KAIST · 컴퓨터과학
홍주리 교수의 연구실은 의료 영상 분할, 신경망 기반 얼굴 랜드마크 검출, 그리고 신뢰성 있는 딥러닝 모델 설계에 중점을 두고 있습니다. 특히 의료 영상에서의 경계 불확실성과 모델의 불확실성 추정 문제를 해결하기 위한 구조 보존형 분할 프레임워크와, 기하학적 제약을 통합한 생성적 적대적 네트워크를 활용한 정밀한 랜드마크 검출 기법을 개발하고 있습니다. 또한, 적대적 공격에 강건한 특징 표현 학습 및 효율적인 앙상블 기반 세분화 기법을 통해 딥러닝의 신뢰성과 정확도를 향상시키는 데 기여하고 있습니다.
표시된 성과는 수집된 데이터 기준으로 산출되며, 일부 차이가 있을 수 있습니다.
In this paper, we propose a novel image segmentation method to tackle two critical problems of medical image, which are (i) ambiguity of structure boundary in the medical image domain and (ii) uncertainty of the segmented region without specialized domain knowledge. To solve those two problems in automatic medical segmentation, we propose a novel structure boundary preserving segmentation framework. To this end, the boundary key point selection algorithm is proposed. In the proposed algorithm, t
Two-dimensional simulations of photon acceleration by using a laser wake field are presented with a fully electromagnetic and relativistic particle-in-cell code. The frequency increase of about 10% is observed, which is saturated mainly by diffraction and dispersion of the laser pulse. Images of electron density and laser field profiles are presented.
Facial landmark detection plays an important role in face analysis tasks. Moreover, it is used as a prerequisite in many facial related applications, the simplicity, as well as effectiveness, is essential in the facial landmark detection. In this paper, we propose an effective facial landmark detection network and an associated learning framework with the geometric prior-generative adversarial network. The geometric prior-generative adversarial network consists of one generator and two discrimin
Interests on the fabrication of microfluidic devices have increased in the fields of micro total analysis system (μ-TAS) and MEMS (Microelectromechanical systems) due to their chemical inertness and high thermal stability. The thermal characterization of the SiCN preceramic polymer, polyvinylsilazane, showed that the cured polymer has ceramic properties at heat treatment temperature of 600 oC or above. In the characterization of the mechanical properties, the characteristic values of the elastic
