Adaptive Object Detection Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods are computationally intensive and inefficient in detecting characteristics of images, particularly in adapting to changes in object detection over sequential image frames.
Innovation Solution
A method and system that perform feature extraction and object detection on image frames, with the ability to modify processing parameters based on characteristics of object detection data from previous frames, such as object appearance, disappearance, movement, or changes in position, to optimize resource consumption and detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If feature extraction and object detection operations are performed on all image frames using fixed parameters, then detection consistency is maintained, but computational resources are wasted on frames where objects have not changed
Solution Approach 1:
The patent applies dynamics by making the feature extraction operation adaptive rather than static. The system dynamically adjusts whether to perform feature extraction on subsequent image frames based on object detection results from previous frames. When no object is detected or the object has moved out of the region of interest, the feature extraction operation is modified or skipped, allowing the system to adapt its processing behavior to the actual content of the images.
Solution Approach 2:
The patent implements parameter changes by modifying at least one parameter of the feature extraction operation in response to characteristics of object detection data. This could involve changing extraction thresholds, resolution levels, or completely disabling extraction for certain frames based on detection outcomes, thereby optimizing resource usage while maintaining detection accuracy when needed.
2Productivity
If feature extraction parameters are adapted based on object detection data, then resource consumption is optimized, but system complexity increases
Solution Approach 1:
The patent applies feedback by using object detection data from previous frames to inform and adjust the feature extraction operation on subsequent frames. The system creates a closed-loop control where detection results feed back into the extraction process, allowing the feature extraction parameters to be modified based on actual detection outcomes. This feedback mechanism enables intelligent resource allocation without requiring complex external control systems.
3Reliability
If feature extraction is performed on every image frame, then no objects are missed, but processing time increases
Solution Approach 1:
The patent applies preliminary action by performing object detection on the current image frame before deciding whether to perform feature extraction on the next frame. This allows the system to anticipate when feature extraction will be necessary and prepare accordingly, skipping extraction operations on frames where objects are unlikely to appear or have already been detected, thus reducing overall processing time while maintaining detection completeness.
Data Source
AI summary
A method of image processing in a computing device comprising an object detection system. The method comprises processing a first image frame, a second image frame, and one or more further image frames, the second image frame having been captured subsequent to the first image frame and the one or more further image frames having been captured subsequent to the second image frame. A feature extraction operation is performed on the first image frame and the second image frame. An object detection operation is performed, using feature data from said feature extraction operation, to produce object detection data. At least one parameter of said feature extraction operation, for processing said one or more further image frames, is modified in response to a characteristic of said object detection data.