We develop scalable methods for various graphics and geometric applications. In addition, we expand our research area to various fields of computer graphics, computer vision, and robotics.

Companies and funding agencies that we’ve worked together

Research Areas

We are currently working on three applications: Rendering, Vision, and Robotics. These applications seem to be very different, but their core engines require scalable proximity computing. As a result, we focus on designing scalable proximity computing and their applications to those high-level applications.

Rendering (papers)

SGVR Lab has great expertise in rendering. We have covered several topics including rasterization, ray-tracing and neural rendering. Please refer to an on-going book by Prof. Yoon to learn more about the topics. Here, we have selected our recent rendering-related work.

Artificial intelligence + rendering

Recently, we are actively studying the neural rendering area where we integrate neural networks for a new paradigm of rendering.

Weakly-Supervised Contrastive Learning in Path Manifold for Monte Carlo Image Reconstruction
In-Young Cho, Yuchi Huo, and Sung-Eui Yoon
TopACM Trans. on Graphics (ToG) (proc. of SIGGRAPH), 2021
Adaptive Incident Radiance Field Sampling and Reconstruction Using Deep Reinforcement Learning
TopACM Trans. on Graphics (ToG), 2020
Chosen as the cover image of the journal issue
Paper Sup. report

Blockchain for 3D models

MeshChain: Secure 3D model and intellectual property management powered by blockchain technology
Hunmin Park, Yuchi Huo, and Sung-Eui Yoon
CGI, 2021

Computer Vision (papers)

Image search is a field where the proximity computing is required. Useful concepts for image search have been developed and extended to various vision applications. Here’s a list of recent papers.

Large-scale image search

Hypergraph Propagation and Community Selection for Objects Retrieval
Guoyuan An, Yuchi Huo, and Sung-Eui Yoon
Top35th Conference on Neural Information Processing Systems (NeurIPS), 2021
Distance Encoded Product Quantization for Approximate K-Nearest Neighbor Search in High-Dimensional Space
Jae-Pil Heo, Zhe Lin, and Sung-Eui Yoon
IEEE Transactions on Pattern Analysis and Machine Intelligence(TPAMI), 2018
Spherical Hashing
Jae-Pil Heo, Youngwoon Lee, Junfeng He, Shih-Fu Chang, and Sung-eui Yoon
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012

Other applications

We are also interested in various topics of computer vision with machine-learning, including optical flow, inpainting, and reflection removal.

Part-based Pseudo Label Refinement for Unsupervised Person Re-identification
Yoonki Cho, Woo Jae Kim, Seunghoon Hong, and Sung-Eui Yoon
TopIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022
Single Image Reflection Removal with Physically-Based Training Images
Soomin Kim, Yuchi Huo, and Sung-Eui Yoon
TopComputer Vision and Pattern Recognition (CVPR), 2020
Oral paper
Unsupervised Learning of Optical Flow with Deep Feature Similarity
Woobin Im, Tae-Kyun Kim, and Sung-Eui Yoon
TopEuropean Conference on Computer Vision (ECCV), 2020

Robotics (papers)

We also have extensively studied the field of robotics, with special expertise in motion planning. Furthermore, we want robots to better understand the environment via sensors, including 3D point cloud (LiDAR) and even sound (microphones). Recently, we are also integrating deep learning based approach to our robotics research. We selected recent papers here.

Mapping & Motion Planning

Confidence-Based Robot Navigation Under Sensor Occlusion with Deep Reinforcement Learning
Hyeongyeol Ryu, Minsung Yoon, Daehyung Park, and Sung-Eui Yoon
TopIEEE International Conference on Robotics and Automation (ICRA), 2022
Outstanding navigation paper finalist
Super Rays and Culling Region for Real-Time Updates on Grid-based Occupancy Maps
Youngsun Kwon, Donghyuk Kim, Inkyu An, and Sung-Eui Yoon
TopIEEE Transactions on Robotics (T-RO), 2019
Source code: Github ZIP
TORM: Fast and Accurate Trajectory Optimization of Redundant Manipulator given an End-Effector Path
Mincheul Kang, Heechan Shin, Donghyuk Kim, and Sung-Eui Yoon
TopIEEE/RSJ Int. Conf. on Intelligent Robots and Systems (IROS), 2020
Source code: Github ZIP

Sound localization

Diffraction- and Reflection-Aware Multiple Sound Source Localization
Inkyu An, Youngsun Kwon, and Sung-Eui Yoon
TopIEEE Transactions On Robotics (T-RO), 2021
Robust Sound Source Localization considering Similarity of Back-Propagation Signals
Inkyu An, Byeongho-Jo, Youngsun Kwon, Jung-woo Choi, and and Sung-Eui Yoon
TopIEEE Int. Conf. on Robotics and Automation (ICRA), 2020

Legacy Techniques

In the history of our lab, we focused on the techniques listed below, which are not buried, but ready to be revisited and re-discovered in a new perspective.

To design scalable applications, we work on various orthogonal technologies: