Kaiwen Yuan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Kaiwen Yuan has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Z. Jane Wang
University of British Columbia
4 shared publications - Mazen Abdelfattah
Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada
2 shared publications - Alan Luo
Siebel School of Computing and Data Sciences, University of Illinois Urbana-Champaign, Urbana, IL, USA
1 shared publication Alireza Alizadegan
University of British Columbia
1 shared publication- Biao Tian
State Key Laboratory of Reliability and Intelligence of Electrical Equipment, School of Health Sciences and Biomedical Engineering, Hebei University of Technology, Tianjin, China
1 shared publication - Bo Chen
Shuguang Hospital
1 shared publication
Latest Publications
- 12 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 33 (10709908) , 331-335
Simple Self-Organizing Map With Vision Transformers
Vision Transformers (ViTs) have demonstrated exceptional performance in various vision tasks. However, they tend to underperform on smaller datasets due to their inherent lack of inductive biases. Current approaches address this limitation implicitly-often by pairing ViTs with pretext tasks or by distilling knowledge from convolutional neural networks (CNNs) to strengthen the prior. In contrast, Self-Organizing Maps (SOMs), a widely adopted self-supervised framework, are inherently structured to preserve topology and spatial organization, making them a promising candidate to directly address the limitations of ViTs in limited or small training datasets. Despite this potential, equipping SOMs with modern deep learning architectures remains largely unexplored. In this study, we conduct a novel exploration on how Vision Transformers (ViTs) and Self-Organizing Maps (SOMs) can empower each other, aiming to bridge this critical research gap. Our findings demonstrate that these architectures can synergistically enhance each other, leading to significantly improved performance in both unsupervised and supervised tasks. Code is publicly available on GitHub https://github.com/aluo7/ViT-SOM.
- 12 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 33 (10709908) , 331-335
Simple Self-Organizing Map With Vision Transformers
Vision Transformers (ViTs) have demonstrated exceptional performance in various vision tasks. However, they tend to underperform on smaller datasets due to their inherent lack of inductive biases. Current approaches address this limitation implicitly-often by pairing ViTs with pretext tasks or by distilling knowledge from convolutional neural networks (CNNs) to strengthen the prior. In contrast, Self-Organizing Maps (SOMs), a widely adopted self-supervised framework, are inherently structured to preserve topology and spatial organization, making them a promising candidate to directly address the limitations of ViTs in limited or small training datasets. Despite this potential, equipping SOMs with modern deep learning architectures remains largely unexplored. In this study, we conduct a novel exploration on how Vision Transformers (ViTs) and Self-Organizing Maps (SOMs) can empower each other, aiming to bridge this critical research gap. Our findings demonstrate that these architectures can synergistically enhance each other, leading to significantly improved performance in both unsupervised and supervised tasks. Code is publicly available on GitHub https://github.com/aluo7/ViT-SOM.
