Publications

Peer-reviewed conference papers, journal articles, and manuscripts under review.

Conference Papers & Preprints

International conferences, workshops, and manuscripts under review

202620 papers

Pair-Relative Partial Equivariance for Rotation-Robust Keypoint Descriptor Matching
Jongmin Lee
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2027
Keypoint DescriptorsPartial EquivarianceRotation-Robust Matching
Extending a Large View Synthesis Model for Multi-view Panoptic Segmentation
Kwonyoung Ryu, In-Jae Lee, Jonghyun Jin, Hyunjee Lee, Jongmin Lee, Jaesik Park
European Conference on Computer Vision (ECCV) 2026
Novel-View SynthesisMulti-View Consistency3D Scene Understanding
Project arXiv
URHead: A Unified UV-Space Representation for Joint Mesh–3DGS Optimization in Head Avatars
Seonghak Lee, Junhee Cho, Jisoo Park, Min-Gyu Park, Jongmin Lee, Ju Hong Yoon, Junseok Kwon
European Conference on Computer Vision (ECCV) 2026
Head Avatars3D Gaussian SplattingUV-Space Representation
Project arXiv
SetPieceRAG: Domain-Specific RAG for Knowledge-Intensive Soccer VQA with Large Language Models
Young Seon Kim, Jongmin Lee
CVPR 2026 Workshop on Computer Vision in Sports (CVSports)
SpotlightRetrieval-Augmented GenerationSoccer VQAMultimodal LLMsSoccerNet
Paper Workshop Poster
Steering Where to Attend: 3D-Aware Attention Bias for Frozen Vision-Language-Action Policies
Jeongjae Lee, Jongmin Lee
Under review
Vision-Language-Action3D-Aware AttentionRobot Manipulation
Potential-Guided Particle Steering for Negation-Constrained Dexterous Grasping
Geonho Kim*, SooGon Kim*, Jongmin Lee (* Equal Contributions)
Under review
Dexterous GraspingNegation ConstraintsSequential Monte CarloInference-Time Steering
Correct or Retain? Selective 3D Template Correction for Single-Anchor Model-Free 6D Pose Estimation
Sihyeong Park, Jongmin Lee
Under review
6D Pose Estimation3D Template CorrectionSelective Refinement
Context Matters: Test-Time View Selection and Adaptation for Model-Free 6D Object Pose Estimation
Minseo Cho, Sihyeong Park, Jongmin Lee
Under review
6D Pose EstimationTest-Time AdaptationView Selection
Wings Wide Shut: Improving Spread-Wing Avian 3D Reconstruction with WingFit
Yunchan Jeon, CheolWon Lee, Jongmin Lee
Under review
Avian 3D ReconstructionArticulated Shape FittingAnimal Pose & Shape
Beyond the Face: Part-Aware Dirichlet Aggregation for Soccer Player Grounding
Youngseon Kim, Yunchan Jeon, Jongmin Lee
Under review
Player GroundingPart-Aware AggregationDirichlet ModelingSports Video Understanding
RelCompat3D: Re-Ranking 3D Scene Graph Relations with Geometric Evidence
Yoohyun Kim, Jongmin Lee
Under review
3D Scene GraphsRelation Re-RankingGeometric CompatibilityOpen-Vocabulary Relations
Locate, Then Orient: Factorized Supervision for Situated Pose Retrieval in 3D Scenes
Cheoloh Park, Jongmin Lee
Under review
Situated Pose RetrievalEmbodied 3D ReasoningFactorized SupervisionSQA3D
Closing the Empirical Equivariance Gap of Multi-View 3D Foundation Models
Seungwook Kim, Jongmin Lee
Under review
3D Foundation ModelsEmpirical EquivarianceTest-Time Symmetric Marginalization
Mode- and Mesh-Aware Candidate Selection for Single-Anchor Model-Free 6D Pose Estimation
Sihyeong Park, Jongmin Lee
Under review
6D Pose EstimationCandidate Re-RankingMesh-Aware Selection
Training-Free Textual Prescription Refinement for Safe Medication Recommendation
Youngseon Kim, Jongmin Lee
Under review
Medication RecommendationLLM Critic & RefinementDrug–Drug Interaction SafetyClinical Decision Support
PanSplat3R: Feed-forward Panoptic 3D Gaussian Fields from Unposed Multi-View Images
Kwonyoung Ryu, In-Jae Lee, Jonghyun Jin, Hyunjee Lee, Jongmin Lee, Jaesik Park
Under review
Feed-Forward 3D Reconstruction3D Gaussian SplattingPanoptic Segmentation
Beyond Family Labels: A Taxonomic Ornstein–Uhlenbeck Prior for Avian 3D Shape Recovery
Yunchan Jeon*, SooGon Kim*, Seungwook Kim, CheolWon Lee, Jongmin Lee (* Equal Contributions)
Under review
Avian 3D Shape RecoveryPhylogenetic PriorsOrnstein–Uhlenbeck ProcessLong-Tail Generalization
ProxySearch: Decoupled Inference-Time Scaling for Diffusion Models via Asymmetric Noise-Rank Transfer
Seungwook Kim, Jongmin Lee
Under review
Diffusion ModelsInference-Time ScalingText-to-Image GenerationDistilled Proxy ModelsNoise Search
From Natural Language to Traffic Assignment: A Tool-Constrained LLM Agent for Stateful Macroscopic Analysis
Soogon Kim, Gyeongjun Kim, Keemin Sohn, Jongmin Lee
Under review
LLM AgentsTraffic AssignmentTravel Demand ModelingTool-Constrained Workflows
DUET: Two-Timescale State Updates for Training-Free Streaming 3D Reconstruction
Vladislav Viktorovich Kim, Jongmin Lee
Under review
Streaming 3D ReconstructionTraining-FreeTwo-Timescale State Updates

