Non-destructive Classification of Phytophthora Blight and Water Stress in Perilla Using Hyperspectral Imaging and 1D-CNN

Tae-An Kang; Chi-Yong An; Youngnam Yoon; Jongsoo Ryu; Hongseok Lee

doi:10.22698/jales.20260010

All Issue

2026 Vol.38, Issue 2 Preview Page Next Page

Non-destructive Classification of Phytophthora Blight and Water Stress in Perilla Using Hyperspectral Imaging and 1D-CNN

30 June 2026. pp. 136-150

PDF XML

Abstract

Phytophthora blight and water stress in perilla (Perilla frutescens L.) exhibit similar foliar symptoms, making visual differentiation challenging. In this study, we developed a non-destructive classification method using hyperspectral imaging (HSI) in conjunction with Partial Least Squares Discriminant Analysis (PLS-DA) and a one-dimensional convolutional neural network (1D-CNN). The “Anyu” perilla cultivar was subjected to four treatments: control, drought, waterlogging, and inoculation with Phytophthora nicotianae MR40622. Hyperspectral images (400－1,000 nm) were acquired daily, and pixel-level spectra were preprocessed using Raw, Savitzky-Golay (SG) smoothing, Standard Normal Variate (SNV), and first-derivative methods. Classification was performed across three stages of stress progression: early, middle, and late. PLS-DA demonstrated limited sensitivity to preprocessing, achieving a maximum F1 score of 0.688 at the late stage when using SNV. Conversely, the 1D-CNN model combined with first-derivative preprocessing consistently yielded the highest performance, reaching F1 scores of 0.527, 0.772, and 0.827, at the early, middle, and late stages, respectively. The performance gap between PLS-DA and 1D-CNN widened from 0.004 to 0.139 as stress increased, indicating that the 1D-CNN effectively captured nonlinear spectral patterns. Saliency map analysis revealed that the chlorophyll absorption band at 660－690 nm contributed most to classification across all stages, while the water absorption band at 900－950 nm exhibited an increased contribution for the drought and waterlogging treatments during the late stage. These results demonstrate that HSI combined with a 1D-CNN serves an effective, non-destructive diagnostic tool for distinguishing between Phytophthora blight and water stress in perilla.

Keywords

1D-CNN

Hyperspectral imaging

Perilla

Phytophthora

Water stress

References

Afroz, T., Lee, H. S., Jeon, Y. K., Sung, J. S., Rhee, J. H., D. A. Awraris, J., Noh, A., Hwang, O. S. Hur, Ro, N. Y., Lee, J. E., Lee, S. M. (2019) Evaluation of different inoculation methods for screening Sclerotinia rot and Phytophthora blight in perilla germplasm. J Crop Sci Biol 22:177-183.

10.1007/s12892-019-0115-0

Barker, M., Rayens, W. (2003) Partial least squares for discrimination. J Chemom 17:166-173.

10.1002/cem.785

Carter, G. A., Knapp, A. K. (2001) Optical properties of leaves in higher plants: linking spectral characteristics to stress and chlorophyll concentration. Am J Bot 88:677-684.

10.2307/2657068

Ding, Y., Zhang, Z., Zhao, X., Hong, D., Cai, W., Yu, C., Yang, N., Cai, W. (2022) Multi-feature fusion: Combining graph neural networks and CNNs for hyperspectral image classification. Neurocomputing 501:246-257.

10.1016/j.neucom.2022.06.031

Fujiwara, Y., Kono, M., Ito, A., Ito, M. (2018) Anthocyanins in perilla plants and dried leaves. Phytochemistry 147:158-166.

10.1016/j.phytochem.2018.01.003

Gitelson, A. A., Gritz, Y., Merzlyak, M. N. (2003) Relationships between leaf chlorophyll content and spectral reflectance, and algorithms for non-destructive chlorophyll assessment in higher plant leaves. J. Plant Physiol 160:271-282.

10.1078/0176-1617-00887

Guo, C., Liu, L., Sun, H., Wang, N., Zhang, K., Zhang, Y., Zhu, J., Li, A., Bai, Z., Liu, X., Dong, H., Li, C. (2022) Predicting Fv/Fm and evaluating cotton drought tolerance using hyperspectral data and a 1D-CNN. Front Plant Sci 13:1007150.

