ML Event Detection in Electronic Game Video Streams

The system divides the complex task of event detection into multiple specialized application modules, each dedicated to a specific game engine (e.g., Unity module, Unreal Engine module). Each module contains game engine-specific data and logic tailored to its target engine, allowing the system to handle different engines independently without increasing overall complexity. This segmentation enables accurate event detection for each engine while maintaining manageable system architecture.

The patent introduces an intermediary layer between the video information input and the event detection logic. This intermediary consists of the application modules that act as mediators, translating raw video data into meaningful events specific to each game engine. The intermediary layer abstracts the complexity of different game engines, allowing the core system to remain simple while achieving high detection accuracy through engine-specific processing.

The system applies local quality by making each application module specialized for its specific game engine, rather than using a uniform approach for all engines. Each module contains game engine-specific data that captures the unique characteristics (art styles, textures, animations) of its target engine. This localized specialization enables accurate event identification for each engine while keeping the overall system structure simple and manageable.

The patent utilizes parameter changes by adapting the detection parameters and data structures within each application module to match the specific characteristics of its target game engine. For example, Unity-specific modules use parameters optimized for Unity's rendering pipeline, while Unreal Engine modules use parameters suited to Unreal's systems. This parameter adaptation allows accurate event detection across diverse engines without requiring a single complex universal system.