Adversarial sample generation method and system for robustness testing of sign recognition systems
Generating and embedding adversarial samples through image converters and optimization algorithms, the robustness of traffic sign recognition system under adversarial sample attack is solved, testing efficiency and coverage are improved, and the reliability of the system in actual operation is enhanced.
CN119580036BActive Publication Date: 2025-04-15武汉市计量标准质量研究院
2 Cites 0 Cited by
Patent Information
- Application Number
- CN202510115505.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-24
- Publication Date
- 2025-04-15
- Estimated Expiration
- 2045-01-24
AI Technical Summary
Technical Problem
Traffic sign recognition systems are difficult to maintain robustness when facing confrontational sample attacks, resulting in potentially serious traffic accident risks.
Method used
Iteratively update transformation parameters through image transformers and optimization algorithms, quickly generate diverse adversarial samples, and embed them in random backgrounds to simulate real scenes to improve test efficiency and coverage.
Benefits of technology
The generated adversarial samples cover a wider range of potential attack patterns and are close to the actual application environment, help test the robustness of the system in actual operation, and ensure the accuracy and targeting of generation of anti-perturbations through an adaptive bounding box filtering mechanism.
✦ Generated by Eureka AI based on patent content.
Abstract
The present invention discloses an adversarial sample generation method and system for robustness testing of a sign recognition system, the method comprising: transforming a standard sign image based on an image transformer and transformation parameters to obtain a first image; embedding the first image into a background image to obtain a second image; performing target detection on the second image based on a target detection model to obtain a plurality of bounding boxes; screening out a target bounding box from a plurality of bounding boxes; selecting a corresponding loss function to process the target bounding box to obtain a loss value; updating the current transformation parameters according to the loss value and using an optimization method, and iteratively processing the standard sign image based on the updated transformation parameters and the above process. The above adversarial sample generation method can quickly generate a variety of adversarial samples, so that the generated adversarial samples are closer to the actual application environment, which is helpful to test the robustness of the system in actual operation.
Need to check novelty before this filing date? Find Prior Art