Vehicle Blind-Spot Detection Using ROI and Multi-Scale Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing blind area detection systems in vehicles suffer from false detections due to challenges in accurately identifying targets at different distances, leading to reduced safety and insufficient reaction time for drivers.
Innovation Solution
A method involving multiple cameras to capture real-time images, determining a region of interest (ROI) based on vanishing points, and using a target detection model to generate detection boxes at different scales, followed by filtering based on overlap values to improve accuracy and reduce false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If cameras are installed around the vehicle to capture real-time images of blind areas, then the coverage of blind area detection is improved, but false detections increase due to difficulty in accurately identifying targets at different distances
Solution Approach 1:
The patent divides the blind area into multiple regions based on vanishing points and generates multiple scaled images (first scaled image and second scaled image) from the original image. This segmentation allows the detection model to process different regions at different scales, improving the ability to accurately identify targets at various distances while maintaining comprehensive coverage.
Solution Approach 2:
The patent introduces a scaling dimension by generating multiple scaled versions of the image (first scaled image with first scaling ratio and second scaled image with second scaling ratio). This adds a new dimension to the detection process, enabling the model to detect targets at different sizes and distances more accurately, thereby reducing false detections while maintaining coverage.
2Measurement precision
If multiple scaled images are generated and processed by the detection model, then target detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary actions by determining the region of interest (ROI) based on vanishing points before generating scaled images. This preliminary identification of critical regions allows the system to focus computational resources on the most important areas, reducing overall processing time while maintaining high detection accuracy through multi-scale processing of the ROI.
3Reliability
If detection boxes are filtered based on overlap values, then false positives are reduced, but the complexity of the detection algorithm increases
Solution Approach 1:
The patent implements a feedback mechanism by calculating overlap values between detection boxes and using this information to filter false positives. The overlap value calculation provides feedback on the quality of detection results, allowing the system to iteratively refine and improve detection accuracy while managing algorithm complexity through a systematic filtering approach.
Data Source
AI summary
A target detection method for blind areas of a vehicle includes acquiring a real-time image of one of the blind areas of a vehicle; determining a region of interest (ROI) image based on the real-time image; determine a image to be detected based on the ROI image and the real-time image; inputting the image to be detected to a target detection model to generate a first detection box in the first scaled image corresponding to a detection target in the real-time image and a second detection box in the second scaled image corresponding to the detection target in the ROI image; and determining a detection result for the blind area of the vehicle based on the first detection box and the second detection box.