Gameplay Video Tracking with Perceptual Hashing for Scene Cuts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is difficult for users to simultaneously play video games and selectively record exciting or surprising content while also summarizing game progress, as identifying key events within the generated gameplay footage is challenging due to the vast amount of user-directed content.
Innovation Solution
An apparatus and method for video tracking that uses perceptual hashing to detect scene cuts and identify notable events in gameplay by generating and comparing image hashes, utilizing a database of hashes for notable events, and employing key points to confirm matches, reducing computational overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If users manually review and identify key events in gameplay footage, then identification accuracy can be maintained, but time consumption increases significantly due to the vast amount of user-directed content
Solution Approach 1:
The patent replaces manual mechanical review of gameplay footage with automated computer-based image processing. The system uses perceptual hashing to generate hash values from gameplay images, compares these hashes to identify scene cuts, and automatically detects key events without human intervention. This substitution of automated computational methods for manual review resolves the contradiction by maintaining identification accuracy through systematic analysis while dramatically reducing the time required to process vast amounts of gameplay content.
2Measurement precision
If comprehensive text detection is performed across all regions of gameplay images, then text identification accuracy improves, but computational overhead increases significantly
Solution Approach 1:
The patent applies local quality by performing text detection selectively in specific regions of gameplay images rather than uniformly across the entire image. The system identifies regions where text is likely to appear (such as UI elements, scoreboards, or notification areas) and concentrates computational resources on these localized areas. This approach maintains text identification accuracy by focusing detection efforts where text actually appears while significantly reducing the overall computational overhead compared to comprehensive full-image analysis.
Data Source
Figure 1
Figure 2A~2B
Figure 2C~2D
AI summary
A method of identifying text that appears within a sequence of images comprises the steps of defining one or more regions of respective images in which text will appear, detecting when text has appeared in one or more of the defined regions, identifying text within a defined region when text has been detected there, and selecting for output identified text that has appeared, or a respective image of the sequence of images that comprise such identified text.