Wen-Chieh Fang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Wen-Chieh Fang has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Chih-Yi Chiu
National Chiayi University
1 shared publication - Chih-Yi Chiu
Department of Computer Science and Information Engineering National Chiayi University Taiwan, R.O.C
1 shared publication - Hsien-Chang Wu
School of Post-Baccalaureate Chinese Medicine, Tzu Chi University, Hualien 970374, Taiwan
1 shared publication - Jiann-Yeou Rau
Department of Geomatics, National Cheng Kung University, Tainan, Taiwan
1 shared publication - Jiao-Chuan Huang
Department of Computer Science and Information Engineering, National Chiayi University, Chiayi City, Taiwan
1 shared publication - Jui-Lin Wang
China Engineering Consultants, Inc., Floor 28, No.185, Section 2, Sinhai Road, Da-an District, Taipei City 106, Taiwan
1 shared publication
Latest Publications
- 6 December 2023
- Published by Springer Nature in Machine Vision and Applications
- Vol. 35 (1) , 1-16
Pixel representations, sampling, and label correction for semantic part detection
- 6 December 2023
- Published by Springer Nature in Machine Vision and Applications
- Vol. 35 (1) , 1-16
Pixel representations, sampling, and label correction for semantic part detection
- 4 August 2023
- Published by Springer Nature in Neural Computing and Applications
- Vol. 35 (28) , 21259-21274
TongueMobile: automated tongue segmentation and diagnosis on smartphones
- 4 August 2023
- Published by Springer Nature in Neural Computing and Applications
- Vol. 35 (28) , 21259-21274
TongueMobile: automated tongue segmentation and diagnosis on smartphones
- 6 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 9 (5) , 1394-1405
Attention on Global–Local Representation Spaces in Recommender Systems
In this study, we present a novel clustering-based collaborative filtering (CF) method for recommender systems. Clustering-based CF methods can effectively deal with data sparsity and scalability problems. However, most of them are applied to a single representation space, which might not characterize complex user–item interactions well. We argue that the user–item interactions should be observed from multiple views and characterized in an adaptive way. To address this issue, we leveraged the global and local properties to construct multiple representation spaces by learning various training datasets and loss functions. An attention network was built to generate a blended representation according to the relative importance of the representation spaces for each user–item pair, providing a flexible way to characterize diverse user–item interactions. Substantial experiments were evaluated on four popular benchmark datasets. The results show that the proposed method is superior to several CF methods where only one representation space is considered.
- 6 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 9 (5) , 1394-1405
Attention on Global–Local Representation Spaces in Recommender Systems
In this study, we present a novel clustering-based collaborative filtering (CF) method for recommender systems. Clustering-based CF methods can effectively deal with data sparsity and scalability problems. However, most of them are applied to a single representation space, which might not characterize complex user–item interactions well. We argue that the user–item interactions should be observed from multiple views and characterized in an adaptive way. To address this issue, we leveraged the global and local properties to construct multiple representation spaces by learning various training datasets and loss functions. An attention network was built to generate a blended representation according to the relative importance of the representation spaces for each user–item pair, providing a flexible way to characterize diverse user–item interactions. Substantial experiments were evaluated on four popular benchmark datasets. The results show that the proposed method is superior to several CF methods where only one representation space is considered.
- Open Access
Attention on Global-Local Representation Spaces in Recommender Systems
In this study, we present a novel clustering-based collaborative filtering (CF) method for recommender systems. Clustering-based CF methods can effectively deal with data sparsity and scalability problems. However, most of them are applied to a single representation space, which might not characterize complex user-item interactions well. We argue that the user-item interactions should be observed from multiple views and characterized in an adaptive way. To address this issue, we leveraged the global and local properties to construct multiple representation spaces by learning various training datasets and loss functions. An attention network was built to generate a blended representation according to the relative importance of the representation spaces for each user-item pair, providing a flexible way to characterize diverse user-item interactions. Substantial experiments were evaluated on four popular benchmark datasets. The results show that the proposed method is superior to several CF methods where only one representation space is considered.
Open AccessAttention on Global-Local Representation Spaces in Recommender Systems
In this study, we present a novel clustering-based collaborative filtering (CF) method for recommender systems. Clustering-based CF methods can effectively deal with data sparsity and scalability problems. However, most of them are applied to a single representation space, which might not characterize complex user-item interactions well. We argue that the user-item interactions should be observed from multiple views and characterized in an adaptive way. To address this issue, we leveraged the global and local properties to construct multiple representation spaces by learning various training datasets and loss functions. An attention network was built to generate a blended representation according to the relative importance of the representation spaces for each user-item pair, providing a flexible way to characterize diverse user-item interactions. Substantial experiments were evaluated on four popular benchmark datasets. The results show that the proposed method is superior to several CF methods where only one representation space is considered.
