Fluorescence Imaging Detects Tomato Disease Early

Spotting Tomato Diseases Early with AI: A New Way to Save Your Crops

Soraya Malik monogram
Soraya Malik in Tech & Innovation October 2026 • 4 min read.

"Fluorescence Imaging and AI Team Up to Detect Tomato Yellow Leaf Curl Before It's Too Late"


Tomato Yellow Leaf Curl Disease (TYLCD) is a major threat to tomato crops worldwide, capable of causing significant yield losses. Early detection is crucial, but traditional methods often rely on visible symptoms that may not appear until the disease is well-established, making timely intervention difficult.

Now, a groundbreaking study explores the use of fluorescence imaging combined with artificial intelligence to identify TYLCD in tomato plants even before they show obvious signs of infection. This innovative approach promises to give farmers a powerful new tool to combat this devastating disease.

This article delves into the details of this research, explaining how the technology works, what the key findings are, and what the implications could be for tomato growers seeking to protect their crops.

AI Search Multiple angles on this topic

Tomato Disease’s Global Stakes

Tomato yellow leaf curl disease (TYLCD), caused by tomato yellow leaf curl virus (TYLCV) and related begomoviruses, has caused serious economic problems in tomato production worldwide. The review identifies whiteflies in the Bemisia tabaci complex as vectors responsible for many TYLCD outbreaks. Separately, a 2025 dataset description projects the global tomato market value to exceed US $166 billion in 2025, with annual production surpassing 180 million metric tons; it reports that disease-related yield losses can range from 30% to nearly 100% during severe epidemics of major pathogens such as late blight.

Why Early Detection Is Difficult

For TYLCD, traditional detection methods are limited by a prolonged asymptomatic infection period and early symptoms that are not distinct. A 2025 study describes conventional approaches as too slow for timely tomato-disease detection and examines deep learning for classification and detection. Another study frames accurate, timely leaf-disease detection as a formidable challenge for effective early intervention in agricultural practice.

How Does Fluorescence Imaging Spot Early Signs of TYLCD?

Fluorescence Imaging Detects Tomato Disease Early

The core of this new method lies in the way healthy and infected plants respond differently to light. Healthy plant tissues naturally emit fluorescence when exposed to certain wavelengths of light. When TYLCD infects a tomato plant, this fluorescence pattern changes, even before visible symptoms like leaf curling or yellowing appear. Researchers used a specialized high-speed camera and carefully chosen filters to capture these subtle changes in fluorescence.

The process involves:

  • Capturing Fluorescence: A high-speed camera records the fluorescence emitted by tomato leaves under specific lighting conditions.
  • Isolating Leaf Pixels:Sophisticated image processing techniques automatically separate the leaf pixels from the background in each image.
  • Extracting Key Features: The system analyzes both the texture of the leaf and the intricate patterns of its veins, using a method called Gray-Level Co-occurrence Matrix (GLCM) to pull out meaningful data.
  • AI-Powered Analysis: Artificial intelligence algorithms then analyze these texture and vein features to distinguish between healthy leaves and those infected with TYLCD.
AI Search Multiple angles on this topic

Fluorescence and Robust Image Features

Research on chlorophyll fluorescence imaging in ten garden shrub species under flooding stress reports more pronounced heterogeneity in veins and mesophyll under stress than in healthy plants, describing this as a common response to biological or abiotic stress. A separate study on texture classification identifies desirable feature-vector properties: discriminating power, noise robustness, invariance to image orientation and lighting changes, compact dimensionality, and implementation efficiency. These sources address plant stress imaging and general texture classification rather than tomato-disease diagnosis specifically.

Comparing Models and Field Adaptation

A review of tomato leaf-disease identification analyzes recent advances using machine-learning models. Another source introduces auto-learning and deep-learning models intended to improve the accuracy and speed of plant-disease diagnosis, while noting that modern methods have often lacked the accuracy, efficiency, and productivity needed for early diagnosis. For non-stationary environments, a 2025 study proposes CTTA-DisDet, a continuous test-time domain adaptation framework combining knowledge distillation with dynamic data augmentation.

Unlike traditional methods that depend on human observation of symptoms, this technique can detect the disease at a much earlier stage, allowing for quicker action.

The Future of Disease Detection: What This Means for Farmers

This research demonstrates the potential of using fluorescence imaging and AI to create a non-destructive method for early TYLCD detection. The system achieved high accuracy in identifying infected leaves, even before they showed visible symptoms.

AI Search Multiple angles on this topic

Building Models for Real-World Conditions

A 2024 study focuses on tomato-disease images captured against intricate backgrounds, where environmental disturbances can make precise detection and identification difficult. A 2025 investigation uses data augmentation to simulate variations in farm images and enrich datasets, aiming to identify seven tomato diseases—including bacterial spot, early blight, and late blight—and healthy leaves. A broader survey describes computer-vision approaches, machine-learning algorithms, and hybrid models as methods being explored for improved detection accuracy and efficiency.

Data and Field Conditions

A 2026 review identifies two practical bottlenecks for reliable tomato leaf-disease detection in real fields: limited data, including small samples for rare diseases, class imbalance, and noisy field images; and multimodal heterogeneity across RGB images, textual symptom descriptions, and spectral cues. The same review describes tomato leaf diseases as causing substantial yield and quality losses worldwide. Separately, a review of TYLCD notes that the whitefly-transmitted disease leads to considerable yield losses.

