Automatic Video Edit Detection Using Machine Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Many users who upload videos to sharing sites lack the knowledge or ability to identify and correct issues such as incorrect rotation, poor lighting, poor contrast, and shakiness, which can hinder the quality of their videos, despite video editing software being capable of addressing these problems.

Innovation Solution

A system that uses machine learning techniques to extract features from videos, classify them for editing needs, and automatically apply suggested edits, including rotation, lighting, stabilization, and color balance adjustments, to improve video quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If video editing software is provided to users, then video quality can be improved, but users lack the knowledge and ability to identify and correct video issues

Engineering Contradiction:
Improvevideo qualityVSAvoiduser operation complexity
Core Design Contradiction:
Manufacturing precisionVSEase of operation

Solution Approach 1:

The system performs automatic video analysis and editing without requiring user intervention. The server automatically detects video issues, generates edit suggestions, and applies corrections, allowing the video to serve itself rather than requiring expert user operation.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

An intermediary system (the server with machine learning models) is introduced between the user and the video editing process. This intermediary automatically analyzes video quality, identifies issues, and applies corrections, shielding users from the complexity of video editing while maintaining quality improvement.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Manufacturing precision

If automatic video editing is implemented, then video quality improvement is achieved, but system complexity increases

Engineering Contradiction:
Improvevideo qualityVSAvoidsystem complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The complex video analysis and editing functions are extracted from the user's local environment and relocated to a remote server. This extraction allows the client device to remain simple while the server handles the computationally intensive machine learning models and video processing operations.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

A server-based intermediary system is introduced to handle the complexity of automatic video editing. The server hosts machine learning models for video analysis and manages the editing process, acting as a mediator between simple client uploads and complex processing operations.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Manufacturing precision

If manual video editing is required, then precise corrections can be applied, but time consumption increases significantly

Engineering Contradiction:
Improveediting precisionVSAvoidediting time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The system performs preliminary analysis of video content using machine learning models to pre-identify issues and generate edit suggestions before user review. This preliminary action automates the time-consuming analysis phase while maintaining precision through algorithmic detection of video problems.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The video editing process is automated to perform corrections without manual intervention. The system automatically detects video issues, determines appropriate corrections, and applies edits, eliminating the time required for manual video editing while maintaining correction precision through sophisticated algorithms.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP2817691B1Automatic detection of suggested video edits
Publication Date: 2024.05.08 GOOGLE LLC
  • EP2817691B1 patent drawingFigure 1A~1B
  • EP2817691B1 patent drawingFigure 2
  • EP2817691B1 patent drawingFigure 3

AI summary

Systems and methods are provided herein relating to video classification. A trained classifier can analyze a video for suggested edits. A plurality of features of the video can be analyzed based that determines whether the video is a good candidate for various type of editing. The suggested edits can be performed automatically or with the authorization of a user who submitted the video. The user can review the edited video and either approve or reject the edits. Using an automated process to suggest and perform video edits can improve the quality of videos within a video data store.