Detection and annotation of plant organs from digitised herbarium scans using deep learning

Abstract

As herbarium specimens are increasingly becoming digitised and accessible in online repositories, advanced computer vision techniques are being used to extract information from them. The presence of certain plant organs on herbarium sheets is useful information in various scientific contexts and automatic recognition of these organs will help mobilise such information. In our study, we use deep learning to detect plant organs on digitised herbarium specimens with Faster R-CNN. For our experiment, we manually annotated hundreds of herbarium scans with thousands of bounding boxes for six types of plant organs and used them for training and evaluating the plant organ detection model. The model worked particularly well on leaves and stems, while flowers were also present in large numbers in the sheets, but were not equally well recognised.

Plant phenotyping relevance

深層学習による植物器官の画像検出モデルを開発し、注釈データで訓練・評価しており、植物形態の取得手法が研究の中心である。

Code and data availability

The paper describes paper-specific public assets: annotated herbarium scan datasets (PANGAEA), author code and trained model on GitHub (Younis 2020), and supplementary annotation files. However, none of these have URLs matching the single allowed URL (the Index Herbariorum 'World's Herbaria 2019' report, which is a cit

No evidence-backed public reproduction asset is currently recorded.