Investigating Fluorescence Imaging for TYLCD

A study investigates whether fluorescence imaging can discriminate leaves affected by tomato yellow leaf curl disease (TYLCD). Its stated focus is feasibility, using texture and leaf-vein features. The supplied source description does not report results, so it supports the research aim but not a conclusion about diagnostic performance.

This offers several key benefits:

<ul><li><b>Early Intervention:</b>Detecting the disease early allows farmers to take swift action to prevent it from spreading, potentially saving entire crops.</li><li><b>Reduced Losses:</b>By stopping the disease in its tracks, farmers can minimize yield losses and economic damage.</li><li><b>More Sustainable Practices:</b>Targeted interventions reduce the need for widespread pesticide applications, promoting more sustainable farming practices.</li></ul>

How fluorescence compares with other tomato-disease approaches

A review also describes spectral-index studies reaching up to perfect accuracy across several tomato diseases, while another reports high accuracy for distinguishing diseases and healthy plants in laboratory and field settings.[1]

The fluorescence study at a glance

PopulationHealthy and TYLCD-infected tomato plants[2]
Main outcomeClassification accuracy for identifying infected leaves[2]
Times cited4[2]

Reading the fluorescence results

The reported performance suggests that the selected image features contained information useful for separating infected from healthy leaves in this study. The especially strong result for BLR is specific to the combined feature group and the reported evaluation, not proof that the same performance transfers to other settings.[2]

The imaging setup used filtered fluorescence captured with a high-speed camera, while the analyzed central subimages contained only leaf pixels. This makes the reported classification about selected leaf imagery, rather than an assessment of an unrestricted whole-scene field image.[2]

What a follow-up would need to establish

  • Would the fluorescence classifiers retain their reported discrimination when tested on new plants and images collected outside the selected central leaf regions?[2]
  • How would the approach perform across field conditions, where hyperspectral studies have already reported results in both laboratory and field settings?[1] [2]
  • Could the method distinguish TYLCD from lookalike disorders, rather than only separating the study’s healthy and infected plants?[2] [3]

Sources

  1. Current State of Hyperspectral Remote Sensing for Early Plant Disease Detection: A Review (pmc.ncbi.nlm.nih.gov)
  2. Discriminating Leaves Affected with Tomato Yellow Leaf Curl through Fluorescence Imaging Using Texture and Leaf Vein Features (American Society of Agricultural and Biological Engineers (ASABE), 2016)
  3. Tomato Yellow Leaf Curl Virus | NC State Extension Publications (content.ces.ncsu.edu)

About this Article -

Written with AI assistance from published research, and reviewed by the Mystum team. See our About page for more information.

Everything You Need To Know

1

How does fluorescence imaging work to detect Tomato Yellow Leaf Curl Disease (TYLCD)?

The core of this innovative approach involves analyzing how healthy and infected tomato plants react to light. Healthy plants emit a natural fluorescence when exposed to specific light wavelengths. However, when the Tomato Yellow Leaf Curl Disease (TYLCD) infects a plant, this fluorescence pattern changes before any visible signs appear. Researchers use a high-speed camera and filters to capture these subtle changes.

2

Why is fluorescence imaging important for identifying Tomato Yellow Leaf Curl Disease (TYLCD)?

Fluorescence imaging is important because it allows for the early detection of Tomato Yellow Leaf Curl Disease (TYLCD). Traditional methods often detect the disease after it is well-established, making treatment difficult. By detecting the disease at an early stage, farmers can take immediate actions, like removing infected plants or applying targeted treatments, before it spreads and causes significant yield losses.

3

What is the role of the Gray-Level Co-occurrence Matrix (GLCM) in this disease detection method?

The Gray-Level Co-occurrence Matrix (GLCM) is a technique used in the image processing of the tomato leaves. This method helps in extracting meaningful data from the leaf's texture and vein patterns. By analyzing these features, the artificial intelligence algorithms can differentiate between healthy leaves and those infected with Tomato Yellow Leaf Curl Disease (TYLCD), even before symptoms are visible.

4

How does artificial intelligence contribute to detecting Tomato Yellow Leaf Curl Disease (TYLCD)?

Artificial intelligence plays a crucial role by analyzing the data extracted from the fluorescence imaging. After capturing the fluorescence images and isolating the leaf pixels, key features like the texture and vein patterns are analyzed by AI algorithms. These algorithms are trained to identify patterns indicative of Tomato Yellow Leaf Curl Disease (TYLCD), thus distinguishing between healthy and infected leaves.

5

What are the potential benefits for farmers using this new disease detection method?

The implications for tomato growers are significant. The use of fluorescence imaging and artificial intelligence provides a non-destructive method for early detection of Tomato Yellow Leaf Curl Disease (TYLCD). Early detection allows farmers to take prompt action, leading to better management and control of the disease. This translates to higher crop yields, reduced losses, and a more sustainable approach to tomato farming.

Newsletter Subscribe

Subscribe to get the latest articles and insights directly in your inbox.