Noise-Resilient Hybrid EfficientNet–Vision Transformer Framework with Adaptive Symmetric Cross-Entropy Loss for Robust Plant Disease Detection Pradeep Gupta, Rakesh Singh Jadon Journal of Applied Science and Technology Trends, 2026 The errors of human annotation and the noise of the environment such as lighting changes, occlusions and cluttered backdrop limit the correct detection of the plant diseases in the field condition. The research hypothesis is to present a robust deep learning model that can withstand noise and be interpretable in controlled and noisy environments to achieve high plant disease classification. The hybrid EfficientNet-Vision Transformer (ViT) network proposed is based on an EfficientNet-B4 branch of CNN and a branch of Vision Transformer (ViT) network, which focuses on capturing fine-grained lesion features and global contexts information. A data augmentation pipeline based on CycleGAN is used to introduce field-style distortions (e.g., (lighting shifts, shadowing, debris and partial occlusions), to be more robust to environmental noise, and an Adaptive Symmetric Cross-Entropy (ASCE) loss identifies and down-weights uncertain samples with normalized prediction entropy. The training is done in two phases, Stage 1 pretraining with clean images of PlantVillage and Stage 2 with increasingly noisy samples. The framework is tested in two different noise conditions, and these include the controlled synthetic label noise with PlantVillage and the real environmental noise with PlantDoc. The proposed model has an accuracy of 94.5% on the clean PlantVillage test set. It achieves 85.0% accuracy on the PlantVillage dataset under the 20% synthetic label noise protocol, outperforming ResNet-50V2 (76.5%), DenseNet-121 (78.9%), and Co-Teaching (79.5%). Macro-precision, macro-recall and macro-F1 of the model on the external PlantDoc field dataset are 0.718, 0.681, 0.681, respectively with a top-1 accuracy of 72.0, which is a manifestation of cross-domain generalization. The lesion-centric Grad-CAM images indicate that the model places emphasis on symptomatic areas of leaves and represses reactions of background soil, shadows, and clutters. The suggested hybrid EfficientNet-ViT architecture offers, in general, a robust and explainable solution to precision agriculture and intelligent crop tracking systems that are resistant to noise.
AppleVit: A Smart Agricultural Software for Apple Leaf Disease Detection Using AI Pradeep Gupta, Rakesh Singh Jadon Journal of Applied Science and Technology Trends, 2025 Apple leaf diseases endanger global apple production at such an intensity that it demands precise detection systems to control disease spread effectively. Traditional inspection methods and Convolutional Neural Network (CNN)-based models face challenges when processing extended image dependencies in leaf images, which subsequently affects their ability to identify diseases accurately. This research develops AppleViT, a lightweight Vision Transformer (ViT)-based model that applies Vision Transformer technology with self-attention approaches to enhance leaf disease classification accuracy and feature extraction within apple leaf detection systems. AppleViT was trained using a public dataset comprising 9,714 apple leaf images, categorized into four classes: Apple Scab, Black Rot, Cedar Apple Rust, and Healthy. The accuracy rate of AppleViT reached 97.8%, which exceeded the ResNet-50 and EfficientNet-B3 and MobileNetV3 models while operating with 1.3 million parameters suitable for precision agriculture real-time usage. The proposed approach demonstrates both high generalization skills alongside precise precision and recall value measurements for disease categories. Future research will create attention visualization features and mobile application compatibility before expanding the architecture to identify multiple diseases across different plant types. AppleViT highlights the potential of Vision Transformer (ViT) technology as a powerful tool to revolutionize plant disease detection for improving crop yield and disease management worldwide.
PlantVitGnet: A Hybrid Model of Vision Transformer and GoogLeNet for Plant Disease Identification Pradeep Gupta, Rakesh Singh Jadon Journal of Phytopathology, 2025 Diseases are one of the major factors that have the potential to reduce plant production, food security and ultimately humanity's survival. Therefore, timely and correct identification of plant diseases is important in ascertaining methods to control the diseases. This paper focuses on the application of Deep Learning in identifying plant diseases, and the research's recommendation is a combination of the Vision Transformer (ViT) and GoogLeNet architectures. The objective of this work is to combine the strengths of both models so as to attain increased accuracy and faster computation. This proves that the proposed model yields a substantial accuracy of 99.20% a, 99.30% precision and 99.10% recall. F1‐score shows the highest performance compared to several state‐of‐the‐art models. For comparison, the Vision Transformer, better known as ViT, attained a 97.80% accuracy, 97.90% precision, 97.70% recall and 97.80% F1 scores, and GoogLeNet attained 98. 60% accuracy, 98. 70% precision, 98.50% recall and 98.60% F1‐score. The present hybrid model substantially enhances the capacity to identify plant diseases, hence providing a comprehensive means of managing the early diseases in the plantations. Due to high performance in the desired indicators, it is applicable for real‐world purposes, controlling crops and increasing their yields.
Effective Contrastive Feature Recycling for 3D Point-Based Cloud Representation Learning Hemlata Arya, Rakesh Singh Jadon, Rajeev Goyal 2025 International Conference on Electrical Communication and Computing Technologies Iconecct 2025, 2025 Reconstruction and Retrieval of an Three dimensional object from point cloud data be a challenging task in computer vision, especially due to heavy dependency on labelled dataset and complexity of current computational methods. This study presents a novel framework where contrastive learning addresses the mention limitations through a dual branch architecture integrated with an free cycling mechanism. the proposed framework leverages self supervised learning for extract the robust representation from an unlabeled point cloud data, drastically reducing annotation requirements while maintaining superior performance. Our feature-recycling mechanism recovers and utilizes the discarded latent features from pooling operations, enriching the overall feature representation to substantially enhance the model efficiency. The framework utilizes contrastive learning to maximize the agreement between augmented views of identical inputs and minimize the similarity with unrelated samples, thus enabling effectively learning from large-scale unlabeled datasets. Extensive experiments on three benchmark datasets clearly identify the superiority of our approach; achieving 86.1.
Conventional versus Explainable Artificial Intelligence-Driven Threat Analysis in SIEM Systems: A Comparative Study Atul Kumar Chauhan, R.S. Jadon, Mir Shahnawaz Ahmad 2025 International Conference on Electrical Communication and Computing Technologies Iconecct 2025, 2025 Security Information and event management (SIEM) systems have evolved from rule-based threat detection to the implementation of advanced machine learning techniques. The continued development of Explainable Artificial Intelligence (XAI) enhances transparency and clarity in the complicated processes of identifying the threat, analysis, despite conventional SIEM systems being optimized for detection accuracy and processing efficiency. In the context of SIEM systems, this paper offers a thorough comparison of explainable AI-driven approaches with traditional approaches. The study breaks down the methodological differences, assesses performance indicators, and identifies the advantages, limitations, and issues of the application of both approaches and, therefore, reveals the research gap, namely the lack of standardized measures to evaluate the explanation, adversarial robustness, and obstacles in operational integration. The paper concludes by highlighting some future research directions that emphasize the importance of human-AI synergies, explainable models, and enhanced XAI methods in enhancing cybersecurity activities.
Hand gesture object recognition based on the combination of fuzzy reasoning method, back propagation algorithm and mamdani classification approach International Journal of Innovative Technology and Exploring Engineering, 2019
Facial emotion recognition through hand gesture and its position surrounding the face International Journal of Engineering and Advanced Technology, 2019
Recognizing facial expressions using eigenspaces V.R. Vijaykumar, P.T. Vanathi, P. Kanagasapathy Proceedings International Conference on Computational Intelligence and Multimedia Applications Iccima 2007, 2008
Noise-Resilient Hybrid EfficientNet–Vision Transformer Framework with Adaptive Symmetric Cross-Entropy Loss for Robust Plant Disease Detection P Gupta, RS Jadon Journal of Applied Science and Technology Trends 7 (1), 209-222 , 2026 2026
Hybrid Point Cloud Encoders for Self-Supervised 3D Shape Representation Learning H Arya, RS Jadon, R Goyal 2026 IEEE Madhya Pradesh Section Conference (MPCON), 429-436 , 2026 2026
Opinion analysis using social media in the automobile industry with machine learning R Shukla, RS Jadon, R Jain AIP Conference Proceedings 3385 (1), 070004 , 2026 2026
Conventional versus Explainable Artificial Intelligence-Driven Threat Analysis in SIEM Systems: A Comparative Study AK Chauhan, RS Jadon, MS Ahmad 2025 International Conference on Electrical, Communication, and Computing … , 2025 2025
Effective Contrastive Feature Recycling for 3D Point-Based Cloud Representation Learning H Arya, RS Jadon, R Goyal 2025 International Conference on Electrical, Communication, and Computing … , 2025 2025
AppleVit: A Smart Agricultural Software for Apple Leaf Disease Detection Using AI P Gupta, RS Jadon Journal of Applied Science and Technology Trends 6 (2), 219-230 , 2025 2025
PLANT Detect Net: a hybrid IoT and deep learning framework for secure plant disease detection and classification P Gupta, RS Jadon Evolving Systems 16 (2), 55 , 2025 2025 Citations: 4
PlantVitGnet: A hybrid model of vision transformer and GoogLeNet for plant disease identification P Gupta, RS Jadon Journal of Phytopathology 173 (2), e70041 , 2025 2025 Citations: 5
Detection, Segmentation, and Classification on Brain Tumor MRI Dataset S Bansal, RS Jadon, SK Gupta International Conference on Information Technology and Artificial … , 2025 2025
Cervical Spine Fracture Classification Using CT Images by Different Optimizers S Bansal, RS Jadon, SK Gupta International Conference on Communication and Intelligent Systems, 567-577 , 2024 2024
A Robust Hybrid Convolutional Network for Tumor Classification Using Brain MRI Image Datasets. S Bansal, RS Jadon, SK Gupta International Journal of Advanced Computer Science & Applications 15 (4) , 2024 2024 Citations: 14
Plant Leaf Disease Identification Using Deep Learning Algorithms P Gupta, RS Jadon International Conference On Innovative Computing And Communication, 733-744 , 2024 2024 Citations: 1
Features extraction and classification using machine learning classifiers for the recognition of lips and tongue cancer S Bansal, RS Jadon, SK Gupta Available at SSRN 4719018 , 2024 2024 Citations: 3
Correction: Characterization of Indian Visual Arts Architecture Ages and sub-ages using ML and Fuzzy-ML algorithms A Sharma, RS Jadon Multimedia Tools and Applications 83 (6), 18637-18637 , 2024 2024
to Classify Indian Visual Arts A Sharma, RS Jadon Proceedings of Third Emerging Trends and Technologies on Intelligent Systems … , 2023 2023
Stock Price Prediction using Modified BPSO for Feature Selection with RNN Variants on Top Tech Companies P Gupta, N Paharia, SK Gupta, RS Jadon 2023 World Conference on Communication & Computing (WCONF), 1-7 , 2023 2023 Citations: 4
Stock Market Prediction using RNN-based Models with Random and Tuned Hyperparameters P Gupta, SK Gupta, RS Jadon International Journal of Computer Applications 185 (21), 12-17 , 2023 2023 Citations: 2
Plant Disease Detection Using Machine Learning Models RS Jadon, P Gupta Proceedings of the KILBY 100 7th International Conference on Computing Sciences , 2023 2023 Citations: 3
