Dataset for Vascular Bundle Detection in Moso Bamboo

© 2026 Lukas Speichinger, Karlsruhe Institute of Technology (KIT), Institue of Engineering Mechanics, Chair of Continuum Mechanics

Description

This repository provides 62 annotated images in the YOLO‑txt format for training a detection model that identifies vascular bundles in the cross-section of Moso bamboo (Phyllostachys pubescens) internodes. The images show radially orientated realizations of the culm wall's mesostrucutre taken by a digital microscop with a 2MP sensor. Each original image was cropped so that its horizontal borders coincide with the inner and outer bounds of the culm wall. Three bundle types are present, VB-I, VB-II and VB-III, which differ by fiber bundle number. Bounding boxes and class labels were assigned manually based on the study of [Xu et al.(2021) ].

Citation

We encourage users to cite this dataset as recommended in the CITATION.cff file as well as the corresponding publication 10.1016/j.conbuildmat.2026.146522

Aknowlegments

The data presented in this repository were generated as part of the author's PhD thesis at the Karlsruhe Institute of Technology (KIT), Institute of Engineering Mechanics, Chair of Continuum Mechanics, in collaboration with the Berliner Hochschule für Technik (BHT).