10.3389/fpls.2022.100715036330250PMC9623111

Hornero, A., Zarco-Tejada, P. J., Quero, J. L., North, P. R., Ruiz-Gómez, F. J., Sánchez-Cuesta, R., Hernández-Clemente, R. (2021) Modeling hyperspectral- and thermal-based plant traits for the early detection of Phytophthora-induced symptoms in oak decline. Remote Sens Environ 263:112570.

10.1016/j.rse.2021.112570

Kang, T. A., An, C. Y., Jeong, H., Park, Y. J., Park, J., Ryu, J., Bae, H., Yoon, Y., Lee, H. (2025) Early detection of sesame leaf disease using hyperspectral imaging and machine learning: a patch-based spatial-spectral integration approach. Frontiers in Food Systems 9:1716907.

10.3389/fsufs.2025.1716907

Kiranyaz, S., Avci. O., Abdeljaber, O., Ince, T., Gabbouj, M., Inman, D. J. (2021) 1D Convolutional Neural Networks and Applications: A survey. Mech Syst Signal Process 151:107398.

10.1016/j.ymssp.2020.107398

Lamour, K. H., Stam, R., Jupe, J., Huitema, E. (2012) The oomycete broad-host-range pathogen Phytophthora capsici. Mol Plant Pathol 13:329-337.

10.1111/j.1364-3703.2011.00754.x22013895PMC6638677

Lee, J., Shin, Y. (2024) Antioxidant Compounds and Activities of Perilla frutescens var. crispa and Its Processed Products. Food Sci Biotechnol 33:1123-1133.

10.1007/s10068-023-01412-z38440683PMC10908715

Lee, Y., Kim, G. W. Jeong, C., Kim, H., Choi, M., Park, S., Lee, T. J., Kim, H. J. (2026) Raman Spectroscopy and One-Dimensional Convolutional Neural Networks for the Forensic Identification of Red Stamp Inks in Questioned Documents. J. Raman Spectrosc 57:80-93.

10.1002/jrs.70018

Lowe, A., Harrison, N., French, A. P. (2017) Hyperspectral image analysis techniques for the detection and classification of the early onset of plant disease and stress. Plant Methods 13:80.

10.1186/s13007-017-0233-z29051772PMC5634902

Peñuelas, J., Pinol, J., Ogaya, R., Filella, I. (1997) Estimation of plant water concentration using the Water Index (WI) (R900/R970). Int J Remote Sens 18:2869-2875.

10.1080/014311697217396

Pscheidt, J. W. (2023) Diagnosis and Management of Phytophthora Diseases. Pacific Northwest Pest Management Handbooks.

Rozenstein, O., Paz-Kagan, T., Salbach, C., Karnieli, A. (2014) Comparing the effect of preprocessing transformations on land-use classification methods derived from spectral soil measurements. IEEE J Sel Top Appl Earth Obs Remote Sens 8:2393-2404.

10.1109/JSTARS.2014.2371920

Shin, H. S., Kim, S. W. (1994) Lipid composition of perilla seeds. J Am Oil Chem Soc 71:619-622.

10.1007/BF02540589

Smith, K. L., Steven, M. D., Colls, J. J. (2004) Use of hyperspectral derivative ratios in the red-edge region to identify plant stress responses to gas leaks. Remote Sens Environ 92:207-217.

10.1016/j.rse.2004.06.002

Wu, J., Handique, U., Graham, J., Johnson, E. (2020) Phytophthora nicotianae infection of citrus leaves and host defense activation compared to root infection. Phytopathology 110:1437-1448.

10.1094/PHYTO-09-19-0343-R

Yu, H., Qiu, J. F., Ma, L. J., Hu, Y. J., Li, P., Wan, J. B. (2017) A review of the phytochemicals and phytopharmacology of Perilla frutescens L. (Labiatae), a traditional edible and medicinal herb in China. Food Chem Toxicol 108:375-391.

10.1016/j.fct.2016.11.023

Information

Publisher :Agriculture and Life Sciences Research Institute, Kangwon National University
Publisher(Ko) :None
Journal Title :Journal of Agricultural, Life and Environmental Sciences
Volume : 38
No :2
Pages :136-150
Received Date : 2026-04-02
Revised Date : 2026-04-03
Accepted Date : 2026-04-13
DOI :https://doi.org/10.22698/jales.20260010

Journal of Agricultural, Life and Environmental Sciences ISSN:2233-8322(Print) 2508-870X(Online)

All Issue