Research
What We Research
Low-level Vision & Computational Photography
[Tasks]
Low-level vision : In-camera Pipeline, Auto White Balance, Super Resolution, etc.
Image / Video Generation & Manipulation
Compositional Understanding & Object-centric Learning
Image Quality Assessment
[Selected Papers]
Attentive Illumination Decomposition Model for Multi-Illuminant White Balance, CVPR 2024 π
Large Scale Multi-Illuminant (LSMI) Dataset for Developing White Balance Algorithm Under Mixed Illumination, ICCV 2021 π
Accelerating Large Image Super-Resolution Networks with Pixel-Level Classification, ECCV 2024
Deep video super-resolution network using dynamic upsampling filters without explicit motion compensation, CVPR 2018 π
Shepherding Slots to Objects: Towards Stable and Robust Object-Centric Learning, CVPR 2023 π
Dense Interspecies Face Embedding, NeurIPS 2022 π
Video Understanding
[Tasks]
Video Segmentation
Streaming Perception, Temporal Action Detection
Vision Language Modeling
Vision for Robotics
[Selected Papers]
VISAGE: Video Instance Segmentation with Appearance-Guided Enhancement, ECCV 2024 π
A Generalized Framework for Video Instance Segmentation, CVPR 2023 π
VISOLO: Grid-Based Space-Time Aggregation for Efficient Online Video Instance Segmentation, CVPR 2022 (Oral) π
Video Object Segmentation using Space-Time Memory Networks, ICCV 2019 (Oral) π
ActionSwitch: Class-agnostic Detection of Simultaneous Actions in Streaming Videos, ECCV 2024
MiniROAD: Minimal Framework for Online Action Detection, ICCV 2023 π
Exploring Scalability of Self-Training for Open-Vocabulary Temporal Action Localization, arxiv 2024 π
3D Vision
[Tasks]
Novel-View Synthesis (Neural Radiance Fields, Gaussian Splatting, etc.)
3D Representation Learning
Animatable 3D Model Reconstruction
Point Cloud Understanding
Non-line-of-sight Imaging
[Selected Papers]
Hierarchically Structured Neural Bones for Reconstructing Animatable Objects from Casual Videos, ECCV 2024
Domain Reduction Strategy for Non-Line-of-Sight Imaging, ECCV 2024
Learning to Enhance Aperture Phasor Field for Non-Line-of-Sight Imaging, ECCV 2024Β
EPINET: A Fully-Convolutional Neural Network Using Epipolar Geometry for Depth from Light Field Images, CVPR 2018 π