Dart Pin Hit Location Identification Using Deep Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image analysis methods for identifying hit locations of dart pins in dart games face challenges such as noise in various installation environments, difficulty in securing a large enough dataset, and complexity in processing hit locations of multiple dart pins.

Innovation Solution

A method and system for generating a training dataset to improve the accuracy of identifying hit locations of dart pins, utilizing a dart game apparatus equipped with cameras to capture images from multiple angles, and employing a deep learning model to process these images and generate hit location information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If rule-based algorithms are used for image analysis, then the system is simple to implement, but accuracy deteriorates due to noise in various installation environments

Engineering Contradiction:
Improveease of implementationVSAvoidhit location identification accuracy
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent replaces rule-based algorithms with a deep learning-based image analysis system. The deep learning model (neural network) automatically learns optimal feature extraction and hit location determination from training data, substituting the manual rule-based approach with an adaptive intelligent system that maintains high accuracy across varying installation environments and lighting conditions.

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

2Measurement precision

If deep learning is used for image analysis, then accuracy improves, but the difficulty of securing a large enough dataset increases

Engineering Contradiction:
Improvehit location identification accuracyVSAvoiddataset size
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent employs data augmentation techniques to create synthetic training data by copying and transforming existing images. The system generates additional training samples through geometric transformations (rotation, flipping, scaling), color adjustments, and noise addition, effectively multiplying the available training data without requiring proportional increases in physical experimentation or data collection.

Inventive Principle:
Principle #26Copying

3Measurement precision

If deep learning is used to process hit locations of multiple dart pins, then identification accuracy improves, but algorithm complexity increases

Engineering Contradiction:
Improvehit location identification accuracyVSAvoidalgorithm complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements a segmentation approach where the deep learning model first identifies and segments individual dart pins in the image, then processes each segmented region separately to determine hit locations. This divides the complex task of analyzing multiple overlapping pins into simpler sub-tasks, making the algorithm more manageable while maintaining accuracy in identifying multiple hit locations simultaneously.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4325440B1Method, computer program, and device for identifying hit location of dart pin
Publication Date: 2025.03.05 PHOENIXDARTS CO LTD
  • EP4325440B1 patent drawingFigure 1
  • EP4325440B1 patent drawingFigure 2
  • EP4325440B1 patent drawingFigure 3

AI summary

Disclosed is a computer program stored in a computer-readable storage medium, the computer program providing a dart pin hit location identification method when being executed by one or more control units of a dart game device, the method including: obtaining at least two dart target photographing images in response to a dart pin throw, in which the dart target photographing image includes an image of a dart target and at least one dart pin hitting the dart target; and generating, by a dart pin location identification network model, hit location information of the at least one dart pin based on the at least two dart target photographing images.