Search

Your search keyword '"Zhu, Kaiyue"' showing total 178 results
Start Over Author "Zhu, Kaiyue"
178 results on '"Zhu, Kaiyue"'

2. Semi-supervised Domain Adaptation for Semantic Segmentation

Author

Chen, Ying, Ouyang, Xu, Zhu, Kaiyue, and Agam, Gady

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Deep learning approaches for semantic segmentation rely primarily on supervised learning approaches and require substantial efforts in producing pixel-level annotations. Further, such approaches may perform poorly when applied to unseen image domains. To cope with these limitations, both unsupervised domain adaptation (UDA) with full source supervision but without target supervision and semi-supervised learning (SSL) with partial supervision have been proposed. While such methods are effective at aligning different feature distributions, there is still a need to efficiently exploit unlabeled data to address the performance gap with respect to fully-supervised methods. In this paper we address semi-supervised domain adaptation (SSDA) for semantic segmentation, where a large amount of labeled source data as well as a small amount of labeled target data are available. We propose a novel and effective two-step semi-supervised dual-domain adaptation (SSDDA) approach to address both cross- and intra-domain gaps in semantic segmentation. The proposed framework is comprised of two mixing modules. First, we conduct a cross-domain adaptation via an image-level mixing strategy, which learns to align the distribution shift of features between the source data and target data. Second, intra-domain adaptation is achieved using a separate student-teacher network which is built to generate category-level data augmentation by mixing unlabeled target data in a way that respects predicted object boundaries. We demonstrate that the proposed approach outperforms state-of-the-art methods on two common synthetic-to-real semantic segmentation benchmarks. An extensive ablation study is provided to further validate the effectiveness of our approach.

Published
2021

4. Mask-based Data Augmentation for Semi-supervised Semantic Segmentation

Author

Chen, Ying, Ouyang, Xu, Zhu, Kaiyue, and Agam, Gady

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Semantic segmentation using convolutional neural networks (CNN) is a crucial component in image analysis. Training a CNN to perform semantic segmentation requires a large amount of labeled data, where the production of such labeled data is both costly and labor intensive. Semi-supervised learning algorithms address this issue by utilizing unlabeled data and so reduce the amount of labeled data needed for training. In particular, data augmentation techniques such as CutMix and ClassMix generate additional training data from existing labeled data. In this paper we propose a new approach for data augmentation, termed ComplexMix, which incorporates aspects of CutMix and ClassMix with improved performance. The proposed approach has the ability to control the complexity of the augmented data while attempting to be semantically-correct and address the tradeoff between complexity and correctness. The proposed ComplexMix approach is evaluated on a standard dataset for semantic segmentation and compared to other state-of-the-art techniques. Experimental results show that our method yields improvement over state-of-the-art methods on standard datasets for semantic image segmentation.

Published
2021

5. Domain Adaptation on Semantic Segmentation for Aerial Images

Author

Chen, Ying, Ouyang, Xu, Zhu, Kaiyue, and Agam, Gady

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Semantic segmentation has achieved significant advances in recent years. While deep neural networks perform semantic segmentation well, their success rely on pixel level supervision which is expensive and time-consuming. Further, training using data from one domain may not generalize well to data from a new domain due to a domain gap between data distributions in the different domains. This domain gap is particularly evident in aerial images where visual appearance depends on the type of environment imaged, season, weather, and time of day when the environment is imaged. Subsequently, this distribution gap leads to severe accuracy loss when using a pretrained segmentation model to analyze new data with different characteristics. In this paper, we propose a novel unsupervised domain adaptation framework to address domain shift in the context of aerial semantic image segmentation. To this end, we solve the problem of domain shift by learn the soft label distribution difference between the source and target domains. Further, we also apply entropy minimization on the target domain to produce high-confident prediction rather than using high-confident prediction by pseudo-labeling. We demonstrate the effectiveness of our domain adaptation framework using the challenge image segmentation dataset of ISPRS, and show improvement over state-of-the-art methods in terms of various metrics.

Published
2020

22. A Novel Atomic-Level Post-Etch-Surface-Reinforcement Process for High-Performance p-GaN Gate HEMTs Fabrication

Author

Wang, Luyu, primary, Zhang, Penghao, additional, Zhu, Kaiyue, additional, Wang, Qiang, additional, Pan, Maolin, additional, Sun, Xin, additional, Huang, Ziqiang, additional, Chen, Kun, additional, Yang, Yannan, additional, Xie, Xinling, additional, Huang, Hai, additional, Hu, Xin, additional, Xu, Saisheng, additional, Wu, Chunlei, additional, Wang, Chen, additional, Xu, Min, additional, and Zhang, David Wei, additional

Published
2023
Full Text
View/download PDF

23. Non-Buffer Epi-AlGaN/GaN on SiC for High-Performance Depletion-Mode MIS-HEMTs Fabrication

Author

Zhang, Penghao, primary, Wang, Luyu, additional, Zhu, Kaiyue, additional, Wang, Qiang, additional, Pan, Maolin, additional, Huang, Ziqiang, additional, Yang, Yannan, additional, Xie, Xinling, additional, Huang, Hai, additional, Hu, Xin, additional, Xu, Saisheng, additional, Xu, Min, additional, Wang, Chen, additional, Wu, Chunlei, additional, and Zhang, David Wei, additional

