Detail of Publication

Text Language	English
Authors	Naoki Yagihara, Yuzuko Utsumi, Daisuke Yamauchi, Yoshinobu Mineyuki, Ichirou Karahara, Masato Hoshino, Kentaro Uesugi, Yoshio Suzuki, Akihisa Takeuchi, and Masakazu Iwamura1
Title	Instance Segmentation of Densely Packed Plant Cells under Sparse Annotation in X-ray ?CT
Journal	ECCV 2026 Workshop on Computer Vision in Plant Phenotyping and Agriculture (CVPPA)
Location	Malmö, Sweden
Reviewed or not	Reviewed
Presentation type	Poster
Month & Year	September 2026
Abstract	Unstained X-ray ?CTcancapture plant tissue non-destructively at the single-cell level. However, its cell-wall contrast is low, and the cells are densely packed. These conditions make cell instance segmentation difficult. We build on the flow-based method cpsam and introduce two components: z-consensus boundary recovery (ZBR), which strengthens weak boundaries by combining information across z-slices and synthetic training data, which improves label efficiency. Together, these compo nents improve segmentation accuracy under sparse annotation.

• Following file is available.
1. CVPPA_2026_yagihara_CameraReady3.pdf (9.1MB)
• Entry for BibTeX

  @InProceedings{Yagihara2026,
    author =	{Naoki Yagihara and Yuzuko Utsumi and Daisuke Yamauchi and Yoshinobu Mineyuki and Ichirou Karahara and Masato Hoshino and Kentaro Uesugi and Yoshio Suzuki and Akihisa Takeuchi and Masakazu Iwamura1},
    title =	{Instance Segmentation of Densely Packed Plant Cells under Sparse Annotation in X-ray ?CT},
    booktitle =	{ECCV 2026 Workshop on Computer Vision in Plant Phenotyping and Agriculture (CVPPA)},
    year =	2026,
    month =	sep,
    location =	{Malmö, Sweden}
  }
  
  
  

Back to list