Game Scene Recognition Using Object Models From Drawing Instructions

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing scene recognition methods in computer graphics technologies consume a large quantity of computing resources, leading to high computing costs and performance issues in terminals, especially when determining complex game scenes in large-scale 3D games.

Innovation Solution

A scene recognition method that determines the target scene based on a target object model derived from vertex information of drawing instructions, reducing the need for extensive computing resources by using a correspondence between object models and scenes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If image recognition technology is used to match current image frames with training set pictures, then scene recognition accuracy is improved, but computing resource consumption increases significantly

Engineering Contradiction:
Improvescene recognition accuracyVSAvoidcomputing resource consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent creates simplified 2D silhouette copies of 3D game objects by projecting vertex information onto a plane. These silhouette images serve as lightweight representations that can be quickly compared against a pre-built training set, avoiding the need for complex full-image recognition while maintaining scene identification accuracy.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent extracts only the essential silhouette features from complex 3D game scenes by projecting vertex data onto a 2D plane. This extraction process isolates the key structural information needed for scene recognition while discarding unnecessary visual details, thereby reducing computing resource requirements.

Inventive Principle:
Principle #2Taking out (Extraction)

2Loss of information

If complex 3D game scenes are rendered and analyzed in real-time, then scene recognition completeness is improved, but processing speed decreases

Engineering Contradiction:
Improvescene recognition completenessVSAvoidprocessing speed
Core Design Contradiction:
Loss of informationVSSpeed

Solution Approach 1:

The patent generates simplified 2D silhouette copies from 3D vertex data, creating a compressed representation that retains essential scene structure information. This copying approach maintains scene recognition completeness while enabling faster processing compared to analyzing full-resolution 3D rendered images.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent performs preliminary projection of vertex information onto a 2D plane before scene recognition. This preliminary action transforms complex 3D data into a simplified format in advance, preparing it for rapid comparison with the training set and improving overall processing speed.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If vertex information from drawing instructions is processed to create object models, then scene recognition efficiency is improved, but system complexity increases

Engineering Contradiction:
Improvescene recognition efficiencyVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent replaces traditional image-based recognition mechanisms with a vertex information processing system. By using drawing instruction data and vertex coordinates directly, the system substitutes complex image analysis with simpler geometric data processing, improving efficiency while managing system complexity through standardized projection algorithms.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP3905204B1Scene recognition method and apparatus, terminal, and storage medium
Publication Date: 2026.03.25 HUAWEI TECH CO LTD
  • EP3905204B1 patent drawingFigure 1~2
  • EP3905204B1 patent drawingFigure 3
  • EP3905204B1 patent drawingFigure 4

AI summary

This application discloses a scene recognition method and apparatus, a terminal, a storage medium, and the like, which may be applied to game scene recognition. The method includes: obtaining one or more drawing instructions, and determining a to-be-drawn target object based on a description parameter carried in the one or more drawing instructions. Because there is a particular relationship between a scene and an object, a corresponding target scene may be determined based on the target object. Compared with a related technology in which a large quantity of computing resources are consumed by using an image recognition technology, the scene recognition method provided in the embodiments of this application effectively reduces system resources.