Traffic Sign Boundary Recognition via Image Segmentation and Pre-trained Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for recognizing traffic signs in smart cars and autonomous driving systems lack accuracy, particularly due to factors like varying lighting conditions, which affect the precise detection of boundary points.
Innovation Solution
A method and apparatus that acquire initial boundary points in images, segment the images into areas, and use a pre-trained traffic sign boundary point recognition model to obtain update boundary points, improving accuracy by training the model with sample images and adjusting parameters based on prediction accuracy rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing technology is used for traffic sign recognition, then the system can detect traffic signs, but the recognition accuracy is insufficient
Solution Approach 1:
The patent divides the image processing into multiple stages: initial boundary point detection, image segmentation based on initial points, and refined boundary point recognition in each segmented region. This segmentation approach allows the system to focus computational resources on specific areas, improving both accuracy and reliability of boundary recognition while managing complexity.
Solution Approach 2:
The patent implements preliminary action by first detecting initial boundary points and segmenting the image before performing detailed boundary recognition. The pre-trained model is prepared in advance with sample images, and the segmentation creates focused regions for subsequent analysis. This preliminary processing establishes a solid foundation that enhances final recognition accuracy and reliability.
2Measurement precision
If image segmentation is performed based on initial boundary points, then the processing can be more focused, but the computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the to-be-processed image into multiple image segmentation areas based on initial boundary points. Each segmentation area is then independently processed by the pre-trained model to obtain update boundary points. This segmentation strategy improves detection precision by focusing on specific regions while managing computational complexity through parallel processing of divided areas.
Solution Approach 2:
The patent uses partial action by applying the pre-trained model only to specific image segmentation areas rather than the entire image. This selective processing approach maintains high boundary point detection precision in critical regions while reducing overall computational complexity compared to processing the complete image uniformly.
3Measurement precision
If a pre-trained model is used for boundary point recognition, then the recognition accuracy improves, but the training and model preparation time increases
Solution Approach 1:
The patent implements preliminary action by training the boundary point recognition model in advance using sample images with known boundary points. The pre-trained model captures boundary characteristics and patterns beforehand, enabling accurate recognition during actual operation without requiring real-time training. This approach achieves high recognition accuracy while minimizing time loss during deployment.
Solution Approach 2:
The patent uses copying by creating a pre-trained model that encapsulates boundary recognition knowledge learned from sample images. This model copy can be repeatedly applied to different images without retraining, maintaining high accuracy while avoiding repeated time investment in training. The model serves as a reusable artifact that transfers learning results across multiple applications.
Data Source
AI summary
A method and an apparatus for recognizing a boundary of a traffic sign are provided. The method includes: acquiring at least one initial boundary point of a traffic sign in a to-be-processed image, the initial boundary point being used to mark a boundary of the traffic sign; dividing the to-be-processed image into at least one image segmentation area based on the at least one initial boundary point, the image segmentation area including at least one initial boundary point; importing, for an image segmentation area in the at least one image segmentation area, the image segmentation area into a pre-trained traffic sign boundary point recognition model to obtain an update boundary point corresponding to the image segmentation area; and determining a boundary point of the traffic sign based on all the update boundary point corresponding to the image segmentation area.