Data augmentation in microscopic images for material data mining
Recent progress in material data mining has been driven by high-capacity models trained on large datasets. However, collecting experimental data (real data) has been extremely costly since the amount of human effort and expertise required. Here, we develop a novel transfer learning strategy to addre...
Tallennettuna:
| Päätekijät: | , , , , , , , , , , , , |
|---|---|
| Aineistotyyppi: | Artikkeli |
| Julkaistu: |
2019
|
| Aiheet: | |
| Linkit: | https://arxiv.org/abs/1905.04711 http://arxiv.org/abs/1905.04711 |
| Tagit: |
Ei tageja, Lisää ensimmäinen tagi!
|
| _version_ | 1879172528558047232 |
|---|---|
| author | Ma, Boyuan Wei, Xiaoyan Liu, Chuni Ban, Xiaojuan Huang, Haiyou Wang, Hao Xue, Weihua Wu, Stephen Gao, Mingfei Shen, Qing Abuassba, Adnan Omer Shen, Haokai Su, Yanjing |
| author_browse | Abuassba, Adnan Omer Ban, Xiaojuan Gao, Mingfei Huang, Haiyou Liu, Chuni Ma, Boyuan Shen, Haokai Shen, Qing Su, Yanjing Wang, Hao Wei, Xiaoyan Wu, Stephen Xue, Weihua |
| author_facet | Ma, Boyuan Wei, Xiaoyan Liu, Chuni Ban, Xiaojuan Huang, Haiyou Wang, Hao Xue, Weihua Wu, Stephen Gao, Mingfei Shen, Qing Abuassba, Adnan Omer Shen, Haokai Su, Yanjing |
| author_sort | Ma, Boyuan |
| building | arXiv |
| collection | Acceso abierto |
| description | Recent progress in material data mining has been driven by high-capacity models trained on large datasets. However, collecting experimental data (real data) has been extremely costly since the amount of human effort and expertise required. Here, we develop a novel transfer learning strategy to address small or insufficient data problem. This strategy realizes the fusion of real and simulated data, and the augmentation of training data in data mining procedure. For a specific task of image segmentation, this strategy can generate synthetic images by fusing physical mechanism of simulated images and "image style" of real images. The result shows that the model trained with the acquired synthetic images and 35% of the real images outperforms the model trained on all real images. As the time required to generate synthetic data is almost negligible, this strategy is able to reduce the time cost of real data preparation by roughly 65%. |
| doi_str_mv | 10.1038/s41524-020-00392-6 |
| first_indexed | 2026-10-15T23:42:45Z |
| format | Article |
| fullrecord | <oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ https://www.openarchives.org/OAI/2.0/oai_dc.xsd"><identifier>arxiv-1905.04711</identifier><datestamp>2020-10-01</datestamp>
<dc:title>Data augmentation in microscopic images for material data mining</dc:title>
<dc:creator>Ma, Boyuan</dc:creator>
<dc:creator>Wei, Xiaoyan</dc:creator>
<dc:creator>Liu, Chuni</dc:creator>
<dc:creator>Ban, Xiaojuan</dc:creator>
<dc:creator>Huang, Haiyou</dc:creator>
<dc:creator>Wang, Hao</dc:creator>
<dc:creator>Xue, Weihua</dc:creator>
<dc:creator>Wu, Stephen</dc:creator>
<dc:creator>Gao, Mingfei</dc:creator>
<dc:creator>Shen, Qing</dc:creator>
<dc:creator>Abuassba, Adnan Omer</dc:creator>
<dc:creator>Shen, Haokai</dc:creator>
<dc:creator>Su, Yanjing</dc:creator>
<dc:subject>Materials Science</dc:subject>
<dc:subject>Computer Vision and Pattern Recognition</dc:subject>
<dc:subject>Image and Video Processing</dc:subject>
<dc:description>Recent progress in material data mining has been driven by high-capacity models trained on large datasets. However, collecting experimental data (real data) has been extremely costly since the amount of human effort and expertise required. Here, we develop a novel transfer learning strategy to address small or insufficient data problem. This strategy realizes the fusion of real and simulated data, and the augmentation of training data in data mining procedure. For a specific task of image segmentation, this strategy can generate synthetic images by fusing physical mechanism of simulated images and "image style" of real images. The result shows that the model trained with the acquired synthetic images and 35% of the real images outperforms the model trained on all real images. As the time required to generate synthetic data is almost negligible, this strategy is able to reduce the time cost of real data preparation by roughly 65%. </dc:description>
