Masato Shirai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Masato Shirai has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Amjad Khan
Department of Plant Sciences; Quaid-i-Azam University; Islamabad 45320; Pakistan
1 shared publication - Atsuko Takano
Institute of Natural & Environmental Sciences, University of Hyogo, Sanda, Japan
1 shared publication - Chan-Ho Park
National Institute of Biological Resources, Incheon, Korea
1 shared publication - Chung-Kun Lee
Department of Botany, Graduate School of Science, Kyoto University , Kitashirakawa-oiwake-cho, Sakyo-ku, Kyoto , Japan
1 shared publication - Diego Tavares Vasques
Botanical Gardens, Graduate School of Science The University of Tokyo Tokyo Japan
1 shared publication Hayato Ogino
1 shared publication
Latest Publications
- Open Access
- 26 August 2026
- Published by Springer Nature in Journal of Plant Research
- p. 1-17
Herbarium specimens in the age of artificial intelligence: from herbarium image identification to Integrative Taxonomic AI
Herbarium specimens are physical, verifiable records that form the basis of taxonomic knowledge and biodiversity research. Their large-scale digitization has produced extensive collections of high-resolution images and associated specimen metadata, creating conditions in which artificial intelligence (AI) can play an important role in plant taxonomy, collection management, and ecological research. Early AI applications have primarily focused on automated species identification based on individual specimen images. Although increasingly accurate, such approaches remain limited by their emphasis on single-specimen label prediction and by treating identification outputs as final analytical decisions. Recent methodological advances—including segmentation-based preprocessing, automated trait extraction, structured extraction of label data, detection of potentially misidentified specimens, and multimodal integration of visual, textual, and genetic information—extend AI applications beyond species identification toward broader analytical frameworks, encompassing taxonomic interpretation as well as ecological and biodiversity research. In these approaches, specimens are placed within a shared analytical space, and identification results are used to support comparisons across multiple specimens rather than being treated as final decisions for single individuals. This multi-specimen perspective enables quantitative examination of species boundaries, morphological variation, data inconsistencies, and taxonomic stability within curated collections. In this context, AI serves not as an ultimate decision-maker but as a decision-support tool embedded in expert-guided workflows and biodiversity knowledge infrastructures. These developments can be summarized as Integrative Taxonomic AI , an approach that employs learned morphospaces to interpret and refine taxonomic categories by integrating multimodal evidence and curated specimen data under expert guidance.
- 26 August 2026
- Published by Springer Nature in Journal of Plant Research
- p. 1-17
Open AccessHerbarium specimens in the age of artificial intelligence: from herbarium image identification to Integrative Taxonomic AI
Herbarium specimens are physical, verifiable records that form the basis of taxonomic knowledge and biodiversity research. Their large-scale digitization has produced extensive collections of high-resolution images and associated specimen metadata, creating conditions in which artificial intelligence (AI) can play an important role in plant taxonomy, collection management, and ecological research. Early AI applications have primarily focused on automated species identification based on individual specimen images. Although increasingly accurate, such approaches remain limited by their emphasis on single-specimen label prediction and by treating identification outputs as final analytical decisions. Recent methodological advances—including segmentation-based preprocessing, automated trait extraction, structured extraction of label data, detection of potentially misidentified specimens, and multimodal integration of visual, textual, and genetic information—extend AI applications beyond species identification toward broader analytical frameworks, encompassing taxonomic interpretation as well as ecological and biodiversity research. In these approaches, specimens are placed within a shared analytical space, and identification results are used to support comparisons across multiple specimens rather than being treated as final decisions for single individuals. This multi-specimen perspective enables quantitative examination of species boundaries, morphological variation, data inconsistencies, and taxonomic stability within curated collections. In this context, AI serves not as an ultimate decision-maker but as a decision-support tool embedded in expert-guided workflows and biodiversity knowledge infrastructures. These developments can be summarized as Integrative Taxonomic AI , an approach that employs learned morphospaces to interpret and refine taxonomic categories by integrating multimodal evidence and curated specimen data under expert guidance.
- Open Access
- 21 August 2026
- Published by Elsevier in Materials Today Communications
- Vol. 56
Deep learning-based grain boundary detection in polycrystalline metals via self-supervised boundary restoration
- 21 August 2026
- Published by Elsevier in Materials Today Communications
- Vol. 56
Open AccessDeep learning-based grain boundary detection in polycrystalline metals via self-supervised boundary restoration
- Open Access
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 178876-178892
SWSformer: Subwindow Shuffle Transformer for Image Restoration
Transformers have demonstrated superior performance over conventional methods in various tasks due to their ability to capture long-range dependencies and adaptively generate weights. However, their computational complexity increases quadratically with the number of tokens, limiting their applicability to high-resolution image tasks. To address this problem, recent image restoration methods have attempted to mitigate this limitation by adopting mechanisms that reduce computational demands at the expense of fully capturing long-range dependencies. In contrast, the window shuffle self-attention (WS-SA) mechanism of the Shuffle Transformer reduces computational complexity without significantly limiting the ability to capture long-range dependencies. Nevertheless, WS-SA suffers from generating excessively sparse self-attention maps with a limited receptive field when applied to high-resolution images. In this work, we propose subwindow shuffle self-attention (SWS-SA), a mechanism that expands the receptive field without increasing the computational complexity of WS-SA. SWS-SA introduces a subwindow-based spatial shuffle to enhance the receptive field. Additionally, we apply average pooling to the query embeddings to further reduce computational complexity. Furthermore, we present the Subwindow Shuffle Transformer (SWSformer), which employs SWS-SA as its core component and integrates effective techniques from related works. To evaluate the performance of SWSformer, we conduct experiments on image denoising, deblurring, and deraining tasks. The experimental results demonstrate that SWSformer achieves state-of-the-art performance. Moreover, we perform comprehensive ablation studies to identify the contributions of individual components and settings to the model’s overall effectiveness.
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 178876-178892
Open AccessSWSformer: Subwindow Shuffle Transformer for Image Restoration
Transformers have demonstrated superior performance over conventional methods in various tasks due to their ability to capture long-range dependencies and adaptively generate weights. However, their computational complexity increases quadratically with the number of tokens, limiting their applicability to high-resolution image tasks. To address this problem, recent image restoration methods have attempted to mitigate this limitation by adopting mechanisms that reduce computational demands at the expense of fully capturing long-range dependencies. In contrast, the window shuffle self-attention (WS-SA) mechanism of the Shuffle Transformer reduces computational complexity without significantly limiting the ability to capture long-range dependencies. Nevertheless, WS-SA suffers from generating excessively sparse self-attention maps with a limited receptive field when applied to high-resolution images. In this work, we propose subwindow shuffle self-attention (SWS-SA), a mechanism that expands the receptive field without increasing the computational complexity of WS-SA. SWS-SA introduces a subwindow-based spatial shuffle to enhance the receptive field. Additionally, we apply average pooling to the query embeddings to further reduce computational complexity. Furthermore, we present the Subwindow Shuffle Transformer (SWSformer), which employs SWS-SA as its core component and integrates effective techniques from related works. To evaluate the performance of SWSformer, we conduct experiments on image denoising, deblurring, and deraining tasks. The experimental results demonstrate that SWSformer achieves state-of-the-art performance. Moreover, we perform comprehensive ablation studies to identify the contributions of individual components and settings to the model’s overall effectiveness.