Sunghwan Hong

Donghoon Ahn

Junwan Kim

The Computer Vision Laboratory (CVLAB), led by Professor Seungryong Kim, is part of the Kim Jaechul Graduate School of AI at KAIST. Research focuses on computer vision, multi-modal learning, and visual computing. Applications include autonomous driving, intelligent surveillance, humanoid robots, AR/VR, Metaverse, and healthcare.

KAIST 김재철 AI 대학원 소속 CVLAB (지도교수: 김승룡). 주요 연구 분야는 컴퓨터 비전, 멀티모달 학습, 시각 컴퓨팅이며, 자율 주행, 지능형 감시, 휴머노이드 로봇, AR/VR, 메타버스, 헬스케어 등에 적용된다.

Vid-CamEdit: Video Camera Trajectory Editing with Generative Rendering from Estimated Geometry

Few-Shot Efficient Training of Diffusion Models for Industrial Anomaly Synthesis

V-Warper: Appearance-Consistent Video Diffusion Personalization via Value Warping

Deep Forcing: Training-Free Long Video Generation with Deep Sink and Participative Compression

UniT: Unified Diffusion Transformer for High-fidelity Text-Aware Image Restoration

RobustVGGT: Emergent Outlier View Rejection in Visual Geometry Grounded Transformers

C3G: Learning Compact 3D Representations with 2K Gaussians

CAMEO: Correspondence-Attention Alignment for Multi-View Diffusion Models

MV-TAP: Tracking Any Point in Multi-View Videos

Learning to Track Any Points from Real-World Motion

Effective Test-Time Scaling of Discrete Diffusion through Iterative Refinement

Lookahead Unmasking Elicits Accurate Decoding in Diffusion Language Models

Lookahead Anchoring: Preserving Character Identity in Audio-Driven Human Animation

Exploring Conditions for Diffusion models in Robotic Control

3D Scene Prompting for Scene-Consistent Camera-Controllable Video Generation

MATRIX: Mask Track Alignment for Interaction-aware Video Generation

TAG: Tangential Amplifying Guidance for Hallucination-Resistant Diffusion Sampling

SpikeMatch: Semi-Supervised Learning with Temporal Dynamics of Spiking Neural Networks

VIRAL: Visual Representation Alignment for Multimodal Large Language Models

PLOT: Pseudo-Labeling via Object Tracking for Scalable Monocular 3D Object Detection

MoAI: Aligned Novel View Image and Geometry Synthesis via Cross-modal Attention Instillation

Text-Aware Image Restoration with Diffusion Models

InterRVOS: Interaction-aware Referring Video Object Segmentation

URECA: Unique Region Caption Anything

A Noise is Worth Diffusion Guidance

Referring Video Object Segmentation via Language-aligned Track Selection

Emergent Temporal Correspondences from Video Diffusion Transformers

Seg4Diff: Unveiling Open-Vocabulary Semantic Segmentation in Text-to-Image Diffusion Transformers

D2USt3R: Enhancing 3D Reconstruction with 4D Point Maps for Dynamic Scenes

Where and How to Perturb: On the Design of Perturbation Guidance in Diffusion and Flow Models

Active Test-time Vision-Language Navigation

First Attentions Last: Better Exploiting First Attentions for Efficient Parallel Training

Leveraging Intrinsic Components for Few-Shot Neural Radiance Fields in Unconstrained Illumination

Self-Evolving Neural Radiance Fields

DiffPort: Adapting Pre-trained Diffusion Models for Generalizable Robot Manipulation

Talk3D: High-Fidelity Talking Portrait Synthesis via Personalized 3D Generative Prior

AM-Adapter: Appearance Matching Adapter for Exemplar-based Semantic Image Synthesis

S4M: Boosting Semi-Supervised Instance Segmentation with SAM

ReMoTE:  A Benchmark for Object Motion Transfer

PF3plat : Pose-Free Feed-Forward 3D Gaussian Splatting

ReSIT: A more Realistic Synthetic Driving Dataset for Multi-Domain Image-to-Image Translation

