Intersection traffic sensing methods, devices, electronic equipment and storage media
By embedding a geometric prior convolution module into the object detection model to extract semantic features and edge gradient features in parallel, the problem of visual data damage in intersection traffic perception is solved, and highly robust object detection is achieved in damaged scenarios.
CN122090246APending Publication Date: 2026-05-26GRG INTELLIGENT TECH SOLUTION CO LTD
View PDF 0 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GRG INTELLIGENT TECH SOLUTION CO LTD
- Filing Date
- 2026-03-27
- Publication Date
- 2026-05-26
Smart Images
Figure CN122090246A_ABST
Abstract
This application discloses a traffic perception method, device, electronic device, and storage medium at an intersection, belonging to the field of image processing technology. The method includes: acquiring a traffic image of a target intersection; inputting the topologically structured traffic image into a pre-constructed target detection model; obtaining semantic features and edge gradient features of the topologically structured traffic image based on the traffic image using multiple topologically structured geometric prior convolutional modules included in the topologically structured target detection model; incorporating the topologically structured semantic features and edge gradient features to obtain enhanced features corresponding to the topologically structured traffic image; and detecting target objects in the topologically structured target intersection based on the enhanced topological features to obtain the target detection result. This application improves the robustness of target detection in intersection scenarios where visual data is impaired.
Need to check novelty before this filing date? Find Prior Art