Hyundai Mobis announced on the 17th that its generative artificial intelligence (AI) was selected as an excellent paper at the Computer Vision and Pattern Recognition (CVPR) conference held in Tennessee, USA.
CVPR is known as one of the top conferences in the world in fields such as image processing, object recognition, and deep learning. A representative from Hyundai Mobis noted, "It is significant that an AI paper has been selected at a conference known for its strict selection criteria."
The generative AI unveiled by Hyundai Mobis can analyze text and image information and precisely transform specific parts. A typical example is the ability to change specific conditions, such as weather, time of day, and road shape, based on road footage taken during the day.
Hyundai Mobis plans to apply this generative AI to analyze and search document text and image information across the company. It will also develop a proprietary platform capable of providing information services, and plans to implement it in research and development, production technology, and IT systems starting in August. This platform is intended for knowledge search, task automation, and image analysis.
Previously, Hyundai Mobis collaborated with Pohang University of Science and Technology to develop natural language processing AI technology specialized for equipment control in manufacturing sites. In traditional manufacturing sites, workers entered complex commands manually for equipment control, but when generative AI is utilized, workers can input only basic commands, and the AI will automatically generate control codes to ensure stable operation of the equipment.
This AI model was recently selected as a core paper at the Association for Computational Linguistics (ACL), an international conference in the field of natural language and computer language processing.
Hyundai Mobis has introduced a safety detection AI model at its business sites that detects workers entering hazardous areas and prevents robot collisions. Additionally, an 'acoustic AI system' that recognizes sounds to assess product quality accuracy has also been developed and is being applied in production sites.