Image traffic signal lamp detection method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of image detection / image classification / neural network, in particular to an image traffic information lamp detection method and device, electronic equipment and a computer readable storage medium, which can be applied to scenes such as aided driving and navigation / automatic driving / smart traffic to promote smart city construction. The image traffic information lamp detection method comprises the following steps: acquiring an original image; reducing the resolution of the original image to obtain a rough detection image; performing target area detection on the coarse detection image by using a coarse-grained detection convolutional neural network to obtain area information of a target area; matching the region information of the target region with the original image and obtaining an original image block corresponding to the target region in the original image; and using a fine-grained detection convolutional neural network to carry out traffic signal lamp area detection on the original image block so as to acquire position and category information of a traffic signal lamp area.

Description

technical field [0001] The present invention relates to the technical field of image detection / image classification / neural network, in particular to an image traffic information light detection method, device, electronic device and computer-readable storage medium, which can be applied to assisted driving and navigation / autonomous driving / intelligent transportation, etc. In the scene, promote the construction of smart city. Background technique [0002] With the continuous development of the social economy and the gradual improvement of the living standards of the public, computer technology has been widely promoted in production and life, especially computer image processing technology, which has become one of the important types of technology in the current computer application field. [0003] In computer image processing technology, traffic signal detection technology is of great significance to assisted driving and navigation / autonomous driving / smart traffic technology, ...

AI Technical Summary

Problems solved by technology

First of all, there is very little shared data on domestic traffic lights, and there is no direction information of traffic lights in foreign shared data, and the differences between Chinese and Western traffic lights also limit the progress of domestic research

Secondly, traffic lights are small targets, and their size is extremely small. Although the existing high-definition cameras can capture higher-resolution pictures, the extremely small relative scale also increases the difficulty of its detection.

The accuracy and robustness of traditional detection methods are poor, and some related technologies generally have the following shortcomings: First, the original high-resolution image (such as 2400×2400) is scaled into a general detection frame size (such as 512×512) and input, However, the loss of image information after shrinking will make it difficult to detect signal lights; the second is to input the original image resolution, which will greatly increase the calculation amount and time cost of the network; the third is to divide the complete original image into multiple image blocks and then input , which also greatly increases the calculation cost, and the calculation of many image blocks is invalid

And some related technologies are difficult to achieve a relatively good balance between detection accuracy and efficiency

Images