Jen-Hui Chuang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Jen-Hui Chuang has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Ardian Umam
Department of Computer Science, National Yang Ming Chiao Tung University, Hsinchu, Taiwan
3 shared publications - Cheng-Kun Yang
Department of Computer Science and Information Engineering, National Taiwan University, Taipei, Taiwan
3 shared publications - Hsin-Yi Chen
Department of Computer Science, National Yang Ming Chiao Tung University, Hsinchu, Taiwan
3 shared publications - Yen-Yu Lin
National yang Ming Chiao Tung University
3 shared publications Yi-Yu Hsieh
Department of Computer Science, National Yang Ming Chiao Tung University, Hsinchu, Taiwan
3 shared publications- Chen-Tao Hsu
Department of Computer Science, National Chiao Tung University, Hsinchu, Taiwan
2 shared publications
Latest Publications
- 27 October 2026
- Published by Elsevier in Computer Vision and Image Understanding
- Vol. 274
PartDistill++: Distilling multiple foundation models for robust 3D shape part segmentation
- 27 October 2026
- Published by Elsevier in Computer Vision and Image Understanding
- Vol. 274
PartDistill++: Distilling multiple foundation models for robust 3D shape part segmentation
- 16 December 2025
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- Vol. 142 (3-4) , 1495-1504
Comparison of spatial and frequency domain methods for detecting linear texture defects on yarn package surface
- 16 December 2025
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- Vol. 142 (3-4) , 1495-1504
Comparison of spatial and frequency domain methods for detecting linear texture defects on yarn package surface
- Open Access
- 22 September 2025
- Published by Springer Nature in Research Square
Vision-Based UAV Navigation Using Multi-Intensity Illumination and Fiducial Markers in GPS-Denied Environments
In hazardous or restricted environments, such as tunnels, industrial facilities, or mountainous terrains, conventional inspection tasks pose significant risks. Leveraging the high mobility of unmanned aerial vehicles (UAVs), such tasks can be performed more safely and efficiently. While GPS provides reliable positioning in outdoor settings, its absence in GPS-denied environments presents substantial challenges for autonomous navigation. This study proposes a vision-based UAV navigation system utilizing fiducial markers and a multi-intensity illumination scheme to enable accurate localization in GPS-denied scenarios. The system operates through a three-stage guidance framework: long-distance, short-distance, and target-shift. A custom illuminator enhances marker visibility under varying lighting conditions, while the YOLOv5 model facilitates target recognition throughout the flight. As the UAV approaches the marker, the ArUco Marker-Based Positioning System (AMBPS) ensures precise positioning and stable hovering. Experimental results demonstrate that the UAV can successfully navigate from an initial position 50 meters away to the predefined location marked by the ArUco board. Across 120 recorded measurements, the system achieved a positioning error within ±0.15 meters. The core technique of this navigation framework has been patented under ROC Patent No. I862216.
- 22 September 2025
- Published by Springer Nature in Research Square
Open AccessVision-Based UAV Navigation Using Multi-Intensity Illumination and Fiducial Markers in GPS-Denied Environments
In hazardous or restricted environments, such as tunnels, industrial facilities, or mountainous terrains, conventional inspection tasks pose significant risks. Leveraging the high mobility of unmanned aerial vehicles (UAVs), such tasks can be performed more safely and efficiently. While GPS provides reliable positioning in outdoor settings, its absence in GPS-denied environments presents substantial challenges for autonomous navigation. This study proposes a vision-based UAV navigation system utilizing fiducial markers and a multi-intensity illumination scheme to enable accurate localization in GPS-denied scenarios. The system operates through a three-stage guidance framework: long-distance, short-distance, and target-shift. A custom illuminator enhances marker visibility under varying lighting conditions, while the YOLOv5 model facilitates target recognition throughout the flight. As the UAV approaches the marker, the ArUco Marker-Based Positioning System (AMBPS) ensures precise positioning and stable hovering. Experimental results demonstrate that the UAV can successfully navigate from an initial position 50 meters away to the predefined location marked by the ArUco board. Across 120 recorded measurements, the system achieved a positioning error within ±0.15 meters. The core technique of this navigation framework has been patented under ROC Patent No. I862216.
