Traffic signal light identification method and device

Abstract

Embodiments of the present application disclose a traffic signal light identification method and apparatus. A specific embodiment of the method comprises the following steps: scaling an image to be processed collected by an image acquisition device according to at least one preset proportion to obtain at least one scaled image; Inputting the at least one zoom image to a pre-trained convolutional neural network for obtaining position information and class information of traffic lights corresponding to each zoom image in the at least one zoom image, wherein the convolutional neural network is used for extracting position information and class information of traffic lights displayed in the image; analyzing The obtained position information and category information to generate at least one candidate traffic signal lamp recognition result, and fusing the generated candidate traffic signal lamp recognition result to generate a traffic signal lamp recognition result corresponding to the imageto be processed. This embodiment improves the accuracy of traffic light identification.

Description

technical field [0001] The present application relates to the field of computer technology, specifically to the field of unmanned driving, and in particular to a traffic signal light recognition method and device. Background technique [0002] Unmanned vehicles are a new type of intelligent vehicles, also known as "wheeled mobile robots", which mainly use the ECU (Electronic Control Unit, electronic control unit), that is, the on-board brain control system to precisely control various parts of the vehicle Combined with calculation and analysis, the fully automatic operation of the vehicle can be realized, and the purpose of unmanned driving of the vehicle can be achieved. [0003] In order to ensure that unmanned vehicles can safely pass through intersections during driving, it is necessary to detect and identify traffic lights to make corresponding decisions. The existing methods are usually based on color space conversion and other methods to identify the color and shape ...

AI Technical Summary

Problems solved by technology

The existing methods are usually based on color space conversion and other methods to identify the color and shape of traffic lights. However, this method cannot take into account lighting changes, weather conditions, etc., resulting in many false detections and missed detections. There are traffic lights. Issues with low recognition accuracy

Images