A research team from Konkuk University has developed an artificial intelligence algorithm that can create precise 3D human models using minimal input images, enhancing real-time processing speed.
Key points
•Konkuk University Kim Wonjun Professor team develops new AI algorithm
•Real-time processing for precise 3D modeling with limited data
•Expected application in VR/AR and digital human content
Professor Wonjun Kim and his team at Konkuk University have devised an AI algorithm capable of generating detailed 3D representations of humans from limited input data. The algorithm leverages the geometric structure and texture complexity of bodies and clothing to restore fine details such as faces or fingers.
Driven by the growing importance of precise 3D human modeling, especially in environments where acquiring extensive video footage is challenging, the team aimed to find ways to accurately represent people with limited input data. This breakthrough allows for detailed 3D models even when using multiple cameras to gather a large volume of images is difficult.
Supported by the Ministry of Science and ICT, this research will be presented at NeurIPS 2026. The algorithm is expected to find applications in virtual and augmented reality as well as digital human content creation.