- Open Access
- 10 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 161854-161864
Pentagon-Match (PMatch): Identification of View-Invariant Planar Feature for Local Feature Matching-Based Homography Estimation
In computer vision, finding correct point correspondence among images plays an important role in many applications, such as image stitching, image retrieval, visual localization, etc. Most of the research works focus on the matching of local features before a sampling method is employed, such as RANSAC, to verify initial matching results via repeated fitting of certain global transformation among the images. In this paper, a novel sampling scheme, Pentagon-Match (PMatch), is proposed for more efficient verification of the correctness of initially matched keypoints using pentagons randomly sampled from them. By ensuring shape and location of some sampled pentagons are view-invariant with various evaluations of cross-ratio (CR), incorrect matches of keypoint can then be identified easily with homography estimated from these pentagons. Experimental results show that highly accurate estimation of homography can be obtained efficiently and more accurately for planar scenes of the HPatches dataset, based on keypoint matching results provided by LoFTR. Besides, a new way of showing inliers (and outlier) for the above matching results and possible extension of the approach for multi-plane situation are also demonstrated.
- 10 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 161854-161864
Open AccessPentagon-Match (PMatch): Identification of View-Invariant Planar Feature for Local Feature Matching-Based Homography Estimation
In computer vision, finding correct point correspondence among images plays an important role in many applications, such as image stitching, image retrieval, visual localization, etc. Most of the research works focus on the matching of local features before a sampling method is employed, such as RANSAC, to verify initial matching results via repeated fitting of certain global transformation among the images. In this paper, a novel sampling scheme, Pentagon-Match (PMatch), is proposed for more efficient verification of the correctness of initially matched keypoints using pentagons randomly sampled from them. By ensuring shape and location of some sampled pentagons are view-invariant with various evaluations of cross-ratio (CR), incorrect matches of keypoint can then be identified easily with homography estimated from these pentagons. Experimental results show that highly accurate estimation of homography can be obtained efficiently and more accurately for planar scenes of the HPatches dataset, based on keypoint matching results provided by LoFTR. Besides, a new way of showing inliers (and outlier) for the above matching results and possible extension of the approach for multi-plane situation are also demonstrated.
- 19 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-11
Alleviating Radial Distortion Effect for Accurate, Iterative Camera Calibration Using Principal Lines
Preparing appropriate images for camera calibration is crucial to accurate calibration results. In this paper, guidelines for preparing such data to alleviate the effect of radial distortion for accurate calibration procedure, using principal lines, are developed through (i) identifying directions of movement of calibration (checkerboard) pattern in an image which will result in maximum influence (deflection of the principal line) on the calibration results, and (ii) inspecting symmetry and monotonicity of such adverse effect in (i). Accordingly, it is suggested that an iterative estimation of principal point with the above calibration procedure is based on linearly independent pairs of nearly parallel principal lines, with one member in each pair corresponding to a 180° rotation (in the image plane) of the other. Experimental results on synthetic data show that more accurate (and robust) calibration can be achieved with the foregoing iterative approach using principal lines compared with a similar implementation using the renowned algebraic (Zhang’s) method which explicitly estimates distortion parameters in the calibration. Moreover, similar results are obtained for real data via cross-validation of different sets of calibrated intrinsic/extrinsic camera parameters using reprojection errors, and from the consistency of flow vector field of camera distortion associated with different sets of intrinsic parameters.
- 19 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-11
Alleviating Radial Distortion Effect for Accurate, Iterative Camera Calibration Using Principal Lines
Preparing appropriate images for camera calibration is crucial to accurate calibration results. In this paper, guidelines for preparing such data to alleviate the effect of radial distortion for accurate calibration procedure, using principal lines, are developed through (i) identifying directions of movement of calibration (checkerboard) pattern in an image which will result in maximum influence (deflection of the principal line) on the calibration results, and (ii) inspecting symmetry and monotonicity of such adverse effect in (i). Accordingly, it is suggested that an iterative estimation of principal point with the above calibration procedure is based on linearly independent pairs of nearly parallel principal lines, with one member in each pair corresponding to a 180° rotation (in the image plane) of the other. Experimental results on synthetic data show that more accurate (and robust) calibration can be achieved with the foregoing iterative approach using principal lines compared with a similar implementation using the renowned algebraic (Zhang’s) method which explicitly estimates distortion parameters in the calibration. Moreover, similar results are obtained for real data via cross-validation of different sets of calibrated intrinsic/extrinsic camera parameters using reprojection errors, and from the consistency of flow vector field of camera distortion associated with different sets of intrinsic parameters.
