Slow-Motion Video Highlight Detection Using ROI Gray Centroids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile devices struggle to accurately and efficiently detect highlight moments in slow-motion video recording due to limited DRAM buffer capacity and challenges in determining velocity thresholds, often missing the highlight moments or requiring extensive computation, which can block the processing pipeline.
Innovation Solution
An electronic device captures a sequence of images at a high frame rate and uses a gray centroid of a region of interest (ROI) to detect the initiation of a highlight moment, storing a plurality of highlight images in memory for slow-motion playback, without relying on complex algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video content is recorded at high frame rate in DRAM buffer, then slow-motion playback quality is improved, but buffer capacity is limited and recording duration is short
Solution Approach 1:
The system pre-records video content at high frame rate into the DRAM buffer before the highlight moment occurs, so that when detection triggers, the buffer already contains sufficient frames for slow-motion playback. This preliminary recording ensures high-quality slow-motion capture without exceeding buffer capacity limits.
2Ease of operation
If manual recording trigger is used, then user control is improved, but highlight moment detection accuracy deteriorates due to human reaction latency
Solution Approach 1:
The system continuously monitors video frames for highlight moment characteristics (such as sudden motion changes or specific visual patterns) and automatically triggers recording when detected. This feedback mechanism eliminates human reaction latency while allowing users to maintain manual control by overriding the automatic trigger if needed.
3Measurement precision
If automatic velocity threshold detection is used, then highlight moment detection accuracy is improved, but computation time increases and processing pipeline is blocked
Solution Approach 1:
The detection process is divided into separate stages: initial simple motion detection to identify potential highlight moments, followed by more sophisticated velocity threshold analysis only on those candidate frames. This segmentation reduces overall computation time and prevents pipeline blocking while maintaining detection accuracy.
Solution Approach 2:
Instead of applying full velocity threshold detection to every frame, the system performs partial detection on selected frames that show preliminary signs of highlight moments. This reduces computational load and processing time while still achieving accurate highlight moment identification.
Data Source
AI summary
Provided are a method for detecting a highlight moment, an electronic device, and a non-transitory computer-readable medium. The electronic device captures a sequence of images of a field of view with a first frame rate, the sequence of images including at least an initiation of the highlight moment. In response to determining, based on at least a gray centroid of a region of interest (ROI) of one image of the sequence of images, that an object appears in the ROI of the sequence of images from the one image, the electronic device determines the one image as an initial image at which the highlight moment is initiated. The electronic device further stores multiple highlight images in association with the highlight moment in the memory of the electronic device. The stored highlight images correspond to the first frame rate.