Multi-Scale Object Detection Using Critical Region Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection systems in unmanned aerial vehicles face challenges in achieving high accuracy and efficiency due to the reduction in image size for timely processing, which compromises detection precision.
Innovation Solution
A multi-scale object detection method that performs parallel object detection processes on both the entire image frame and critical regions, utilizing different processors (GPU and CPU) to enhance accuracy while maintaining real-time operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the image size is reduced to complete detection by the target time, then the processing time is reduced, but the accuracy in detecting objects is lowered
Solution Approach 1:
The patent divides the image processing into two segments: critical regions (where objects are likely to be located based on previous frame information) and non-critical regions. The critical region extractor identifies and extracts only the critical regions from the current frame, performing detailed object detection only on these smaller regions while using reduced-resolution processing for the rest of the image. This segmentation allows the system to maintain high detection accuracy in critical areas while reducing overall processing time and computational load.
2Measurement precision
If object detection is performed on the entire image frame with high resolution, then the detection accuracy is improved, but the power consumption increases
Solution Approach 1:
The patent extracts and processes only the critical regions that contain potential objects of interest, rather than processing the entire high-resolution image frame. The critical region extractor identifies regions where objects are likely to be located based on information from previous frames, and the multi-scale object detector performs detailed analysis only on these extracted critical regions. This extraction approach maintains high detection accuracy where needed while significantly reducing the computational power and energy consumption required for object detection.
3Productivity
If parallel processing is used for multiple detection processes, then the productivity is improved, but the device complexity increases
Solution Approach 1:
The patent implements dynamic parallel processing where the system adaptively assigns different processing tasks to different processors based on the current frame's characteristics. The critical region extractor and multi-scale object detector operate in parallel on different regions of the image, with the GPU handling computationally intensive multi-scale detection on critical regions while the CPU manages overall coordination and non-critical region processing. This dynamic parallel architecture improves processing throughput while managing system complexity through specialized task distribution.
Data Source
AI summary
There is provided a multi-scale object detection device. The device includes an image frame acquisition unit for acquiring a plurality of consecutive image frames, a critical region extractor for extracting at least one second critical region from a current image frame based on at least one first critical region extracted from a previous image frame among the consecutive image frames, a multi-scale object detector whose operation involves a first object detection process for the current image frame and a second object detection process for the at least one second critical region, and an object detection integration unit for integrating the results of the first and second object detection processes.