Dr Ir Ike Fibriani ST MT presenting her dissertation at the Doctoral Promotion Open Session at the Department of Electrical Engineering ITS

ITS Campus, ITS News — Institut Teknologi Sepuluh Nopember (ITS) never stops producing outstanding doctoral graduates. Becoming the first in the world, a new doctor who graduated from the ITS Department of Electrical Engineering (DEE) Dr Ir Ike Fibriani ST MT created an innovation that combines kinship tests with microexpressions.

At the Open Session of New Doctoral Promotion of DEE ITS, Tuesday (2/25), Ike explained that through this breakthrough, the kinship relationship between parents and their children can be detected through microexpressions. “Through microexpressions, namely spontaneous human expressions, it can determine the relationship of one descendant,” she explained.

This Jember-born woman revealed how this research unites the correlation between kinship and microexpressions. Departing from the lack of research related to both of these things, her dissertation entitled Kinship Detection Using Hybrid Vision Transformer Mamba Based on Microexpressions is expected to be used as an innovation in the world of face-recognition. This breakthrough is in accordance with point 9 of the Sustainable Development Goals (SDG), namely Industry, Innovation, and Infrastructure.

The process of how the kinship detection system works using the Vision Transformer and Mamba methods initiated by Dr Ir Ike Fibriani ST MT, a doctoral graduate of ITS

Ike introduced this system as ViTMa, which originated from a combination of vision transformers and mamba, which is the method used to optimize the system. In the process, Vision Transformer (ViT) works by understanding the face parts and capturing unique patterns and microexpression that form spontaneously on the face.

Furthermore, according to Ike, ViT plays a role in extracting features in facial expressions between children and parents by dividing pictures into 16 patches as inputs. This input works using a self-attention mechanism that can make it possible to read micro details or subtle changes in expressions that are important for kinship recognition.

Ir Ike Fibriani ST MT explaining about her dissertation entitled Kinship Detection Using Hybrid Vision Transformer Mamba Based on Microexpressions

The feature extraction results are then optimized using the Mamba method to increase the accuracy of the results and determine the kinship relationship. “After Mamba’s optimization, there will be a result that will determine whether or not a kinship relationship exists,” said Ike. The Jember University lecturer also explained that the results of the ViTMa system has a high level of accuracy, as high as 85.29 percent, outperforming traditional methods such as CNN and VGGFace.

In the future, the ViTMa system can also be used for various fields, such as forensics, psychology, document verification, and/or safety and surveillance systems. With the extent of benefits and fields that can be achieved, Ike believes that ViTMa has a great potential to be developed and continues to be optimized repeatedly. “I hope this system can be used for the wider community,” he ended. (ITS PUBLIC RELATIONS)

Reporter: Nailah Rifdah Zakiyah