Visual Persona: Foundation Model for Full-Body Human Customization

Seurat: From Moving Points to Depth

Cross-View Completion Models are Zero-shot Correspondence Estimators

Exploring Temporally-Aware Features for Point Tracking

Identity-preserving Distillation Sampling by Fixed-Point Iterator

ControlFace: Harnessing Facial Parametric Control for Face Rigging

Mixture of Submodule for Domain Adaptive Person Search

DiffFace: Diffusion-based Face Swapping with Facial Guidance

Semantically Complex Audio to Video Generation with Audio Source Separation

Multi-Granularity Video Object Segmentation

Robustnet: Improving domain generalization in urban-scene segmentation via instance selective whitening

Context-aware emotion recognition networks

Cost aggregation with 4d convolutional swin transformer for few-shot segmentation

Cat-seg: Cost aggregation for open-vocabulary semantic segmentation

Fcss: Fully convolutional self-similarity for dense semantic correspondence

Let 2d diffusion model know 3d-consistency for robust text-to-3d generation

Improving sample quality of diffusion models using self-attention guidance

Unified multi-spectral pedestrian detection based on probabilistic fusion networks

Cats: Cost aggregation transformers for visual correspondence

Partmix: Regularization strategy to learn part discovery for visible-infrared person re-identification

Unified depth prediction and intrinsic image decomposition from a single image via joint convolutional neural fields

DASC: Dense adaptive self-correlation descriptor for multi-modal and multi-spectral correspondence

Recurrent Transformer Networks for Semantic Correspondence

Diffface: Diffusion-based face swapping with facial guidance

Multi-modal recurrent attention networks for facial expression recognition

Multi-spectral pedestrian detection based on accumulated object proposal with fully convolutional networks

Dunit: Detection-based unsupervised image-to-image translation

High-precision depth estimation with the 3d lidar and stereo fusion

High-precision depth estimation using uncalibrated LiDAR and stereo fusion

3d gan inversion with pose optimization

Kwanghoon Sohn

Dongbo Min

Seokju Cho

Sunok Kim

Junyoung Seo

Jisu Nam

Sangryul Jeon

Stephen Lin

Jiyoung Lee

Heeseong Shin

Bumsub Ham

Susung Hong

Wooseok Jang

Gyuseong Lee

Jaewoo Jung

Min-Seop Kwak

Seungchul Ryu

Honggyu An

Jaehoon Ko

Seungryong Kim

B.S. in Electrical and Electronics Engineering

Ph.D. in Electrical and Electronics Engineering

Associate Professor

Sangwon Jung

Postdoc Researcher

Yeji Choi

Seok Yeong Lee

Ph.D Student

M.S./Ph.D Integrated Student

Minseop Kwak

Ph.D. Student

Jin Hyeon Kim

Sangbeom Lim

Jini Yang

Inès Hyeonsu Kim

M.S. Student

Jisang Han

Siyoon Jin

Soowon Son

Heeji Yoon

Woojeong Jin

Seongchan Kim

Jaewon Min

Chaehyun Kim

Jiwon Kang

Joungbin Lee

Jiyoung Kim

Jung Yi

Hyunkoo Lee

Minkyung Kwon

Eunbeen Hong

Jinhyeok Choi

Junghyun Park

Jaeho Lee

Jinhyuk Jang

Jahyeok Koo

Paul Hyunbin Cho

Mungyeom Kim

Sanghyun Lee

Seonghu Jeon

Undergraduate Student

Junghyun Hwang

Minjae Kim

Dahyun Chung

Hyunwook Choi

Taehwan Kim

Jaeeun Lee

Seohyeon Park

Jiho Park

Jaewoo Yu

Kyungjin Oh

Junsu Kim

Jaeyeong Kim

Hyunah Ko

Sangoh Kim

Youngjin Shin

Hyeonseo Yu

Jinyoung Park

CVLAB

교수님 소개

김승룡

연구실 소개

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

Vid-CamEdit: Video Camera Trajectory Editing with Generative Rendering from Estimated Geometry

Few-Shot Efficient Training of Diffusion Models for Industrial Anomaly Synthesis

Emergent Temporal Correspondences from Video Diffusion Transformers

연구실 보유 특허

대학별 입시끈

우리 연구실 모집

교수님 소개

김승룡

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

Vid-CamEdit: Video Camera Trajectory Editing with Generative Rendering from Estimated Geometry

Few-Shot Efficient Training of Diffusion Models for Industrial Anomaly Synthesis

Emergent Temporal Correspondences from Video Diffusion Transformers

연구실 보유 특허

최근 3년간 연구실적 높은 연구실

대학별 입시끈

우리 연구실 모집