- 9 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 8 (10) , 6163-6170
Multi-Modal Streaming 3D Object Detection
Modern autonomous vehicles rely heavily on mechanical LiDARs for perception. Current perception methods generally require $360^\circ$ point clouds, collected sequentially as the LiDAR scans the azimuth and acquires consecutive wedge-shaped slices. The acquisition latency of a full scan ( $\sim 100ms$ ) may lead to outdated perception which is detrimental to safe operation. Recent streaming perception works proposed directly processing LiDAR slices and compensating for the narrow field of view (FOV) of a slice by reusing features from preceding slices. These works, however, are all based on a single modality and require past information which may be outdated. Meanwhile, images from high-frequency cameras can support streaming models as they provide a larger FoV compared to a LiDAR slice. However, this difference in FoV complicates sensor fusion. We propose an innovative camera-LiDAR streaming 3D object detection framework that uses camera images instead of past LiDAR slices to provide an up-to-date, dense, and wide context for streaming perception. The proposed method outperforms prior streaming models and powerful full-scan baselines on the challenging NuScenes benchmark in detection accuracy and end-to-end runtime. Our method is shown to be robust to missing camera images, narrow LiDAR slices, and small camera-LiDAR miscalibration. Project website: mmstream.github.io
- 9 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 8 (10) , 6163-6170
Multi-Modal Streaming 3D Object Detection
Modern autonomous vehicles rely heavily on mechanical LiDARs for perception. Current perception methods generally require $360^\circ$ point clouds, collected sequentially as the LiDAR scans the azimuth and acquires consecutive wedge-shaped slices. The acquisition latency of a full scan ( $\sim 100ms$ ) may lead to outdated perception which is detrimental to safe operation. Recent streaming perception works proposed directly processing LiDAR slices and compensating for the narrow field of view (FOV) of a slice by reusing features from preceding slices. These works, however, are all based on a single modality and require past information which may be outdated. Meanwhile, images from high-frequency cameras can support streaming models as they provide a larger FoV compared to a LiDAR slice. However, this difference in FoV complicates sensor fusion. We propose an innovative camera-LiDAR streaming 3D object detection framework that uses camera images instead of past LiDAR slices to provide an up-to-date, dense, and wide context for streaming perception. The proposed method outperforms prior streaming models and powerful full-scan baselines on the challenging NuScenes benchmark in detection accuracy and end-to-end runtime. Our method is shown to be robust to missing camera images, narrow LiDAR slices, and small camera-LiDAR miscalibration. Project website: mmstream.github.io
- 6 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 8 (2) , 944-950
A Simple Self-Supervised IMU Denoising Method for Inertial Aided Navigation
Inertial Measurement Unit (IMU) plays an important role in inertial aided navigation on robots. However, raw IMU data could be noisy, especially for low-cost IMUs, and thus requires efficient pre-processing or denoising before applying further navigation algorithms. Conventional IMU denoising approaches are mostly hand-crafted and may face concerns such as sensor modelling errors and generalization issues. Several recent works leverage deep neural networks (DNNs) to tackle this problem and achieve promising results. However, currently reported deep learning methods are based on supervised learning, requiring sufficient and accurate annotations. While in real-world applications, such annotations can be expensive or unavailable, making these methods not practical. To address the above research gap, we propose incorporating self-supervised learning and future-aware inference for IMU denoising. The end-to-end navigation evaluation results on EuRoC and TUM-VI datasets are promising. The code will be publicly available at https://github.com/KleinYuan/IMUDB .
- 6 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 8 (2) , 944-950
A Simple Self-Supervised IMU Denoising Method for Inertial Aided Navigation
Inertial Measurement Unit (IMU) plays an important role in inertial aided navigation on robots. However, raw IMU data could be noisy, especially for low-cost IMUs, and thus requires efficient pre-processing or denoising before applying further navigation algorithms. Conventional IMU denoising approaches are mostly hand-crafted and may face concerns such as sensor modelling errors and generalization issues. Several recent works leverage deep neural networks (DNNs) to tackle this problem and achieve promising results. However, currently reported deep learning methods are based on supervised learning, requiring sufficient and accurate annotations. While in real-world applications, such annotations can be expensive or unavailable, making these methods not practical. To address the above research gap, we propose incorporating self-supervised learning and future-aware inference for IMU denoising. The end-to-end navigation evaluation results on EuRoC and TUM-VI datasets are promising. The code will be publicly available at https://github.com/KleinYuan/IMUDB .