- 28 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 26 (15209210) , 7775-7786
Unsupervised Point Cloud Co-Part Segmentation via Co-Attended Superpoint Generation and Aggregation
We propose a co-part segmentation method that takes a set of point clouds of the same category as input where neither a ground truth label nor a prior network is required. With difficulties caused by the label absence, we formulate the co-part segmentation task into two subtasks, including superpoint generation and part aggregation. In the first subtask, our superpoint generation network divides each point cloud into homogeneous partitions, each called superpoint, while in the second subtask, these superpoints are further aggregated into a few semantic parts via our part aggregation network. We introduce the coupled attention blocks in the part aggregation network to explicitly enforce semantic consistency in the segmentation by exploiting intra-, inter-, and paired-cloud geometrical information by minimizing the devised intra-, inter-, and paired-cloud losses, respectively. The intra-cloud loss triggers a semantic segmentation in each point cloud, while the inter-cloud loss considers all clouds to enforce their semantic consistency. The paired-cloud loss is designed to ensure that each part of one point cloud can be discriminatively reconstructed from the superpoints of another point cloud. We perform experiments on two benchmark datasets, ShapeNet part and COSEG, and provide quantitative and qualitative results to demonstrate the superiority of our method over existing methods. We also show that the proposed method can help several downstream tasks, including semi-supervised part segmentation and data augmentation for shape classification. The code for this work will be publicly available upon the paper's publication.
- 28 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 26 (15209210) , 7775-7786
Unsupervised Point Cloud Co-Part Segmentation via Co-Attended Superpoint Generation and Aggregation
We propose a co-part segmentation method that takes a set of point clouds of the same category as input where neither a ground truth label nor a prior network is required. With difficulties caused by the label absence, we formulate the co-part segmentation task into two subtasks, including superpoint generation and part aggregation. In the first subtask, our superpoint generation network divides each point cloud into homogeneous partitions, each called superpoint, while in the second subtask, these superpoints are further aggregated into a few semantic parts via our part aggregation network. We introduce the coupled attention blocks in the part aggregation network to explicitly enforce semantic consistency in the segmentation by exploiting intra-, inter-, and paired-cloud geometrical information by minimizing the devised intra-, inter-, and paired-cloud losses, respectively. The intra-cloud loss triggers a semantic segmentation in each point cloud, while the inter-cloud loss considers all clouds to enforce their semantic consistency. The paired-cloud loss is designed to ensure that each part of one point cloud can be discriminatively reconstructed from the superpoints of another point cloud. We perform experiments on two benchmark datasets, ShapeNet part and COSEG, and provide quantitative and qualitative results to demonstrate the superiority of our method over existing methods. We also show that the proposed method can help several downstream tasks, including semi-supervised part segmentation and data augmentation for shape classification. The code for this work will be publicly available upon the paper's publication.
- Open Access
- 23 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 51320-51330
FedDAD: Federated Domain Adaptation for Object Detection
Training an object detection model often requires numerous annotated images on a centralized host, which may violate user privacy and data confidentiality. Federated learning (FL) resolves this issue by allowing multiple clients, e.g., cameras, to collaboratively train a model while protecting user privacy. However, models trained with FL may fail to be generalized for new target domain due to domain shift when the data between source and target domains are statistically different. In this work, we formulate a real world object detection problem as a source-free multi-domain adaptation problem in FL architecture. Moreover, we propose an adaptive FL algorithm, called FedDAD (Federated Domain Adaptive Detector), which aggregates models with dynamic attention targeting the unsupervised domain on server, and utilize instance-level alignment to alleviate the effects of scene variation on clients. Experiments are conducted on various datasets and the results show that FedDAD has significant improvements over state-of-the-art FL aggregation algorithms.
- 23 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 51320-51330
Open AccessFedDAD: Federated Domain Adaptation for Object Detection
Training an object detection model often requires numerous annotated images on a centralized host, which may violate user privacy and data confidentiality. Federated learning (FL) resolves this issue by allowing multiple clients, e.g., cameras, to collaboratively train a model while protecting user privacy. However, models trained with FL may fail to be generalized for new target domain due to domain shift when the data between source and target domains are statistically different. In this work, we formulate a real world object detection problem as a source-free multi-domain adaptation problem in FL architecture. Moreover, we propose an adaptive FL algorithm, called FedDAD (Federated Domain Adaptive Detector), which aggregates models with dynamic attention targeting the unsupervised domain on server, and utilize instance-level alignment to alleviate the effects of scene variation on clients. Experiments are conducted on various datasets and the results show that FedDAD has significant improvements over state-of-the-art FL aggregation algorithms.
