Inchul Kim

Inchul (Incheol) Kim

I am a PhD student of the Visual Computing Lab at the Korea Advanced Institute of Science and Technology (KAIST), South Korea, under supervision of Prof. .

Current research fields: hyperspectral imaging, learning-based image reconstruction, computational imaging

Email / CV / Google Scholar / Github

Publications

	Self-Calibrating, Fully Differentiable NLOS Inverse Rendering , Inchul Kim, , , , Proceedings of ACM SIGGRAPH Asia, 2023 Project / Paper / Supplemental / Code / ACM DL / bibtex We present a flexible, differentiable non-line-of-sight (NLOS) imaging and rendering pipeline that automatically estimates system parameters and the geometry of hidden objects.
	Spatio-Focal Bidirectional Disparity Estimation from a Dual-Pixel Image , , Inchul Kim, IEEE Computer Vision and Pattern Recognition (CVPR), 2023 Project / Paper / Supplemental / Code / IEEE Xplore / bibtex We estimate bidirectional disparities from a dual-pixel image by a network using self-supervised learning. Our method based on the dual-pixel optics produces more physically correct results.
	Motion parallax for 360° RGBD video , Incheol Kim, , , , , IEEE Transactions on Visualization and Computer Graphics (TVCG), Proc. IEEE VR, 2019 (Best journal paper nominee) Project / Paper / Supplemental / Slides / Code / IEEE Xplore / bibtex Our VR video viewer exploits motion parallax cues to reduce VR sickness without additional hardware requirements. Our system provides a compelling viewing experience with reduced discomfort.
	Non-local Haze Propagation with an Iso-Depth Prior Incheol Kim, Computer Vision, Imaging and Computer Graphics - Theory and Applications, 2019 Project / Paper / Supplemental / Code / Springer DL / bibtex More experiments and validations have been added on top of our non-local haze regularization paper in 2017.
	Compact Single-Shot Hyperspectral Imaging Using a Prism , Incheol Kim, , ACM Transactions on Graphics (TOG), presented at SIGGRAPH Asia, 2017 Project / Paper / Supplemental1 / Supplemental2 / Slides / ACM DL / bibtex Press release: EurekAlert! / Photonics / R&D Magazine Patent registered: US Patent / European Patent / PCT Patent Our compact snapshot hyperspectral imaging (HSI) system enables hand-held imaging spectroscopy, and shows strong performance over large HSI systems.
	Dehazing using Non-Local Regularization with Iso-Depth Neighbor-Fields Incheol Kim, International Conference on Computer Vision, Theory and Applications (VISAPP), 2017 (Oral Presentation) Project / Paper / Supplemental / Code / bibtex Our haze regularization exploits non-local information from nearest-neighbor fields and shows the state-of-the-art performance.

Teaching

My website has been built upon Jon Barron's code.