20251 paper

EXAONE 4.0 VL: Vision-Language Foundation Model for Enterprise AI Agent
Jongmin Lee, LG AI Research
LG AI Research Talk Concert 2025
Vision-Language ModelFoundation ModelsEnterprise AIDocument Understanding
Announcement Video

20241 paper

3D Equivariant Pose Regression via Direct Wigner-D Harmonics Prediction
Jongmin Lee, Minsu Cho
Conference on Neural Information Processing Systems (NeurIPS) 2024
3D Rotation EstimationSO(3) EquivarianceWigner-D HarmonicsFrequency-Domain Learning
Paper Code Project

20231 paper

Learning Rotation-Equivariant Features for Visual Correspondence
Jongmin Lee, Byungjin Kim, Seungwook Kim, Minsu Cho
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2023
Rotation EquivarianceLocal Feature DescriptorsVisual CorrespondenceGroup-Equivariant CNNs
Paper Code Project

20221 paper

Self-Supervised Equivariant Learning for Oriented Keypoint Detection
Jongmin Lee, Byungjin Kim, Minsu Cho
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2022
Oriented Keypoint DetectionRotation EquivarianceSelf-Supervised Learning
Paper Code Project

20212 papers

Self-supervised Learning of Image Scale and Orientation Estimation
Jongmin Lee, Yoonwoo Jeong, Minsu Cho
British Machine Vision Conference (BMVC) 2021
Scale & Orientation EstimationSelf-Supervised LearningLocal Features
Paper Code
Learning to Distill Convolutional Features Into Compact Local Descriptors
Jongmin Lee, Yoonwoo Jeong, Seungwook Kim, Juhong Min, Minsu Cho
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2021
Local DescriptorsKnowledge DistillationFeature Matching
Paper

20201 paper

Learning to Compose Hypercolumns for Visual Correspondence
Juhong Min, Jongmin Lee, Jean Ponce, Minsu Cho
European Conference on Computer Vision (ECCV) 2020
Semantic CorrespondenceHypercolumn FeaturesDynamic Layer Selection
Paper Code Project

20191 paper

Hyperpixel Flow: Semantic Correspondence with Multi-layer Neural Features
Juhong Min, Jongmin Lee, Jean Ponce, Minsu Cho
International Conference on Computer Vision (ICCV) 2019
Semantic CorrespondenceHyperpixel FlowMulti-Layer FeaturesRegularized Hough Matching
Paper Code Project

20181 paper

Attentive Semantic Alignment with Offset-Aware Correlation Kernels
Paul Hongsuck Seo, Jongmin Lee, Deunsol Jung, Bohyung Han, Minsu Cho
European Conference on Computer Vision (ECCV) 2018
Semantic AlignmentAttentionOffset-Aware CorrelationGeometric Transformation Estimation
Paper Code Project

Publications

Pair-Relative Partial Equivariance for Rotation-Robust Keypoint Descriptor Matching

Extending a Large View Synthesis Model for Multi-view Panoptic Segmentation

URHead: A Unified UV-Space Representation for Joint Mesh–3DGS Optimization in Head Avatars

SetPieceRAG: Domain-Specific RAG for Knowledge-Intensive Soccer VQA with Large Language Models

Steering Where to Attend: 3D-Aware Attention Bias for Frozen Vision-Language-Action Policies

Potential-Guided Particle Steering for Negation-Constrained Dexterous Grasping

Correct or Retain? Selective 3D Template Correction for Single-Anchor Model-Free 6D Pose Estimation

Context Matters: Test-Time View Selection and Adaptation for Model-Free 6D Object Pose Estimation

Wings Wide Shut: Improving Spread-Wing Avian 3D Reconstruction with WingFit

Beyond the Face: Part-Aware Dirichlet Aggregation for Soccer Player Grounding

RelCompat3D: Re-Ranking 3D Scene Graph Relations with Geometric Evidence

Locate, Then Orient: Factorized Supervision for Situated Pose Retrieval in 3D Scenes

Closing the Empirical Equivariance Gap of Multi-View 3D Foundation Models

Mode- and Mesh-Aware Candidate Selection for Single-Anchor Model-Free 6D Pose Estimation

Training-Free Textual Prescription Refinement for Safe Medication Recommendation

PanSplat3R: Feed-forward Panoptic 3D Gaussian Fields from Unposed Multi-View Images

Beyond Family Labels: A Taxonomic Ornstein–Uhlenbeck Prior for Avian 3D Shape Recovery

ProxySearch: Decoupled Inference-Time Scaling for Diffusion Models via Asymmetric Noise-Rank Transfer

From Natural Language to Traffic Assignment: A Tool-Constrained LLM Agent for Stateful Macroscopic Analysis

DUET: Two-Timescale State Updates for Training-Free Streaming 3D Reconstruction

EXAONE 4.0 VL: Vision-Language Foundation Model for Enterprise AI Agent

3D Equivariant Pose Regression via Direct Wigner-D Harmonics Prediction

Learning Rotation-Equivariant Features for Visual Correspondence

Self-Supervised Equivariant Learning for Oriented Keypoint Detection

Self-supervised Learning of Image Scale and Orientation Estimation

Learning to Distill Convolutional Features Into Compact Local Descriptors

Learning to Compose Hypercolumns for Visual Correspondence

Hyperpixel Flow: Semantic Correspondence with Multi-layer Neural Features

Attentive Semantic Alignment with Offset-Aware Correlation Kernels