ASICLAND Registers 'Edge AI Facial Recognition Device' Patent 

ASICLAND (CEO Jong-min LEE) announced on the 8th that they have registered a patent for an 'Edge AI-based facial recognition device.'

The device uses a 3D camera to generate a depth map image corresponding to the user's face and a 2D camera to create a 2D facial image, enabling the recognition of the user's face.

Facial recognition technology involves detecting facial images within still images or videos captured through cameras. After detecting the facial images, pre-processing steps such as face alignment, brightness correction, and image normalization are applied. Subsequently, features are extracted from the pre-processed facial image to identify the specific face.

The issue with existing technology lies in its inability to improve the fundamental accuracy and reliability of facial recognition, as it primarily verifies the three-dimensional aspects of a user's face through distance sensors or depth maps.

In this regard, Jong-min Lee, CEO of ASICLAND, stated, “The first neural network model is trained to calculate the similarity between depth map images corresponding to the user's face and pre-registered depth map images. Simultaneously, the second neural network model, also trained, computes the similarity between 2D facial images corresponding to the user's face and pre-registered 2D facial images. These approaches collectively enable diverse facial recognition techniques,” and added, “The final facial recognition result is obtained by considering multiple results together, thus preventing spoofing attacks using photos or videos and enhancing the accuracy and reliability of facial recognition.”

The company has also registered a patent for 'On-Chip System Data Management Device and Method.' This technology is expected to contribute to power savings in on-chip systems and can be applied to various devices that use limited power, such as mobile devices, sensor nodes, and robots.

Source: ETNEWS (link)