A Study on AI-Based Diverse Bacterial Blight Symptoms by Water-Soaked and Necrotic Lesions in Pomegranate Leafs and Fruits Using Image Analysis on Computer Vision
Journal of Intelligent Decision Making and Information Science · 10.10.2026 · 3 (17s) · 330-351
DOI: 10.17762/jidmis.v3.5377 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Intelligent Decision Making and Information Science, online erschienen am 10.10.2026. Verfasst von Naveen M V. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Conventional visual disease diagnosis is laborious, subjective and time consuming and the image analysis of pomegranate bacterial blight is complex due to the significant variation in colour, texture, morphology, size and spatial distribution of water-soaked and necrotic lesions, both within and between leaves and fruits. General disease classification is the major concern of existing image-based methods, which also have received very limited attention in terms of symptom-level analysis of the disease in both respective plant organs. This work presents an Artificial Intelligence (AI) computer vision approach for water-soaked lesion detection and characterization of various bacterial blight disease outcomes, focusing on water-soaked and necrotic lesions. The proposed framework identifies and segments the lesions from the images using the K-Means algorithm, detects the symptoms, estimates the disease-affected area, extracts colour, texture and morphological features, classifies after multi-class SVM using stratified training and testing. Experimental analysis consisted of 965 images with 555 leaf and 410 fruit samples with 676 training images, 145 validation images and 144 testing images: of which 360 were healthy images, 320 were water soaked images and 285 were necrotic images. Results of k means segmentation gave K=4, objective value=2.9 and Peak Silhouette value = 0.71. The intensity span by the histogram equalisation technique was reduced from 900 to 70. The multi-class SVM yields an F1 score of 0.945 for healthy samples, 0.920 for water-soaked samples and 0.975 for necrotic samples, achieving a macro-average F1 score of ~0.947. The framework illustrates the possible benefits and potentials of automated image analysis at the symptom level that could be realized to recognize and quantify pomegranate diseases automatically.
Zitieren
APA
V, N. M. (2026). A Study on AI-Based Diverse Bacterial Blight Symptoms by Water-Soaked and Necrotic Lesions in Pomegranate Leafs and Fruits Using Image Analysis on Computer Vision. Journal of Intelligent Decision Making and Information Science, 3(17s), 330-351. https://doi.org/10.17762/jidmis.v3.5377
Vancouver
V NM. A Study on AI-Based Diverse Bacterial Blight Symptoms by Water-Soaked and Necrotic Lesions in Pomegranate Leafs and Fruits Using Image Analysis on Computer Vision. Journal of Intelligent Decision Making and Information Science. 2026;3(17s):330-351. doi:10.17762/jidmis.v3.5377
BibTeX
@article{V2026,
author = {V, Naveen M},
title = {A Study on AI-Based Diverse Bacterial Blight Symptoms by Water-Soaked and Necrotic Lesions in Pomegranate Leafs and Fruits Using Image Analysis on Computer Vision},
journal = {Journal of Intelligent Decision Making and Information Science},
year = {2026},
volume = {3},
number = {17s},
pages = {330--351},
doi = {10.17762/jidmis.v3.5377}
}
EUVIMED