A signal light region automatic labeling method, device, storage medium and program product

By acquiring the spatial location and orientation information of the traffic light area and combining it with the calibration parameters of the roadside camera, the traffic light area is automatically marked in the image, which solves the problem of low efficiency of manual marking in the existing technology and realizes efficient and accurate automatic marking of traffic light areas.

CN122416451APending Publication Date: 2026-07-17TUS CLOUD CONTROL (BEIJING) TECH LTD

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
TUS CLOUD CONTROL (BEIJING) TECH LTD
Filing Date
2026-03-13
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

In existing technologies, traffic light area marking relies on manual methods, resulting in low marking efficiency and a large workload in large-scale road network deployment scenarios. Furthermore, frequent adjustments or additions of cameras require remarking, increasing system deployment and maintenance costs.

Method used

By acquiring the spatial location and orientation information of the traffic light area, and combining it with the calibration parameters of the roadside camera, the spatial location information of the traffic light area is projected into the pixel coordinate system of the image. By utilizing the matching relationship between the traffic light orientation information and the camera observation direction, the target traffic light marking area is automatically determined.

Benefits of technology

It enables batch automatic labeling of traffic light areas, improving labeling efficiency, reducing manual labeling costs, and enhancing the accuracy and reliability of labeling results. It is suitable for large-scale traffic perception system deployment with multiple intersections and multiple cameras.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122416451A_ABST
    Figure CN122416451A_ABST
Patent Text Reader

Abstract

本申请提供一种信号灯区域自动标注方法、设备、存储介质及程序产品,方法包括:获取信号灯区域信息,信号灯区域信息包括信号灯区域的空间位置信息及信号灯朝向信息;获取路侧相机采集的图像以及路侧相机的标定参数;根据标定参数,将信号灯区域的空间位置信息投影至图像的像素坐标系中,得到图像中的候选信号灯标注区域;根据信号灯朝向信息与路侧相机的观测方向之间的匹配关系,从候选信号灯标注区域中确定目标信号灯标注区域。本申请能够利用已有的信号灯空间信息与相机标定参数,对多个信号灯对象及多路侧相机采集的图像自动执行投影标注处理,从而实现对图像中信号灯区域的批量自动标注,降低了大规模路网部署场景下的人工标注成本。
Need to check novelty before this filing date? Find Prior Art