The method utilizes RGB, chlorophyll fluorescence (CF), and infrared (IR) thermal imaging to diagnose herbicidal activity and modes of action (MOAs) in a fast, non-destructive manner. The innovative combination of these imaging techniques, along with machine learning algorithms, enables the identification of herbicide effects within just a few days, achieving up to 100% accuracy by the third day of treatment.
The discovery and evaluation of herbicides require significant resources, and many existing diagnostic techniques are limited by their complexity, time requirements, and destructiveness. In recent years, spectral imaging technologies—especially those involving RGB, IR thermal, and CF images—have gained attention for their potential in plant phenotyping, aiding in the detection of biotic and abiotic stresses, including herbicide activity. These imaging methods capture key plant responses such as chlorophyll status, temperature changes, and fluorescence signals, which are critical for diagnosing herbicide effects. However, integrating multiple spectral data types for herbicide screening has not been widely explored until now.
A study (DOI: 10.1016/j.plaphe.2025.100038) published in Plant Phenomics on 7 June 2025 by Do-Soon Kim’s team, Seoul National University, offers the potential to reduce the time and costs involved in herbicide research, while also enhancing the efficiency and accuracy of herbicide screening.
The study employed RGB, CF, and IR thermal imaging to analyze the spectral responses of oilseed rape (Brassica napus) treated with different herbicides, including propanil, oxyfluorfen, mesotrione, and glyphosate, which represent PSII, PPO, HPPD, and EPSPS inhibitors, respectively. These images were analyzed to estimate indices such as Normalized Difference Index (NDI), Excess Green (ExG), PSII quantum yield, and temperature index. A two-way ANOVA revealed significant effects of both herbicide treatment and time on all spectral indices. Distinct differences were observed between herbicide-treated plants and untreated controls, with spectral indices like NDI, ExG, and temperature index showing significant changes from 1 day after treatment (DAT), while PSII quantum yield changed from 6 hours after treatment (HAT). The results indicated that different herbicides induced varied spectral responses over time, with PPO causing the most rapid and severe changes, particularly in NDI and ExG, leading to wilting by 4 DAT. In contrast, glyphosate (EPSPS) and mesotrione (HPPD) exhibited gradual changes in spectral indices, with minimal impact at early time points. Propanil (PSII), on the other hand, showed subtle and slower reductions in NDI and ExG, but significant recovery in PSII quantum yield by 6 DAT.The CF images revealed herbicide-induced stress as early as 6 HAT, with significant differences between herbicide treatments even when no visible symptoms were present in RGB images. Propanil and oxyfluorfen showed the most rapid decline in PSII quantum yield, while mesotrione and glyphosate exhibited slower responses. In IR thermal images, all herbicides caused an increase in temperature index, with oxyfluorfen showing the most significant rise in leaf temperature within 1 DAT. Machine learning analysis of the spectral indices further demonstrated that herbicide modes of action could be diagnosed with high accuracy, achieving 100% accuracy by 3 DAT. The combination of spectral indices, particularly PSII quantum yield and temperature index, significantly improved diagnostic accuracy, confirming the potential of this method for rapid and precise herbicide screening.
This study highlights the potential of spectral image analysis and machine learning to revolutionize herbicide screening and mode-of-action diagnostics. By combining RGB, CF, and IR imaging, researchers can quickly and accurately evaluate the effectiveness and mechanism of herbicides. This technique not only reduces the time and cost associated with herbicide screening but also holds promise for more efficient agricultural practices by accelerating the development of new herbicides.
###
References
DOI
10.1016/j.plaphe.2025.100038
Original URL
https://doi.org/10.1016/j.plaphe.2025.100038
Funding information
This work was carried out with the support of “Cooperative Research Program for Agricultural Science & Technology Development (Project No. RS-2024-00397586)”, Rural Development Administration, Republic of Korea.
About Plant Phenomics
Plant Phenomics is dedicated to publishing novel research that will advance all aspects of plant phenotyping from the cell to the plant population levels using innovative combinations of sensor systems and data analytics. Plant Phenomics aims also to connect phenomics to other science domains, such as genomics, genetics, physiology, molecular biology, bioinformatics, statistics, mathematics, and computer sciences. Plant Phenomics should thus contribute to advance plant sciences and agriculture/forestry/horticulture by addressing key scientific challenges in the area of plant phenomics.