- Open Access
- 4 November 2022
- Published by Springer Nature in Research Square
Shi-style Steaming and Bathing Decoction on Knee Osteoarthritis by Decreasing Synovial Fibrosis and Angiogenesis
Aim of the study: A discussion of the effect of Shi-style steaming and bathing decoction on synovial fibrosis and vascular hyperplasia. Materials and methods An anterior cruciate ligament transection (ACLT method) was used to replicate the knee osteoarthritis (KOA) model. After 4 weeks, and synovial tissues were taken for hematoxylin-eosin staining (HE) and Krenn score to verify whether the KOA model was successfully established. After the 4-week successful modeling, each group used different interventions for 4 weeks, while the sham-operated and KOA model groups were not intervened. At the end of the intervention, HE staining and Masson staining were used and Krenn scores and positive fiber areas were calculated.Femoral artery blood flows in the right lower limb was detected with a color Doppler ultrasound machine. The protein expression level of VEGFA was determined by enzyme-linked immunosorbent assay and immunohistochemical assay, and the quantitatively analyze the mRNA of VEGFA using Real-time PCR . Results The animals in Shi-style steaming and bathing decoction herbal bath treatment group generally performed well. HE staining, Krenn score and Masson staining indicated that Shi-style steaming and bathing decoction herbal baths reduced the number of synovial inflammatory cells, attenuated synovial fibrosis, decreased blood flow velocity, alleviated vascular resistance, and decreased the protein expression level of VEGFA in vivo and the mRNA expression level. Conclusions In the rabbit KOA model, Shi-style steaming and bathing decoction herbal bath at 40°C can alleviate synovial inflammation, fibrosis and revascularization, and improve local blood circulation of the knee joint. Therefore, it is feasible for the treatment of knee osteoarthritis.
- 4 November 2022
- Published by Springer Nature in Research Square
Open AccessShi-style Steaming and Bathing Decoction on Knee Osteoarthritis by Decreasing Synovial Fibrosis and Angiogenesis
Aim of the study: A discussion of the effect of Shi-style steaming and bathing decoction on synovial fibrosis and vascular hyperplasia. Materials and methods An anterior cruciate ligament transection (ACLT method) was used to replicate the knee osteoarthritis (KOA) model. After 4 weeks, and synovial tissues were taken for hematoxylin-eosin staining (HE) and Krenn score to verify whether the KOA model was successfully established. After the 4-week successful modeling, each group used different interventions for 4 weeks, while the sham-operated and KOA model groups were not intervened. At the end of the intervention, HE staining and Masson staining were used and Krenn scores and positive fiber areas were calculated.Femoral artery blood flows in the right lower limb was detected with a color Doppler ultrasound machine. The protein expression level of VEGFA was determined by enzyme-linked immunosorbent assay and immunohistochemical assay, and the quantitatively analyze the mRNA of VEGFA using Real-time PCR . Results The animals in Shi-style steaming and bathing decoction herbal bath treatment group generally performed well. HE staining, Krenn score and Masson staining indicated that Shi-style steaming and bathing decoction herbal baths reduced the number of synovial inflammatory cells, attenuated synovial fibrosis, decreased blood flow velocity, alleviated vascular resistance, and decreased the protein expression level of VEGFA in vivo and the mRNA expression level. Conclusions In the rabbit KOA model, Shi-style steaming and bathing decoction herbal bath at 40°C can alleviate synovial inflammation, fibrosis and revascularization, and improve local blood circulation of the knee joint. Therefore, it is feasible for the treatment of knee osteoarthritis.
- 6 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 38 (5) , 3203-3218
LiCaS3: A Simple LiDAR–Camera Self-Supervised Synchronization Method
Recent advances in robotics and deep learning demonstrate promising 3-D perception performances via fusing the light detection and ranging (LiDAR) sensor and camera data, where both spatial calibration and temporal synchronization are generally required. While the LiDAR–camera calibration problem has been actively studied during the past few years, LiDAR–camera synchronization has been less studied and mainly addressed by employing a conventional pipeline consisting of clock synchronization and temporal synchronization. The conventional pipeline has certain potential limitations, which have not been sufficiently addressed and could be a bottleneck for the potential wide adoption of low-cost LiDAR–camera platforms. Different from the conventional pipeline, in this article, we propose the LiCaS3, the first deep-learning-based framework, for the LiDAR–camera synchronization task via self-supervised learning. The proposed LiCaS3 does not require hardware synchronization or extra annotations and can be deployed both online and offline. Evaluated on both the KITTI and Newer College datasets, the proposed method shows promising performances. The code will be publicly available at https://github.com/KleinYuan/LiCaS3.
