Dynamic Stream Selection via AI Event Prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current systems for live game broadcasting, such as esports, require manual selection and arrangement of multiple streams by a team of observers, leading to cumbersome and expensive processes, often resulting in missed important events and a less-than-optimal viewer experience due to the complexity of coordinating multiple feeds in real time.

Innovation Solution

A system that dynamically selects and arranges content for broadcast by predicting events of interest using machine learning and deep neural networks, analyzing video, audio, and player input data to assign priority values to streams, allowing for automatic or semi-automatic selection and arrangement to maximize the inclusion of interesting events.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual selection and arrangement of streams by a team of observers is used, then content selection can be made with human judgment, but the process becomes cumbersome and expensive

Engineering Contradiction:

Improvecontent selection accuracyVSAvoidobserver team complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces the mechanical system of manual human observation and selection with an automated computer-based system that uses machine learning models and event detection algorithms to analyze multiple streams and determine which events to broadcast, eliminating the need for large teams of human observers

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service by allowing the broadcast system to automatically select and arrange streams based on detected events and predicted viewer interest, without requiring continuous human intervention or coordination

Inventive Principle:

Principle #25Self-service

2Adaptability or versatility

If manual coordination of multiple feeds in real time is performed, then flexible content selection is possible, but important events may be missed due to coordination complexity

Engineering Contradiction:

Improvecontent selection flexibilityVSAvoidevent capture reliability

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system implements feedback by continuously monitoring multiple streams, detecting events in real-time, and dynamically adjusting which streams are selected for broadcast based on the detected events and their predicted interest level, ensuring important events are captured

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system performs preliminary action by pre-defining multiple candidate streams and having the event detection system ready to immediately select and switch to relevant streams when events are detected, ensuring no important moments are missed

Inventive Principle:

Principle #10Preliminary action

3Area of stationary object

If equally divided spatial arrangement of streams is used, then all streams are visible, but viewer engagement with important events is reduced

Engineering Contradiction:

Improvestream visibility areaVSAvoidviewer engagement

Core Design Contradiction:

Area of stationary objectVSProductivity

Solution Approach 1:

The patent applies local quality by allocating different display areas to different streams based on their importance and predicted viewer interest, with more important events receiving larger or more prominent display areas while less important events receive smaller areas

Inventive Principle:

Principle #3Local quality

4Ease of operation

If customized view with different selections and arrangements is implemented, then viewer experience improves, but manual curation effort increases

Engineering Contradiction:

Improveviewer experience qualityVSAvoidcuration time

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The patent replaces manual curation with automated systems that use machine learning models to predict viewer interest and automatically arrange streams in customized views, eliminating the time-consuming manual curation process while maintaining high viewer experience quality

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20230149819A1Dynamically selecting from multiple streams for presentation by predicting events using artificial intelligence

Publication Date: 2023.05.18 NVIDIA CORP

AI summary

Approaches presented herein provide for dynamic selection and presentation of content for a broadcast or transmission, such as to provide content that is most likely to be of interest to a viewer. This can be accomplished, at least in part, by predicting occurrences of events of interest in one or more sources of content, such as one or more input media streams. The occurrences can be predicted using various sources of content or data, as may include non-game video, player input, and player-agnostic game data. Various input data streams can be analyzed to predict the probability of one or more events occurring over a future period of time, and these probabilities can be used to assign priority values to the various streams. These priority values can be used to determine which streams to include in a broadcast, as well as how to arrange or feature those streams in the broadcast.