Embedded Files
EV Lab | Designing Efficient Intelligence Through Deep Understanding
EV Lab은 데이터의 본질과 도메인 특성을 고려하여 효율적이고 지능적인 AI 모델 개발을 추구합니다.
EV Lab은 데이터의 본질과 도메인 특성을 고려하여 효율적이고 지능적인 AI 모델 개발을 추구합니다.
단순히 모델의 규모를 확장하는 방식에서 벗어나,
영상처리·컴퓨터 비전·의료 영상·생체 신호 등 각 도메인의 prior 정보를 모델 설계에 적극적으로 반영합니다.
단순히 모델의 규모를 확장하는 방식에서 벗어나,
영상처리·컴퓨터 비전·의료 영상·생체 신호 등 각 도메인의 prior 정보를 모델 설계에 적극적으로 반영합니다.
이를 통해 계산 효율성과 성능을 동시에 향상시키고,
복잡한 문제를 가장 합리적이고 최적화된 방식으로 해결하는 AI 시스템을 구축하고자 합니다.
이를 통해 계산 효율성과 성능을 동시에 향상시키고,
복잡한 문제를 가장 합리적이고 최적화된 방식으로 해결하는 AI 시스템을 구축하고자 합니다.
EV Lab pursues the development of efficient and intelligent AI models
based on a deep understanding of data essence and domain characteristics.
EV Lab pursues the development of efficient and intelligent AI models
based on a deep understanding of data essence and domain characteristics.
Rather than simply scaling model size, we actively integrate prior knowledge from domains like
image processing, computer vision, medical imaging, and bio-signals into our architectures.
Rather than simply scaling model size, we actively integrate prior knowledge from domains like
image processing, computer vision, medical imaging, and bio-signals into our architectures.
Through this, we aim to simultaneously improve computational efficiency and performance,
building AI systems that solve complex problems in the most rational and optimized manner.
Through this, we aim to simultaneously improve computational efficiency and performance,
building AI systems that solve complex problems in the most rational and optimized manner.
Research Topics
Research Topics
Computer Vision / Image Processing
Computer Vision / Image Processing
Medical Imaging / Bio-Signal Procesing
Medical Imaging / Bio-Signal Procesing
On-device AI
On-device AI
Vision-Language-Action
Vision-Language-Action
NEWS
NEWS
(01/2026) Our paper on efficient coarse-to-fine approach in deblurring has reached 1,000 citations on Google Scholar.
(01/2026) Junghyun, Jungwoo, and Eunkang have joined EV Lab as undergraduate research interns.
(12/2025) Jaeyoung will join ETRI as a Winter Research Trainee to work on VLA models at the Robot Foundation Model Research Lab.
(08/2025) Our team has been selected for the '서울 바이오-AI 융합 지원 사업' (with Pukyong National University, Gachon University & MEDAI).
(06/2025) Our work on efficient federated learning with weight standardization and distribution-aware quantization has been accepted to ICCV 2025.
(05/2025) Our team has been selected for the '2025년 서울바이오허브 병원기반 의료데이터 활용 지원 사업' (with Seoul St. Mary’s Hospital & MEDAI).
(05/2025) Our team won 2nd place in the Hanwha Systems AI Challenge: Infrared Sensor Instance Segmentation, and was invited to present at the Thermal Infrared in Robotics Workshop at ICRA 2025.
(05/2025) Our work on depth estimation using 2D Hilbert curves has been accepted to ICML 2025.
(03/2025) Suheum, Jaeyoung, and Seojin have joined EV Lab as undergraduate research interns.
(02/2025) EV Lab was established at Konkuk University, School of Computer Science & Engineering.
Page updated
Google Sites
Report abuse