Non-Reference AI for Video Original Resolution Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing metadata-based methods for determining video resolution are unreliable in distinguishing between genuine and fake 4K videos, and traditional MOS comparison methods are time-consuming and expensive.
Innovation Solution
A non-reference video-based AI model trained through deep learning on a dataset of videos with various resolutions, measuring quality score differences to predict the original resolution by downscaling and upscaling video clips, excluding quality defects and using tolerance steps to ensure accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If metadata-based methods are used to determine video resolution, then the resolution can be easily and quickly obtained, but the method cannot distinguish between genuine and fake 4K videos
Solution Approach 1:
The patent introduces a third-party AI model as an intermediary to verify video resolution authenticity. The model takes downscaled video clips as input and outputs quality scores that indicate whether the video is genuine or fake 4K, resolving the contradiction between fast metadata-based detection and accurate authenticity verification
Solution Approach 2:
The patent replaces the traditional mechanical/metadata-based resolution reading method with an AI-based quality assessment system. Instead of relying on metadata fields, the system uses deep learning models to analyze video content and determine authenticity, substituting a simple reading operation with an intelligent analysis system
2Measurement precision
If MOS comparison methods are used to identify fake 4K videos, then resolution authenticity can be accurately assessed, but the process is time-consuming and expensive
Solution Approach 1:
The patent performs preliminary actions by training the AI model offline on a large dataset of genuine and fake 4K videos. This pre-training enables the model to make rapid predictions on new videos without requiring time-consuming MOS comparisons, transferring the time investment from the authentication phase to the model development phase
Solution Approach 2:
The patent creates a computational model that replicates the complex MOS comparison process. The AI model learns from numerous MOS comparison examples and creates an internal representation that allows it to quickly assess video authenticity without actually performing repeated MOS comparisons, effectively copying the decision-making process in a more efficient form
Data Source
AI summary
A method for predicting the original resolution of video contents using a non-reference video-based AI model built by training a training dataset of video contents having various resolutions through a deep learning method to overcome the limitations of existing metadata-based resolution determination methods. The method includes dividing the video contents that are resolution prediction targets into video clips of a fixed time length; downscaling the resolution of each of the video clips to a predetermined resolution; measuring the quality of video clips that have been downscaled using a non-reference video-based AI model in order from low to high resolution, and calculating a quality score difference between video clips of two neighboring resolutions in order from low to high resolution; predicting a resolution of each of the video clips based on the quality score difference; and aggregating the resolutions of each of the video clips to predict an original resolution of the video contents.