- 22 October 2022
- Published by Springer Nature
- p. 596-611
Point MixSwap: Attentional Point Cloud Mixing via Swapping Matched Structural Divisions
- 22 October 2022
- Published by Springer Nature
- p. 596-611
Point MixSwap: Attentional Point Cloud Mixing via Swapping Matched Structural Divisions
- Open Access
Visualizing and Alleviating the Effect of Radial Distortion on Camera Calibration Using Principal Lines
Preparing appropriate images for camera calibration is crucial to obtain accurate results. In this paper, new suggestions for preparing such data to alleviate the adverse effect of radial distortion for a calibration procedure using principal lines are developed through the investigations of: (i) identifying directions of checkerboard movements in an image which will result in maximum (and minimum) influence on the calibration results, and (ii) inspecting symmetry and monotonicity of such effect in (i) using the above principal lines. Accordingly, it is suggested that the estimation of principal point should based on linearly independent pairs of nearly parallel principal lines, with a member in each pair corresponds to a near 180-degree rotation (in the image plane) of the other. Experimental results show that more robust and consistent calibration results for the foregoing estimation can actually be obtained, compared with the renowned algebraic methods which estimate distortion parameters explicitly.
Open AccessVisualizing and Alleviating the Effect of Radial Distortion on Camera Calibration Using Principal Lines
Preparing appropriate images for camera calibration is crucial to obtain accurate results. In this paper, new suggestions for preparing such data to alleviate the adverse effect of radial distortion for a calibration procedure using principal lines are developed through the investigations of: (i) identifying directions of checkerboard movements in an image which will result in maximum (and minimum) influence on the calibration results, and (ii) inspecting symmetry and monotonicity of such effect in (i) using the above principal lines. Accordingly, it is suggested that the estimation of principal point should based on linearly independent pairs of nearly parallel principal lines, with a member in each pair corresponds to a near 180-degree rotation (in the image plane) of the other. Experimental results show that more robust and consistent calibration results for the foregoing estimation can actually be obtained, compared with the renowned algebraic methods which estimate distortion parameters explicitly.
- Open Access
- 14 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 8816-8823
Fusion of Multi-Intensity Image for Deep Learning-Based Human and Face Detection
For ordinary IR-illuminators in nighttime surveillance system, insufficient illumination may cause misdetection for faraway object while excessive illumination leads to over-exposure of nearby object. To overcome these two problems, we use the MI3 image dataset, which is established by multi-intensity IR-illumination (MIIR), as our benchmark dataset for modern object detection methods. We first provide complete annotations for the MI3 as its current ground-truth is incomplete. Then, we use these multi-intensity illuminated IR videos to evaluate several widely used object detectors, i.e., SSD, YOLO, Faster R-CNN, and Mask R-CNN, by analyzing the effective range of different illumination intensities. By including a tracking scheme, as well as developing of a new fusion method for different illumination intensities to improve the performance, the proposed approach may serve as a new benchmark of face and object detection for a wide range of distances. The new dataset (Dataset is available: https://ieee-dataport.org/documents/mi3 ) with more complete annotations and source codes (Codes are available: https://github.com/thesuperorange/deepMI3 ) is available online.
- 14 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 8816-8823
Open AccessFusion of Multi-Intensity Image for Deep Learning-Based Human and Face Detection
For ordinary IR-illuminators in nighttime surveillance system, insufficient illumination may cause misdetection for faraway object while excessive illumination leads to over-exposure of nearby object. To overcome these two problems, we use the MI3 image dataset, which is established by multi-intensity IR-illumination (MIIR), as our benchmark dataset for modern object detection methods. We first provide complete annotations for the MI3 as its current ground-truth is incomplete. Then, we use these multi-intensity illuminated IR videos to evaluate several widely used object detectors, i.e., SSD, YOLO, Faster R-CNN, and Mask R-CNN, by analyzing the effective range of different illumination intensities. By including a tracking scheme, as well as developing of a new fusion method for different illumination intensities to improve the performance, the proposed approach may serve as a new benchmark of face and object detection for a wide range of distances. The new dataset (Dataset is available: https://ieee-dataport.org/documents/mi3 ) with more complete annotations and source codes (Codes are available: https://github.com/thesuperorange/deepMI3 ) is available online.