<dc:description>17 pages, technical report</dc:description>
<dc:date>2019-05-12</dc:date>
<dc:date>2019-10-28</dc:date>
<dc:type>text</dc:type>
<dc:identifier>https://arxiv.org/abs/1905.04711</dc:identifier>
<dc:identifier>http://arxiv.org/abs/1905.04711</dc:identifier>
<dc:identifier>npj computational materials 2020</dc:identifier>
<dc:identifier>doi:10.1038/s41524-020-00392-6</dc:identifier>
</oai_dc:dc> |
| id | arxiv-1905.04711 |
| institution | arXiv |
| last_indexed | 2026-10-15T23:42:45Z |
| publishDate | 2019 |
| publishDateRange | 2019 |
| publishDateSort | 2019 |
| record_format | oai_dc |
| spelling | arxiv-1905.047112020-10-01 Data augmentation in microscopic images for material data mining Ma, Boyuan Wei, Xiaoyan Liu, Chuni Ban, Xiaojuan Huang, Haiyou Wang, Hao Xue, Weihua Wu, Stephen Gao, Mingfei Shen, Qing Abuassba, Adnan Omer Shen, Haokai Su, Yanjing Materials Science Computer Vision and Pattern Recognition Image and Video Processing Recent progress in material data mining has been driven by high-capacity models trained on large datasets. However, collecting experimental data (real data) has been extremely costly since the amount of human effort and expertise required. Here, we develop a novel transfer learning strategy to address small or insufficient data problem. This strategy realizes the fusion of real and simulated data, and the augmentation of training data in data mining procedure. For a specific task of image segmentation, this strategy can generate synthetic images by fusing physical mechanism of simulated images and "image style" of real images. The result shows that the model trained with the acquired synthetic images and 35% of the real images outperforms the model trained on all real images. As the time required to generate synthetic data is almost negligible, this strategy is able to reduce the time cost of real data preparation by roughly 65%. 17 pages, technical report 2019-05-12 2019-10-28 text https://arxiv.org/abs/1905.04711 http://arxiv.org/abs/1905.04711 npj computational materials 2020 doi:10.1038/s41524-020-00392-6 |
| spellingShingle | Materials Science Computer Vision and Pattern Recognition Image and Video Processing Ma, Boyuan Wei, Xiaoyan Liu, Chuni Ban, Xiaojuan Huang, Haiyou Wang, Hao Xue, Weihua Wu, Stephen Gao, Mingfei Shen, Qing Abuassba, Adnan Omer Shen, Haokai Su, Yanjing Data augmentation in microscopic images for material data mining |
| title | Data augmentation in microscopic images for material data mining |
| title_full | Data augmentation in microscopic images for material data mining |
| title_fullStr | Data augmentation in microscopic images for material data mining |
| title_full_unstemmed | Data augmentation in microscopic images for material data mining |
| title_short | Data augmentation in microscopic images for material data mining |
| title_sort | data augmentation in microscopic images for material data mining |
| topic | Materials Science Computer Vision and Pattern Recognition Image and Video Processing |
| url | https://arxiv.org/abs/1905.04711 http://arxiv.org/abs/1905.04711 |
| work_keys_str_mv | AT maboyuan dataaugmentationinmicroscopicimagesformaterialdatamining AT weixiaoyan dataaugmentationinmicroscopicimagesformaterialdatamining AT liuchuni dataaugmentationinmicroscopicimagesformaterialdatamining AT banxiaojuan dataaugmentationinmicroscopicimagesformaterialdatamining AT huanghaiyou dataaugmentationinmicroscopicimagesformaterialdatamining AT wanghao dataaugmentationinmicroscopicimagesformaterialdatamining AT xueweihua dataaugmentationinmicroscopicimagesformaterialdatamining AT wustephen dataaugmentationinmicroscopicimagesformaterialdatamining AT gaomingfei dataaugmentationinmicroscopicimagesformaterialdatamining AT shenqing dataaugmentationinmicroscopicimagesformaterialdatamining AT abuassbaadnanomer dataaugmentationinmicroscopicimagesformaterialdatamining AT shenhaokai dataaugmentationinmicroscopicimagesformaterialdatamining AT suyanjing dataaugmentationinmicroscopicimagesformaterialdatamining |