ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial Intelligence

Mango productivity suffers greatly from leaf diseases, leading to economic and food security issues. Current visual inspection methods are slow and subjective. Previous Deep-Learning (DL) solutions have shown promise but suffer from imbalanced datasets, modest generalization, and limited interpretab...

Full description

Saved in:
Bibliographic Details
Main Authors: Abdullah Al Noman, Amira Hossain, Anamul Sakib, Jesika Debnath, Hasib Fardin, Abdullah Al Sakib, Rezaul Haque, Md. Redwan Ahmed, Ahmed Wasif Reza, M. Ali Akber Dewan
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Computers
Subjects:
Online Access:https://www.mdpi.com/2073-431X/14/5/171
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850127419058421760
author Abdullah Al Noman
Amira Hossain
Anamul Sakib
Jesika Debnath
Hasib Fardin
Abdullah Al Sakib
Rezaul Haque
Md. Redwan Ahmed
Ahmed Wasif Reza
M. Ali Akber Dewan
author_facet Abdullah Al Noman
Amira Hossain
Anamul Sakib
Jesika Debnath
Hasib Fardin
Abdullah Al Sakib
Rezaul Haque
Md. Redwan Ahmed
Ahmed Wasif Reza
M. Ali Akber Dewan
author_sort Abdullah Al Noman
collection DOAJ
description Mango productivity suffers greatly from leaf diseases, leading to economic and food security issues. Current visual inspection methods are slow and subjective. Previous Deep-Learning (DL) solutions have shown promise but suffer from imbalanced datasets, modest generalization, and limited interpretability. To address these challenges, this study introduces the ViX-MangoEFormer, which combines convolutional kernels and self-attention to effectively diagnose multiple mango leaf conditions in both balanced and imbalanced image sets. To benchmark against ViX-MangoEFormer, we developed a stacking ensemble model (MangoNet-Stack) that utilizes five transfer learning networks as base learners. All models were trained with Grad-CAM produced pixel-level explanations. In a combined dataset of 25,530 images, ViX-MangoEFormer achieved an F1 score of 99.78% and a Matthews Correlation Coefficient (MCC) of 99.34%. This performance consistently outperformed individual pre-trained models and MangoNet-Stack. Additionally, data augmentation has improved the performance of every architecture compared to its non-augmented version. Cross-domain tests on morphologically similar crop leaves confirmed strong generalization. Our findings validate the effectiveness of transformer attention and XAI in mango leaf disease detection. ViX-MangoEFormer is deployed as a web application that delivers real-time predictions, probability scores, and visual rationales. The system enables growers to respond quickly and enhances large-scale smart crop health monitoring.
format Article
id doaj-art-dd3b7c97b1e141b1a8fbdf89f3205aea
institution OA Journals
issn 2073-431X
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Computers
spelling doaj-art-dd3b7c97b1e141b1a8fbdf89f3205aea2025-08-20T02:33:42ZengMDPI AGComputers2073-431X2025-05-0114517110.3390/computers14050171ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial IntelligenceAbdullah Al Noman0Amira Hossain1Anamul Sakib2Jesika Debnath3Hasib Fardin4Abdullah Al Sakib5Rezaul Haque6Md. Redwan Ahmed7Ahmed Wasif Reza8M. Ali Akber Dewan9Department of Information Technology, Westcliff University, Irvine, CA 92614, USADepartment of Computer Science, Westcliff University, Irvine, CA 92614, USADepartment of Business Administration, International American University, 3440 Wilshire Blvd STE 1000, Los Angeles, CA 90010, USADepartment of Computer Science, Westcliff University, Irvine, CA 92614, USADepartment of Engineering Management, Westcliff University, Irvine, CA 92614, USADepartment of Information Technology, Westcliff University, Irvine, CA 92614, USADepartment of Computer Science and Engineering, East West University, Dhaka 1212, BangladeshDepartment of Computer Science and Engineering, East West University, Dhaka 1212, BangladeshDepartment of Computer Science and Engineering, East West University, Dhaka 1212, BangladeshSchool of Computing and Information Systems, Athabasca University, Athabasca, AB T9S 3A3, CanadaMango productivity suffers greatly from leaf diseases, leading to economic and food security issues. Current visual inspection methods are slow and subjective. Previous Deep-Learning (DL) solutions have shown promise but suffer from imbalanced datasets, modest generalization, and limited interpretability. To address these challenges, this study introduces the ViX-MangoEFormer, which combines convolutional kernels and self-attention to effectively diagnose multiple mango leaf conditions in both balanced and imbalanced image sets. To benchmark against ViX-MangoEFormer, we developed a stacking ensemble model (MangoNet-Stack) that utilizes five transfer learning networks as base learners. All models were trained with Grad-CAM produced pixel-level explanations. In a combined dataset of 25,530 images, ViX-MangoEFormer achieved an F1 score of 99.78% and a Matthews Correlation Coefficient (MCC) of 99.34%. This performance consistently outperformed individual pre-trained models and MangoNet-Stack. Additionally, data augmentation has improved the performance of every architecture compared to its non-augmented version. Cross-domain tests on morphologically similar crop leaves confirmed strong generalization. Our findings validate the effectiveness of transformer attention and XAI in mango leaf disease detection. ViX-MangoEFormer is deployed as a web application that delivers real-time predictions, probability scores, and visual rationales. The system enables growers to respond quickly and enhances large-scale smart crop health monitoring.https://www.mdpi.com/2073-431X/14/5/171Vision Transformer (ViT)explainable AI (XAI)ensemble learningprecision agriculturemango leaf classification
spellingShingle Abdullah Al Noman
Amira Hossain
Anamul Sakib
Jesika Debnath
Hasib Fardin
Abdullah Al Sakib
Rezaul Haque
Md. Redwan Ahmed
Ahmed Wasif Reza
M. Ali Akber Dewan
ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial Intelligence
Computers
Vision Transformer (ViT)
explainable AI (XAI)
ensemble learning
precision agriculture
mango leaf classification
title ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial Intelligence
title_full ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial Intelligence
title_fullStr ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial Intelligence
title_full_unstemmed ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial Intelligence
title_short ViX-MangoEFormer: An Enhanced Vision Transformer–EfficientFormer and Stacking Ensemble Approach for Mango Leaf Disease Recognition with Explainable Artificial Intelligence
title_sort vix mangoeformer an enhanced vision transformer efficientformer and stacking ensemble approach for mango leaf disease recognition with explainable artificial intelligence
topic Vision Transformer (ViT)
explainable AI (XAI)
ensemble learning
precision agriculture
mango leaf classification
url https://www.mdpi.com/2073-431X/14/5/171
work_keys_str_mv AT abdullahalnoman vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT amirahossain vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT anamulsakib vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT jesikadebnath vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT hasibfardin vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT abdullahalsakib vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT rezaulhaque vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT mdredwanahmed vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT ahmedwasifreza vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence
AT maliakberdewan vixmangoeformeranenhancedvisiontransformerefficientformerandstackingensembleapproachformangoleafdiseaserecognitionwithexplainableartificialintelligence