Characterization of Indian Visual Arts Architecture Ages and sub-ages using ML and Fuzzy-ML algorithms A Sharma, RS Jadon Multimedia Tools and Applications 82 (10), 15493-15513 , 2023 2023 Citations: 3
Lips and tongue cancer classification using deep learning neural network S Bansal, RS Jadon, SK Gupta 2023 6th International Conference on Information Systems and Computer … , 2023 2023 Citations: 9
MOST CITED SCHOLAR PUBLICATIONS
A review of vision based hand gestures recognition GRS Murthy, RS Jadon International Journal of Information Technology and Knowledge Management 2 … , 2009 2009 Citations: 519
Hand gesture recognition using neural networks GRS Murthy, RS Jadon 2010 IEEE 2nd international advance computing conference (IACC), 134-138 , 2010 2010 Citations: 128
Effectiveness of eigenspaces for facial expressions recognition GRS Murthy, RS Jadon International Journal of Computer Theory and Engineering 1 (5), 638 , 2009 2009 Citations: 93
A fuzzy theoretic approach for video segmentation using syntactic features RS Jadon, S Chaudhury, KK Biswas Pattern Recognition Letters 22 (13), 1359-1369 , 2001 2001 Citations: 72
Biometric: case study S Jaiswal, SS Bhadauria, RS Jadon Journal of Global Research in Computer Science 2 (10), 19-48 , 2011 2011 Citations: 54
A review on class imbalance problem: Analysis and potential solutions S Maheshwari, RC Jain, RS Jadon International journal of computer science issues (IJCSI) 14 (6), 43-51 , 2017 2017 Citations: 44
Role of artificial intelligence in enterprise information security: a review M Dhingra, M Jain, RS Jadon 2016 fourth international conference on parallel, distributed and grid … , 2016 2016 Citations: 41
Movies genres classifier using neural network SK Jain, RS Jadon 2009 24th International Symposium on Computer and Information Sciences, 575-580 , 2009 2009 Citations: 30
Acoustic domain classification and recognition through ensemble based multilevel classification S Rathor, RS Jadon Journal of Ambient Intelligence and Humanized Computing 10 (9), 3617-3627 , 2019 2019 Citations: 24
RETRACTED CHAPTER: Real-life facial expression recognition systems: A review SJ Goyal, AK Upadhyay, RS Jadon, R Goyal Smart Computing and Informatics: Proceedings of the First International … , 2017 2017 Citations: 24
An implementation of frequent pattern mining algorithm using dynamic function S Joshi, RS Jadon, RC Jain International Journal of Computer Applications 9 (9), 0975-8887 , 2010 2010 Citations: 17
Generic video classification: An evolutionary learning based fuzzy theoretic approach RS Jadon, S Chaudhury, KK Biswas small 7, 0.13 , 2010 2010 Citations: 15
Video on Demand: An Overview SK Jain, RS Jadon Jamia Millia Islamia (A Central University): Feb , 2003 2003 Citations: 15
A Robust Hybrid Convolutional Network for Tumor Classification Using Brain MRI Image Datasets. S Bansal, RS Jadon, SK Gupta International Journal of Advanced Computer Science & Applications 15 (4) , 2024 2024 Citations: 14
The art of domain classification and recognition for text conversation using support vector classifier S Rathor, RS Jadon International Journal of Arts and Technology 11 (3), 309-324 , 2019 2019 Citations: 14
Recognizing facial expressions using eigenspaces GRS Murthy, RS Jadon International Conference on Computational Intelligence and Multimedia … , 2007 2007 Citations: 14
Characterization of tumor region using SOM and Neuro Fuzzy techniques in Digital Mammography A Ahirwar, RS Jadon International Journal of Computer Science and Information Technology 3 (1 … , 2011 2011 Citations: 13
Text indpendent speaker recognition using wavelet cepstral coefficient and butter worth filter S Rathor, RS Jadon 2017 8th International Conference on Computing, Communication and Networking … , 2017 2017 Citations: 12
Sequential pattern mining using formal language tools S Joshi, RS Jadon, RC Jain International Journal of Computer Science Issues (IJCSI) 9 (5), 316 , 2012 2012 Citations: 12
Audio based movies characterization using neural network S Jain, RS Jadon International Journal of Computer Science and Applications 1 (2), 87-90 , 2008 2008 Citations: 11