Video Region of Interest Detection via Frame Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting regions of interest (ROI) in videos suffer from inaccurate detection results due to mismatched tracking and detection results, leading to false or missed detections.
Innovation Solution
A method and apparatus that categorize video frames into detection and tracking frames, using algorithms like SSD and Kalman filtering to detect and track ROI, and updating the tracking ROI based on detection results from previous frames to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If tracking is performed on every frame, then continuity is maintained, but detection accuracy deteriorates due to mismatched tracking and detection results
Solution Approach 1:
The patent segments the video processing into two distinct types of frames: detection picture frames where ROI detection is performed, and tracking picture frames where tracking is performed. This segmentation allows each frame type to be processed according to its specific requirements, resolving the contradiction between maintaining tracking continuity and ensuring detection accuracy.
Solution Approach 2:
The patent uses ROI detection results from detection picture frames as feedback to correct and update the tracking results from tracking picture frames. This feedback mechanism ensures that tracking continuity is maintained while detection accuracy is improved by continuously refining the tracking results based on actual detection findings.
2Measurement precision
If ROI detection is performed on all frames, then detection accuracy is maintained, but computing resource consumption increases
Solution Approach 1:
The patent segments frames into detection picture frames and tracking picture frames, applying ROI detection only to detection picture frames while using tracking for tracking picture frames. This segmentation significantly reduces computing resource consumption while maintaining detection accuracy through the selective application of detection algorithms only where necessary.
Solution Approach 2:
The patent uses tracking results as a copy or approximation of ROI detection results for tracking picture frames, avoiding the need to perform computationally intensive ROI detection on all frames. This copying approach maintains sufficient detection accuracy while dramatically reducing computing resource consumption.
Data Source
AI summary
The present disclosure provides a method and apparatus for detecting a region of interest in a video, a device and a storage medium. The method may include: acquiring a current to-be-processed frame from a picture frame sequence of a video; detecting a region of interest (ROI) in the current to-be-processed frame, in response to determining that the current to-be-processed frame is a detection picture frame, to determine at least one ROI in the current to-be-processed frame; and updating a to-be-tracked ROI, based on the ROI in the current to-be-processed frame and a tracking result determined by a pre-order tracking picture frame; and tracking the current to-be-processed frame based on the existing to-be-tracked ROI, in response to determining that the current to-be-processed frame is a tracking picture frame, to determine at least one tracking result as the ROI of the current to-be-processed frame.