- Open Access
- 24 August 2017
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLII-2/W6, 311-318
BRIDGE CRACK DETECTION USING MULTI-ROTARY UAV AND OBJECT-BASE IMAGE ANALYSIS
Bridge is an important infrastructure for human life. Thus, the bridge safety monitoring and maintaining is an important issue to the government. Conventionally, bridge inspection were conducted by human in-situ visual examination. This procedure sometimes require under bridge inspection vehicle or climbing under the bridge personally. Thus, its cost and risk is high as well as labor intensive and time consuming. Particularly, its documentation procedure is subjective without 3D spatial information. In order cope with these challenges, this paper propose the use of a multi-rotary UAV that equipped with a SONY A7r2 high resolution digital camera, 50 mm fixed focus length lens, 135 degrees up-down rotating gimbal. The target bridge contains three spans with a total of 60 meters long, 20 meters width and 8 meters height above the water level. In the end, we took about 10,000 images, but some of them were acquired by hand held method taken on the ground using a pole with 2–8 meters long. Those images were processed by Agisoft PhotoscanPro to obtain exterior and interior orientation parameters. A local coordinate system was defined by using 12 ground control points measured by a total station. After triangulation and camera self-calibration, the RMS of control points is less than 3 cm. A 3D CAD model that describe the bridge surface geometry was manually measured by PhotoscanPro. They were composed of planar polygons and will be used for searching related UAV images. Additionally, a photorealistic 3D model can be produced for 3D visualization. In order to detect cracks on the bridge surface, we utilize object-based image analysis (OBIA) technique to segment the image into objects. Later, we derive several object features, such as density, area/bounding box ratio, length/width ratio, length, etc. Then, we can setup a classification rule set to distinguish cracks. Further, we apply semi-global-matching (SGM) to obtain 3D crack information and based on image scale we can calculate the width of a crack object. For spalling volume calculation, we also apply SGM to obtain dense surface geometry. Assuming the background is a planar surface, we can fit a planar function and convert the surface geometry into a DSM. Thus, for spalling area its height will be lower than the plane and its value will be negative. We can thus apply several image processing technique to segment the spalling area and calculate the spalling volume as well. For bridge inspection and UAV image management within a laboratory, we develop a graphic user interface. The major functions include crack auto-detection using OBIA, crack editing, i.e. delete and add cracks, crack attributing, 3D crack visualization, spalling area/volume calculation, bridge defects documentation, etc.
- 24 August 2017
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLII-2/W6, 311-318
Open AccessBRIDGE CRACK DETECTION USING MULTI-ROTARY UAV AND OBJECT-BASE IMAGE ANALYSIS
Bridge is an important infrastructure for human life. Thus, the bridge safety monitoring and maintaining is an important issue to the government. Conventionally, bridge inspection were conducted by human in-situ visual examination. This procedure sometimes require under bridge inspection vehicle or climbing under the bridge personally. Thus, its cost and risk is high as well as labor intensive and time consuming. Particularly, its documentation procedure is subjective without 3D spatial information. In order cope with these challenges, this paper propose the use of a multi-rotary UAV that equipped with a SONY A7r2 high resolution digital camera, 50 mm fixed focus length lens, 135 degrees up-down rotating gimbal. The target bridge contains three spans with a total of 60 meters long, 20 meters width and 8 meters height above the water level. In the end, we took about 10,000 images, but some of them were acquired by hand held method taken on the ground using a pole with 2–8 meters long. Those images were processed by Agisoft PhotoscanPro to obtain exterior and interior orientation parameters. A local coordinate system was defined by using 12 ground control points measured by a total station. After triangulation and camera self-calibration, the RMS of control points is less than 3 cm. A 3D CAD model that describe the bridge surface geometry was manually measured by PhotoscanPro. They were composed of planar polygons and will be used for searching related UAV images. Additionally, a photorealistic 3D model can be produced for 3D visualization. In order to detect cracks on the bridge surface, we utilize object-based image analysis (OBIA) technique to segment the image into objects. Later, we derive several object features, such as density, area/bounding box ratio, length/width ratio, length, etc. Then, we can setup a classification rule set to distinguish cracks. Further, we apply semi-global-matching (SGM) to obtain 3D crack information and based on image scale we can calculate the width of a crack object. For spalling volume calculation, we also apply SGM to obtain dense surface geometry. Assuming the background is a planar surface, we can fit a planar function and convert the surface geometry into a DSM. Thus, for spalling area its height will be lower than the plane and its value will be negative. We can thus apply several image processing technique to segment the spalling area and calculate the spalling volume as well. For bridge inspection and UAV image management within a laboratory, we develop a graphic user interface. The major functions include crack auto-detection using OBIA, crack editing, i.e. delete and add cracks, crack attributing, 3D crack visualization, spalling area/volume calculation, bridge defects documentation, etc.
- 1 May 2015
- Published by Springer Nature
- p. 683-691
iDianNao: Recommending Volunteer Opportunities to Older Adults
- 1 May 2015
- Published by Springer Nature
- p. 683-691
iDianNao: Recommending Volunteer Opportunities to Older Adults