Table 3.

Comparison of the proposed method and output compared to other recently published methods.

Method	Overview	Output
Proposed method	A convolutional neural network based on semantic segmentation and image processing tool for morphometric calculations of stomata plus the automatic estimation of gsmax Applied to Poplar and Wheat	Pixelwise detection Count Density Pore measurements Guard cell measurements gsmax estimate
Toda et al., 2018 DeepStomata	Developed software comprising histogram of gradients (HOG) detection of stomata followed by region classification by a CNN. Used for stomatal pore quantification. Applied to Dayflower	Pixelwise detection Count Density Classification between open and closed stomata Pore measurements
Bhugra et al., 2019	Detects and quantifies stomata using a CNN and a series of image processing techniques Applied to Rice using scanning electron microscopy (SEM) images	Bounding box detection Count Density
Fetter et al., 2019 StomataCounter	A CNN for counting stomata, which detects bounding boxes that encapsulate the stomata Applied to Ginkgo and Poplar	Bounding box detection Stomata count Density
Andayani et al., 2020	Uses a CNN and image processing for classifying stomata into one of two groups belonging to either turmeric or ginger	Classification
Casado-García et al., 2020 LabelStoma	Use YOLO (Redmon and Farhadi, 2018) to detect bounding boxes Applied to Common Bean, Barley, and Soybean	Bounding box detection Stomata count Density
Kwong et al., 2021	A CNN applied specifically towards detecting stomata from Oil Palm	Bounding box detection Count Density
Toda et al., 2021	A platform that supports real time stomata detection when directly connected to a microscope Applied to Wheat- N.B. measurements of bounding boxes allow morphometric calculations of stomata when orientated parallel or perpendicular to the field of view	Bounding box detection Count Density Bounding box measures
Zhu et al., 2021	Applies R-CNN, U-Net, and image processing to calculate stomatal index Applied to Wheat	Bounding box detection Counts of stomata and epidermal cells Stomatal index calculation
Gómez-de-Mariscal et al., 2021 DeepImageJ	A plugin for the widely used ImageJ application. Brings a sophisticated method for integrating deep learning with ImageJ. A user friendly interface which supports a wide range of phenotyping tasks	Dependent on the network but also on the user for defining and selecting the best choice for their needs. Will give detection and possible measurements but no automatic calculation of indices without an additional step