Shielded traffic sign detection method and system based on MR-SSD

Abstract

The invention discloses an MR-SSD-based occluded traffic sign detection method and system, and the method comprises the steps: carrying out the image filling through a collaborative modulation generative adversarial network, and carrying out the image enhancement at the same time; constructing an MR-SSD feature extraction network, extracting a feature map in a backbone module, inputting the feature map into a feature fusion module for feature fusion, and performing classification and positioning by a prediction module to obtain a preliminary prediction classification and positioning result; adjusting according to the preliminarily predicted classification confidence value and the intersection-to-union ratio IoU between the priori frame and the ground truth value frame belonging to the same area to obtain a secondarily predicted classification and positioning result; classifying the category to which the shielded traffic sign belongs and the positioning result to obtain a final predicted classification and positioning result; continuously carrying out network training; inputting an image to be recognized into the trained network, judging the type of the traffic sign contained in the image, and framing the specific position. The method can effectively improve the detection precision and speed of the shielded traffic sign.

Description

technical field [0001] The invention relates to an MR-SSD-based occluded traffic sign detection method and system, and belongs to the technical field of image processing technology in computer vision. Background technique [0002] Traffic sign detection is an important part of the vehicle driving assistance system, and solving the problem of traffic sign detection is of great significance to the development of the vehicle driving field. In the vehicle driving system, the weather environment of the traffic sign will have a great influence on the decision-making made by the driving system. Bad weather conditions will lead to problems such as low visibility, and snow, messy leaves, and sand and dust accumulation will block the traffic at the same time. In order to effectively reduce violations and traffic accidents, the recognition system needs to recognize the correct blocked traffic sign indication in real time. [0003] In 2017, a method for calculating the occlusion probab...

AI Technical Summary

Problems solved by technology

[0011] In order to overcome the defect that the accuracy of the existing traffic sign detection method is limited under the influence of two factors of bad conditions and occlusion, which leads to a decrease in the detection accuracy of traffic signs, the present invention provides an MR-SSD-based occluded traffic sign detection method And system, effectively improve the detection accuracy and speed of blocked traffic signs

Images