Object Recognition Liveness Detection Neural Network Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current face verification technologies face challenges in accurately distinguishing between real and fake images, particularly in liveness detection, which is crucial for secure biometric authentication.

Innovation Solution

A processor-implemented object recognition method using a deep neural network that includes multiple layers with connection weights, capable of performing object recognition and liveness detection by training neural networks for feature extraction and verification, allowing for the differentiation between real and fake images through a series of computational processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional face verification methods are used, then the authentication process is simple, but the accuracy in distinguishing real and fake images is insufficient

Engineering Contradiction:
Improveliveness detection accuracyVSAvoidverification system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The verification process is divided into multiple stages: initial face verification using a first neural network, followed by liveness detection using a second neural network. This segmentation allows each network to specialize in specific tasks, improving overall accuracy while managing complexity through modular architecture.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a temporal dimension by capturing multiple images at different time points and comparing them. This multi-temporal approach adds a new dimension to the verification process, enabling detection of liveness characteristics that single-frame methods cannot detect, thereby improving accuracy without proportionally increasing complexity.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If multiple neural networks are used for verification and liveness detection, then the detection accuracy improves, but the processing time increases

Engineering Contradiction:
Improveface verification accuracyVSAvoidauthentication processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs face verification using the first neural network before initiating the more computationally intensive liveness detection process. This preliminary verification filters out clearly non-matching cases early, avoiding unnecessary processing time for the second network while maintaining high overall accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent applies liveness detection selectively rather than to all verification cases. By using the second neural network only when needed (e.g., when the first verification is borderline or suspicious), the system achieves high accuracy where necessary while minimizing overall processing time through partial application of the more intensive detection method.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3499412B1Object recognition and liveness detection and apparatus
Publication Date: 2023.02.01 SAMSUNG ELECTRONICS CO LTD
  • EP3499412B1 patent drawingFigure 1
  • EP3499412B1 patent drawingFigure 2
  • EP3499412B1 patent drawingFigure 3

AI summary

An object recognition apparatus and method are provided. The apparatus comprises a processor configured to verify a target image using an object model and based on reference intermediate data extracted by a partial layer of the object model as used in an object recognition of an input image, in response to a failure of a verification of the input image after a success of the object recognition of the input image, and perform an additional verification of the target image in response to the target image being verified in the verifying of the target image.