What are Stomata? Stomata (singular stoma) refers to the complex consisting of a pair of specialised cells, called guard cells, surrounding a pore on the surface of above-ground plant organs. Through changes in turgor pressure of the guard cells, the pore aperture adjusts to facilitate, or restrict, the exchange of gases between the plant and the atmosphere. Stomatal traits are generally measured using direct or indirect methods. The former may include analysis of stomatal geometry using microscope-based imagery, whilst the latter often informs stomatal function; encompassing gas exchange, leaf temperature or conductance measurements.
Why are Stomata important? Plants respond to signals such as an increased light intensity and humidity to increase pore aperture and thus open stomata. Whilst this enables the uptake of carbon dioxide required for photosynthesis, water vapour is simultaneously lost. As such, carbon fixation must be balanced against water loss. Whilst stomata occupy less than 5% of the leaf epidermal surface, they account for up to 98% of gas exchange. Therefore, understanding their function is integral to analysing plant physiology, and accurately characterising stomata traits are vitally important for plant improvement programs.
Deep learning for the analysis of stomata Deep learning, a subset of machine learning, uses artificial neural networks to perform a task, where ‘deep’ refers to the use of multiple layers in the network. Deep learning is particularly effective for image analysis, such as the direct measurement of stomatal traits from microscope-based images. Through this, automated measurement can be made on traits such as stomatal counts or density, area measurements and/ or morphology. Deep learning requires an initial annotated dataset for training, and network structure can be adapted according to the requirements of the image analysis task.
What is StomataHub? Founded by Dr Jonathon Gibbs, a post doctorial researcher at the University of Nottingham, StomataHub aims to encourage the collaboration and sharing of data and resources for stomatal analysis. Plant phenotyping is often cited as a bottleneck to improvement programmes due to lengthy timeframes associated with analysis, or lack of automated tools. However with the advent of deep learning technologies, our current analysis capacity is limited by data availability
StomataHub aims to create a centralised location for datasets in a generic format; deploy a web-based detector (Under development as of 4th April 2024) which requires no technical expertise to use; a ‘peoples’ page to help form collaborations and outreach in particular looking to increase the uptake of such methods by biologists, and; serve as a place to act as an introduction to those looking to pursue research in this area.
Stomata hub hosts published datasets to support the creation of deep learning architectures for the analysis of stomata. We hope this will continue to grow as researchers from other institutions reach out to provide their own datasets and form collaborations. The existing datasets hosted here come with annotations ranging from simple points through to detailed segmentation masks. Each is aimed at providing valuable training and testing data for machine learning and computer vision applications, including classification, object localisation, and segmentation. If you have a dataset you would like to share or have any questions about existing datasets please do get in touch here.
All available datasets have been shared with the permission of the authors. Please ensure you include original citations in your work.
12 Datasets | 21,128 Annotated Images | 10,785MBs of data |
Year | Title | Authors |
---|---|---|
2024 | Labeled temperate hardwood tree stomatal image datasets from seven taxa of Popul... | Wang, Jiaxin; Renninger, Heidi J.; Ma, Qin..... |
2023 | Rotating Stomata Measurement Based on Anchor-Free Object Detection and Stomata C... | Zhang, Fan; Wang, Bo; Lu, Fuhao; Zhang, Xinhong..... |
2023 | Automated plant species identification from the stomata images using deep neural... | Dey, Biplob; Ahmed, Romel; Ferdous, Jannatul; Haqu..... |
2023 | Automated estimation of stomatal number and aperture in haskap (Lonicera caerule... | Meng, Xiangji; Nakano, Arisa; Hoshino, Yoichiro..... |
2023 | RotatedStomataNet: a deep rotated object detection network for directional stoma... | Yang, Xiaohui; Wang, Jiahui; Li, Fan; Zhou, Chengl..... |
2023 | Image-Based Quantification of Arabidopsis thaliana Stomatal Aperture from Leaf I... | Takagi, Momoko; Hirata, Rikako; Aihara, Yusuke; Ha..... |
2023 | An automatic plant leaf stoma detection method based on YOLOv5 | Li, Xin; Guo, Siyu; Gong, Linrui; Lan, Yuan..... |
2023 | StomataTracker: Revealing circadian rhythms of wheat stomata with in-situ video ... | Sun, Zhuangzhuang; Wang, Xiao; Song, Yunlin; Li, Q..... |
2023 | Rapid non-destructive method to phenotype stomatal traits | Pathoumthong, Phetdalaphone; Zhang, Zhen; Roy, Stu..... |
2023 | StoManager1: An Enhanced, Automated, and High-throughput Tool to Measure Leaf St... | Wang, Jiaxin; Renninger, Heidi J.; Ma, Qin; Jin, S..... |