by takingthe dimension and aspect ratio clustering center as a discontinuity point. According to the anchor point generation method, the labeled bounding box is normalized, and the aspect ratio and the size of the labeled bounding box are clustered, and the generated anchor point comprises the geometric attribute of the target, so that region suggestions can be extracted, and the target detection precision is improved.

Description

technical field [0001] The invention relates to the field of target detection, in particular to an anchor point generation method based on geometric attributes. Background technique [0002] With the rapid development of deep learning, models based on convolutional neural networks are widely used in the detection of objects in scenes. Generally, object detection methods based on convolutional neural networks can be divided into two-stage detection methods and one-stage detection methods. The two-stage detection method first extracts the region proposal, then outputs the region proposal to the detector and predicts the target location and category in the region, such as Faster RCNN (Ren Shaoqing et al., Faster RCNN: towards real-time object detection with region proposal networks, IEEE Trans on Pattern Analysis and Machine Intelligence, 2017,39(6):1137-1149); the one-stage detection method does not need to extract region proposals, and directly extracts target candidate regi...

AI Technical Summary

Problems solved by technology

Although deep models such as Faster RCNN and YOLO have good applicability to the detection of general targets, they do not consider prior knowledge such as target attributes, making them still have the problem of unsatisfactory detection accuracy in the application of specific scenarios.

Images