Soongjin Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Soongjin Kim has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Jaesik Park
Seoul National University
3 shared publications - Jungeon Kim
POSTECH South Korea
3 shared publications - Seungyong Lee
POSTECH, Pohang, Republic of Korea
3 shared publications - Jaewon Kam
POSTECH
1 shared publication
Latest Publications
- 29 October 2024
- Published by Springer Nature
- p. 328-346
Deep Cost Ray Fusion for Sparse Depth Video Completion
- 29 October 2024
- Published by Springer Nature
- p. 328-346
Deep Cost Ray Fusion for Sparse Depth Video Completion
- Open Access
Deep Cost Ray Fusion for Sparse Depth Video Completion
In this paper, we present a learning-based framework for sparse depth video completion. Given a sparse depth map and a color image at a certain viewpoint, our approach makes a cost volume that is constructed on depth hypothesis planes. To effectively fuse sequential cost volumes of the multiple viewpoints for improved depth completion, we introduce a learning-based cost volume fusion framework, namely RayFusion, that effectively leverages the attention mechanism for each pair of overlapped rays in adjacent cost volumes. As a result of leveraging feature statistics accumulated over time, our proposed framework consistently outperforms or rivals state-of-the-art approaches on diverse indoor and outdoor datasets, including the KITTI Depth Completion benchmark, VOID Depth Completion benchmark, and ScanNetV2 dataset, using much fewer network parameters.
Open AccessDeep Cost Ray Fusion for Sparse Depth Video Completion
In this paper, we present a learning-based framework for sparse depth video completion. Given a sparse depth map and a color image at a certain viewpoint, our approach makes a cost volume that is constructed on depth hypothesis planes. To effectively fuse sequential cost volumes of the multiple viewpoints for improved depth completion, we introduce a learning-based cost volume fusion framework, namely RayFusion, that effectively leverages the attention mechanism for each pair of overlapped rays in adjacent cost volumes. As a result of leveraging feature statistics accumulated over time, our proposed framework consistently outperforms or rivals state-of-the-art approaches on diverse indoor and outdoor datasets, including the KITTI Depth Completion benchmark, VOID Depth Completion benchmark, and ScanNetV2 dataset, using much fewer network parameters.
- 11 November 2022
- Published by Springer Nature
- p. 257-274
CostDCNet: Cost Volume Based Depth Completion for a Single RGB-D Image
- 11 November 2022
- Published by Springer Nature
- p. 257-274
CostDCNet: Cost Volume Based Depth Completion for a Single RGB-D Image