Video Player Format Selection Using ML for Quality and Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content sharing platforms face inefficiencies in video format selection due to suboptimal transcoding strategies, leading to unnecessary bandwidth usage without perceptual quality improvement, and the computational burden of calculating quality scores for multiple transcoded versions is infeasible on large scales.
Innovation Solution
A machine learning model, trained on historical video data, predicts quality scores for transcoded videos across various resolutions and configurations, enabling efficient format selection that optimizes bandwidth usage and user experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If quality scores are calculated for multiple transcoded versions of videos, then video quality assessment accuracy is improved, but computational burden increases making it infeasible on large scales
Solution Approach 1:
The system pre-calculates and stores quality scores for multiple transcoded versions of videos during an offline processing stage. This preliminary action allows the platform to have quality assessment results ready before users request video playback, eliminating the need for real-time quality score calculations and reducing online computational burden.
Solution Approach 2:
The system creates and stores quality score metadata as a separate copy alongside the video files. Instead of recalculating quality scores from scratch during video serving, the platform uses these pre-generated quality score copies to quickly determine optimal video formats, significantly reducing computational requirements during actual video delivery.
2Quantity of substance
If transcoding is performed to reduce storage requirements, then storage efficiency is improved, but bandwidth requirements for serving video increase
Solution Approach 1:
The system dynamically selects video formats and resolutions based on real-time conditions including user device capabilities, network bandwidth availability, and pre-calculated quality scores. This dynamic adaptation allows the platform to serve appropriately sized video files, optimizing the trade-off between storage efficiency and bandwidth consumption rather than using fixed transcoding strategies.
Solution Approach 2:
The system changes key parameters including video resolution, bitrate, and format based on quality score thresholds and user preferences. By adjusting these parameters intelligently, the platform can reduce storage requirements through efficient transcoding while simultaneously optimizing bandwidth usage by serving videos at appropriate quality levels without over-transcoding.
3Adaptability or versatility
If multiple transcoded versions are maintained for different resolutions, then adaptability to user devices is improved, but device complexity for format selection increases
Solution Approach 1:
The system uses pre-calculated quality scores as feedback to automatically determine the optimal video format for each user request. Instead of requiring complex real-time analysis or user input, the platform queries the stored quality scores, compares them against user preferences and device capabilities, and automatically selects the best format, simplifying the format selection process while maintaining high adaptability.
Solution Approach 2:
The system enables automatic format selection that serves itself by utilizing pre-computed quality metrics. The quality scores embedded in the video metadata automatically guide the format selection process without requiring external intervention or complex decision-making algorithms, reducing system complexity while maintaining versatility across different user devices and conditions.
Data Source
AI summary
A system and methods are disclosed for optimal format selection for video players based on visual quality. The method includes generating a plurality of reference transcoded versions of a reference video, obtaining quality scores for frames of the plurality of reference transcoded versions of the reference video, generating a first training input comprising a set of color attributes, spatial attributes, and temporal attributes of the frames of the reference video, and generating a first target output for the first training input, wherein the first target output comprises the quality scores for the frames of the plurality of reference transcoded versions of the reference video. The method further includes providing the training data to train a machine learning model on (i) a set of training inputs comprising the first training input and (ii) a set of target outputs comprising the first target output.