- 6 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 38 (5) , 3203-3218
LiCaS3: A Simple LiDAR–Camera Self-Supervised Synchronization Method
Recent advances in robotics and deep learning demonstrate promising 3-D perception performances via fusing the light detection and ranging (LiDAR) sensor and camera data, where both spatial calibration and temporal synchronization are generally required. While the LiDAR–camera calibration problem has been actively studied during the past few years, LiDAR–camera synchronization has been less studied and mainly addressed by employing a conventional pipeline consisting of clock synchronization and temporal synchronization. The conventional pipeline has certain potential limitations, which have not been sufficiently addressed and could be a bottleneck for the potential wide adoption of low-cost LiDAR–camera platforms. Different from the conventional pipeline, in this article, we propose the LiCaS3, the first deep-learning-based framework, for the LiDAR–camera synchronization task via self-supervised learning. The proposed LiCaS3 does not require hardware synchronization or extra annotations and can be deployed both online and offline. Evaluated on both the KITTI and Newer College datasets, the proposed method shows promising performances. The code will be publicly available at https://github.com/KleinYuan/LiCaS3.
- 25 September 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 5 (4) , 6956-6963
RGGNet: Tolerance Aware LiDAR-Camera Online Calibration With Geometric Deep Learning and Generative Model
Accurate LiDAR-camera online calibration is critical for modern autonomous vehicle and robot platforms. Dominant methods heavily rely on hand-crafted features, which are not scalable in practice. With the increasing popularity of deep learning (DL), a few recent efforts have demonstrated the advantages of DL for feature extraction on this task. However, their reported performances are not sufficiently satisfying yet. We believe one improvement can be the problem formulation with proper consideration of the underneath geometry. Besides, existing online calibration methods focus on optimizing the calibration error while overlooking the tolerance within the error bounds. To address the research gap, a DL-based LiDAR-camera calibration method, named as the RGGNet, is proposed by considering the Riemannian geometry and utilizing a deep generative model to learn an implicit tolerance model. When evaluated on KITTI benchmark datasets, the proposed method demonstrates superior performances to the state-of-the-art DL-based methods. The code will be public available at https://github.com/KleinYuan/RGGNet.
- 25 September 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 5 (4) , 6956-6963
RGGNet: Tolerance Aware LiDAR-Camera Online Calibration With Geometric Deep Learning and Generative Model
Accurate LiDAR-camera online calibration is critical for modern autonomous vehicle and robot platforms. Dominant methods heavily rely on hand-crafted features, which are not scalable in practice. With the increasing popularity of deep learning (DL), a few recent efforts have demonstrated the advantages of DL for feature extraction on this task. However, their reported performances are not sufficiently satisfying yet. We believe one improvement can be the problem formulation with proper consideration of the underneath geometry. Besides, existing online calibration methods focus on optimizing the calibration error while overlooking the tolerance within the error bounds. To address the research gap, a DL-based LiDAR-camera calibration method, named as the RGGNet, is proposed by considering the Riemannian geometry and utilizing a deep generative model to learn an implicit tolerance model. When evaluated on KITTI benchmark datasets, the proposed method demonstrates superior performances to the state-of-the-art DL-based methods. The code will be public available at https://github.com/KleinYuan/RGGNet.
- 1 September 2020
- Published by Elsevier in Neurocomputing
- Vol. 406, 215-223
Xnet: Task-specific attentional domain adaptation for satellite-to-aerial scene
- 1 September 2020
- Published by Elsevier in Neurocomputing
- Vol. 406, 215-223
Xnet: Task-specific attentional domain adaptation for satellite-to-aerial scene
- 19 October 2016
- Published by Springer Nature
- p. 637-657
Design, Fabrication, and Robust Control of Miniaturized Optical Image Stabilizers
- 19 October 2016
- Published by Springer Nature
- p. 637-657
Design, Fabrication, and Robust Control of Miniaturized Optical Image Stabilizers