Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Cross Disease Similarity Awareness Learning (CDSAL) with DenseNet-EfficientNet embedding fusion for high-precision

C Vanmathi1, R Mangayarkarasi2, Rajat Agrawal2

  • 1School of Computer Science Engineering and Information Systems, Vellore Institute of Technology, Vellore, India. vanmathi.c@vit.ac.in.

Scientific Reports
|May 21, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

A sustainable driven intrusion detection model for green CPS using ISP analysis and energy aware deep ensemble learning.

Scientific reports·2026
Same author

Development of an Anticancer Fungal Paclitaxel-Loaded Hydrogel System via 3D Bioprinting: A Next-Generation Biomedical Platform.

ACS omega·2026
Same author

Effective photocatalysis and toxicity investigations in zebra fish embryos using highly selective hydrogen sensors and supercapacitor electrode materials for SnO<sub>2</sub>-PANI nanocomposites.

Nanoscale advances·2026
Same author

Polysaccharide synergy of alginate/kappa carrageenan with microalgal system for sustainable heavy metal remediation.

International journal of biological macromolecules·2025
Same author

Identifying and modeling barriers to design thinking implementation in healthcare sector: an ISM-MICMAC approach.

Journal of health organization and management·2025
Same author

Ocular Delivery of Viral Vectors for Retinal and Choroidal Vascular Diseases: A Review.

American journal of ophthalmology·2025

Cross Disease Similarity Awareness Learning (CDSAL) enhances tomato leaf disease detection using explainable deep learning. This framework achieves 99.77% accuracy, effectively distinguishing visually similar diseases for reliable agricultural diagnostics.

Area of Science:

  • Agricultural Science
  • Computer Science
  • Machine Learning

Background:

  • Tomato leaf diseases pose a significant threat to crop yield.
  • Accurate detection of visually similar diseases like Leaf Miner, Tomato Spotted Wilt Virus (TSWV), and nutrient deficiencies is challenging.
  • Existing methods often struggle with superimposed disease patterns and lack explainability.

Purpose of the Study:

  • To develop a robust, explainable deep learning framework for multiclass tomato leaf disease detection.
  • To address the challenge of differentiating visually overlapping diseases and nutrient deficiencies.
  • To improve the accuracy and reliability of agricultural diagnostics through advanced machine learning.

Main Methods:

  • Proposed Cross Disease Similarity Awareness Learning (CDSAL) framework utilizing multi-domain feature learning and inter-disease similarity modeling.

Related Experiment Videos

  • Developed a class-level Cross Disease Similarity Matrix for structured inter-disease proximity representation.
  • Employed DenseNet121 and EfficientNet-B0 for feature extraction, with persistent centroid updates and similarity-aware regularization.
  • Utilized HSV green masking, morphological cleaning, contour extraction, resizing, and extensive data augmentation for image preprocessing.
  • Incorporated Grad-CAM for explainable, disease-specific activation signatures.

    • Main Results:

    • Achieved 99.77% classification accuracy on unseen tomato leaf disease samples.
    • Demonstrated high resilience to visual confounding among similar disease categories.
    • Successfully generated predictions, identified disease proximity, and provided explainable features.

    Conclusions:

    • CDSAL provides a novel and effective approach for accurate and explainable multiclass tomato leaf disease detection.
    • The framework's ability to model inter-disease similarity significantly improves classification of visually overlapping conditions.
    • CDSAL facilitates reliable decision-making in agricultural diagnostics by offering interpretable insights.