Kookmin University team develops technology to optimize 3D object detection models for self-driving
A Kookmin University research team has developed SharedKD, a technology that efficiently compresses 3D object detection models for autonomous driving, in collaboration with Hyundai Motor. The university said Tuesday that the findings were presented at the 2026 Design Automation Conference, a prestigious international conference in the field of design automation, held in Long Beach, California,…
A team of researchers from Kookmin University, in partnership with Hyundai Motor, has created SharedKD, a technology aimed at optimizing 3D object detection models specifically for self-driving cars. The innovative technique was presented at the 2026 Design Automation Conference, an esteemed international event held in Long Beach, California, from July 26-29.
Professor Kim Jang-ho, from the university's College of Computer Science, spearheaded the research team, with two master's students, Cho Hyun-joon and An Sang-ho, from the Graduate School of AI and SW, playing significant roles in the project.
What sets SharedKD apart from traditional knowledge distillation methods is its unique approach. Rather than employing separate teacher and student models, SharedKD utilizes the entire network of a single 3D object detection model as the teacher and a subnetwork generated through pruning as the student. During the training process, SharedKD dynamically chooses crucial structures based on gradients, allowing for efficient compression of the 3D object detection models.
Written by urgent.news from The Korea Times's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
