Automated Machine Learning for High-Throughput Image-Based Plant Phenotyping

Joshua C.O Koh,Surya Kant,German Spangenberg

doi:10.3390/rs13050858

Joshua C.O Koh, Surya Kant + Show 1 more

Open Access

https://doi.org/10.3390/rs13050858

Copy DOI

Abstract

Automated machine learning (AutoML) has been heralded as the next wave in artificial intelligence with its promise to deliver high-performance end-to-end machine learning pipelines with minimal effort from the user. However, despite AutoML showing great promise for computer vision tasks, to the best of our knowledge, no study has used AutoML for image-based plant phenotyping. To address this gap in knowledge, we examined the application of AutoML for image-based plant phenotyping using wheat lodging assessment with unmanned aerial vehicle (UAV) imagery as an example. The performance of an open-source AutoML framework, AutoKeras, in image classification and regression tasks was compared to transfer learning using modern convolutional neural network (CNN) architectures. For image classification, which classified plot images as lodged or non-lodged, transfer learning with Xception and DenseNet-201 achieved the best classification accuracy of 93.2%, whereas AutoKeras had a 92.4% accuracy. For image regression, which predicted lodging scores from plot images, transfer learning with DenseNet-201 had the best performance (R2 = 0.8303, root mean-squared error (RMSE) = 9.55, mean absolute error (MAE) = 7.03, mean absolute percentage error (MAPE) = 12.54%), followed closely by AutoKeras (R2 = 0.8273, RMSE = 10.65, MAE = 8.24, MAPE = 13.87%). In both tasks, AutoKeras models had up to 40-fold faster inference times compared to the pretrained CNNs. AutoML has significant potential to enhance plant phenotyping capabilities applicable in crop breeding and precision agriculture.

Highlights

High-throughput plant phenotyping (HTP) plays a crucial role in meeting the increasing demand for large-scale plant evaluation in breeding trials and crop management systems [1,2,3]
Both transfer learning with pretrained convolutional neural network (CNN) and AutoKeras performed strongly in the image classification task (Table 2)
CNNs ranged from 91.6% to 93.2% classification accuracy, with Xception

Summary

Introduction

High-throughput plant phenotyping (HTP) plays a crucial role in meeting the increasing demand for large-scale plant evaluation in breeding trials and crop management systems [1,2,3]. Concurrent with the development of various ground-based and aerial (e.g., unmanned aerial vehicle (UAV)) HTP systems is the rise in use of imaging sensors for phenotyping purposes. The meteoric rise in big image data arising from HTP systems necessitates the development of efficient image processing and analytical pipelines. Conventional image analysis pipelines typically involve computer vision tasks (e.g., wheat head counting using object detection), which are addressed through the development of signal processing and/or machine learning (ML) algorithms. These algorithms are sensitive to image-quality (e.g., illumination, sharpness, distortion) variations and do Remote Sens.

Methods

Results

Discussion

Conclusion