Saliency-Based Video Editing for Accurate Highlight Selection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video editing technologies inaccurately identify highlight moments, often including non-interesting segments and excluding interesting segments due to issues with blur and shakiness in video footage.

Innovation Solution

A system that identifies salient regions within video frames based on their size and saliency criteria, determining salient segments for inclusion in a video edit, using deep learning models and saliency maps to enhance the accuracy of video segment selection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If automatic video editing is performed without saliency analysis, then the editing process is simple and fast, but the video segments selected do not accurately represent interesting content

Engineering Contradiction:

Improveaccuracy of video segment selectionVSAvoidcomplexity of editing system

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The video is divided into individual frames, and each frame is analyzed separately to identify salient regions. This segmentation allows the system to process complex visual information in manageable units, improving measurement precision without overwhelming system complexity

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Saliency analysis is performed in advance on all video frames before final segment selection. By pre-identifying salient regions and their sizes, the system prepares selection criteria that guide subsequent editing decisions, improving accuracy while organizing complexity into distinct preprocessing and decision stages

Inventive Principle:

Principle #10Preliminary action

2Reliability

If video segments are selected based on traditional criteria, then the editing process is quick, but interesting segments are excluded and non-interesting segments are included

Engineering Contradiction:

Improvereliability of segment selectionVSAvoidtime for video analysis

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system uses the size of salient regions as a key parameter to determine whether a frame contains interesting content. By changing from traditional criteria to saliency-based size thresholds, the system improves reliability of segment selection while maintaining efficient processing through clear quantitative thresholds

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If saliency analysis is performed on every video frame, then segment selection accuracy is improved, but processing time increases significantly

Engineering Contradiction:

Improveprecision of salient region identificationVSAvoidvideo processing speed

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces exhaustive mechanical analysis of every frame with a more efficient computational approach using saliency maps and size-based filtering. This substitution maintains precision in identifying interesting content while significantly improving processing speed through optimized algorithms

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20250210063A1Video edits using saliency

Publication Date: 2025.06.26 GOPRO INC

AI summary

Salient regions within video frames of a video may be identified. Sizes of salient regions within video frames may be determined and used to identify saliency frames in the video. Salient segments of the video may be identified using the saliency fames in the video. The salient segments of the video may be used to generate a video edit.