Advancements and Breakthroughs with FACE TRUST Part 2.



FACE TRUST: A New Era in Face Recognition AI


FACE TRUST’s vision transcends the current technological constraints of facial recognition AI, with the aim of constructing a society wherein everyone is secure and protected. In simpler terms, facial recognition AI will not be limited to mere recognition and analysis of human faces; it will broaden its technological applicability to reshape everyday life and society. Our objective is to enhance societal well-being by comprehensively analyzing and validating human faces, surrounding contexts, and environments, all grounded in more dependable data.

To attain this vision, Alchera continuously enhances performance through data diversification, ongoing learning, and model optimization. We also introduce various methodologies across multiple industries, ensuring our position at the vanguard of technology. In this edition, we will delve into Alchera's advancements and breakthroughs in facial recognition AI, heralding the advent of a new era for this technology.


How Has Face Recognition AI Advanced?


During the early stages of facial recognition AI, which marked its inception as version 1.0, a manual approach was employed. This approach involved individuals manually specifying and mapping facial shapes and features. One prominent technique during this period was *Template Matching, which entailed manually measuring point-to-point ratios and feature points to compare pixel values. While it required human intervention and lacked accuracy, it served as a pivotal starting point, laying the foundation for facial recognition technology.
*Template Matching: A method used to locate specific patterns or templates within an image by comparing the template to the image to identify matching sections.



With the advent of machine learning (2.0), automated facial recognition AI utilizing machine learning and computer vision technology emerged. Algorithms were developed using *Multi-pie Dataset to train on diverse data and discern patterns, allowing for the recognition of facial features, patterns, and relationships. However, there were limitations due to the scarcity of available data and a limited number of IDs at the time, resulting in some challenges regarding performance generalization and reliability of results.
*Multi-pie Dataset: A dataset consisting of facial images captured under various conditions, used for the development of facial recognition algorithms.



The emergence of deep learning (3.0) brought about a revolutionary change. Deep learning-based facial recognition algorithms, leveraging *Convolution Neural Networks (CNNs), allowed for more precise feature extraction and classification. Training on extensive real-world facial datasets significantly improved accuracy. This era marked a period of substantial performance enhancement, addressing the limitations in performance generalization and reducing errors encountered during machine learning. Moreover, this era introduced model benchmarking and performance evaluations under conditions closely resembling real-world environments, expanding facial recognition AI's application and utility in various domains.
*Convolution Neural Network (CNN): A deep learning algorithm specialized in extracting features and classifying images.



Additionally, advancements in graphics processing units and computer hardware significantly improved data processing speeds, enabling the efficient handling of large datasets. As a result, facial recognition AI achieved high-performance generalization (4.0), ensuring accurate and fast recognition even with only 40% of the face exposed. This period marked a shift from evaluating the technology's performance to assessing the value of products and quality. Facial recognition AI firmly established itself as a technology that could be tailored to meet user demands.


FACE TRUST: The Current State of Face Recognition AI


How has facial recognition AI advanced? Alchera's FACE TRUST represents the present state of facial recognition AI as a technology that everyone can rely on (5.0). It is expanding its applications across various industries by focusing on developing tailored solutions that address user and industry-specific needs, all built on a foundation of trusted technology. To achieve this, Alchera utilizes pre-trained models from major global IT companies like Google and Facebook, enhancing facial recognition AI models by adding or modifying learning layers.

In particular, facial recognition AI plays a pivotal role in access control and the finance sector, wherein it strengthens identity verification and user authentication. Expectations are high for its continued expansion in the areas of identity verification, personal data protection, and security management by enhancing security levels for facilities and authentication methods, preventing fraudulent activities and unauthorized access.


FACE TRUST 6.0: Leading the New Era of Face Recognition AI


Alchera does not settle for the present but is actively preparing for a new era in facial recognition AI through FACE TRUST 6.0. It aims to maximize the value of facial biometric information to further solidify its current applications. Thus far, facial recognition AI has significantly contributed to societal benefits by recognizing and assessing individuals, verifying their identities, and distinguishing between authorized and unauthorized individuals, serving as a core technology in authentication and security systems.

However, in the future, it will extend its value as a technology (6.0) that supports risk prevention and early response through the use of extensive learning data, thereby bringing about a paradigm shift in various industries. By embracing context learning, which involves recognizing not only a person's face, but also comprehensive information about the subject and their surrounding environment and circumstances, Alchera aims to enhance individuals' psychological well-being through technological advancements.


Surrounding Image Analysis:

When recognizing a subject, it analyzes the surrounding elements such as background, other objects, clothing, and hair.. This allows for more precise identification and an understanding of the interaction between the subject and their environment.

Dynamic Change Analysis:

Utilizing dynamic data, it analyzes the subject's movements, behavioral characteristics, gait, and more. It tracks the subject's previous movements and can predict their next actions and movements.

Social Context Analysis:

Considering the subject's facial expressions, posture, and perceived location, it assesses the subject's situation and emotions. This enables the understanding of the subject's actions and identity within a social context.


Face Recognition AI: Realizing Safety-Assured Technology


To ensure a safer life through a "guaranteed technology" for security, it is essential to surpass the current limitations of facial recognition AI with more advanced technological capabilities.

Alchera is committed to continuous research and efforts aimed at improving facial recognition performance, even for recognition coverage of less than 5%, and detecting changes in appearance spanning over 50 years. Moreover, we seek to integrate stable facial recognition AI technology into all camera-equipped devices, not limited to facial recognition terminal types.

Through this endeavor, we aspire to support not only personal identity verification, but also the analysis of movement patterns, prediction of future actions, accident prevention, and establishment of a swift response system. In the next installment, we will explore potential innovations in the Mobility, Healthcare, and ESG fields and delve into the roadmap outlined by FACE TRUST.


[Related Contetns]
[Article] FACE TRUST, A Technology Everyone Can Trust