Published
2023
Full Text
View/download PDF

27. Critical Issues of Vanadium‐Based Cathodes Towards Practical Aqueous Zn‐Ion Batteries.

Author

Jiang, Weikang, Zhu, Kaiyue, and Yang, Weishen

Subjects
*CATHODES, *ENERGY density, *AQUEOUS electrolytes, *PHASE transitions, *STORAGE batteries, *FRIENDSHIP
Abstract

Aqueous zinc‐ion batteries (ZIBs) are gaining significant attention for their numerous advantages, including high safety, high energy density, affordability, and environmental friendliness. However, the development of ZIBs has been hampered by the lack of suitable cathode materials that can store Zn2+ with high capacity and reversibility. Currently, vanadium‐based materials with tunnel or layered structures are widely researched owing to their high theoretical capacity and diversified structures. However, their long‐term cycling stability is unsatisfactory because of material dissolution, phase transformation, and restrictive kinetics in aqueous electrolytes, which limits their practical applications. Different from previous reviews on ZIBs, this review specifically addresses the critical issues faced by vanadium‐based cathodes for practical aqueous ZIBs and proposes potential solutions. Focusing on vanadium‐based cathodes, their ion storage mechanisms, the critical parameters affecting their performance, and the progress made in addressing the aforementioned problems are also summarized. Finally, future directions for the development of practical aqueous ZIB are suggested. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

41. Effects of Different Modified Starches on the Quality of Mochi Bread.

Author

ZHU Kaiyue, JIANG Caiyan, WEN Chengrong, SHANG Shan, TONG Yi, BAI Yushi, and JIANG Pengfei

Subjects
BREAD, BREAD quality, RHEOLOGY, STARCH
Abstract

In order to explore and optimize the formula of mochi bread, the effects of the different compounding ratios of the pregelatinized acetylated distarch adipate to hydroxypropyl distarch phosphate, acetylated distarch or phosphate distarch on the rheological properties and microstructure, as well as the specific volume, sensory score, and textural properties of mochi bread, were investigated. The results showed that with the increase of the pregelatinized starch content, the storage modulus and loss modulus of the mochi bread increased, with the maximum initial moduli exceeding 12 500 Pa and 4 000 Pa, respectively. In the hydroxypropyl modified starch and phosphate modified starch groups, the increase of the addition amount of the pregelatinized starch increased the hardness of mochi bread to 7 571.47 g and 6 017.17 g, respectively, and raised the chewiness to 2 435.48 and 2 840.05, respectively, whilst decreasing the specific volume gradually, and making the microstructure clearer. In the acetylated modified starch group, the increase of the content of pregelatinized starch made the hardness and chewiness of mochi bread decreased firstly then increased, whilst the specific volume first increased then decreased. When the pregelatinized starch content was 20%, the sensory score of mochi bread was the highest (reaching 86.25).The results of this study can provide more raw material source options for mochi bread and lays a theoretical basis for its processing. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

50. Studies on the Deodorization Method of Tilapia.

Author

Jiang Pengfei, Chen Yao, Jin Wengang, Zhang Yuying, Huang Yizhen, Zhu Kaiyue, and Dong Xiuping

Subjects
DEODORIZATION, ION mobility spectroscopy, FLAVOR, ELECTRONIC noses, WINE flavor & odor, TILAPIA, DIMETHYL sulfide, VINEGAR
Abstract

Purpose: To determine the best deodorization process by comparing the deodorization effects of different treatment methods on tilapia meat. Methods: The sensory evaluation method of sorting was adopted, combined with the electronic nose and gas chromatography ion mobility spectroscopy (GC-IMS) data to select the best tilapia deodorization method. Results: The sensory analysis of tilapia meat treated with cooking wine compound method had the best sensory effect. Electronic nose analysis masking method, cooking wine compound method, and embedding method had significant changes in methyl and inorganic sulfide compounds in fish meat after deodorizing treatment. GC-IMS analysis identified 42 kinds of volatile compounds. Fish meat processed by the tomato compound method, fermentation method, masking method, and cooking wine compound method all formed their characteristic flavor regions on the fingerprint. The fish meat of the embedding method and the blank group had similar fingerprints. In the samples treated by tomato compound method and fermentation method, the relative contents of aldehydes that cause fishy smell, ketones and alcohols that enhance fishy smell were relatively high. The content of low molecular carbon chain aldehydes in tilapia meat after deodorization by masking method and cooking and wine complex method was the least, and the deodorization effect was the best. The results showed that in the tilapia meat treated with cooking wine, white vinegar and ginger juice, the relative content of aldehydes in the sample was reduced due to the introduction of flavor substances by ginger juice and cooking wine and the absorption of white vinegar to reduce fishy substances, and the best deodorizing effect. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results