Publications
Our published research in top-tier journals and conferences.
A Deep Learning Approach to Early Drowning Detection for Child Safety using ResNet and Flower Pollination Algorithm
IEEE Transactions / Applied Computer Vision, 2026
This paper presents an automated computer vision framework using customized ResNet architectures and Flower Pollination optimization algorithms for real-time drowning detection in swimming pools to ensure child safety.
Modelling of Innovative Approaches for Drowning Prevention: Customized CNNs and Optimization of Binary Chimps for Early Detection
Springer Visual Computing Journal, 2026
Novel deep learning pipeline utilizing customized Convolutional Neural Networks (CNNs) coupled with Binary Chimps Optimization for early aquatic surveillance and emergency alerting.
Deep Neural Networks for Medical Image Classification: Automated Breast Cancer and COVID-19 Diagnosis
IEEE Journal of Biomedical and Health Informatics, 2025
An end-to-end explainable AI medical imaging model evaluating X-Ray and ultrasound images for highly accurate multi-class pathology classification.
Deep Attention-Guided Network for Medical Image Segmentation
IEEE Transactions on Medical Imaging, 2025
We propose a novel attention-guided deep neural network architecture for precise medical image segmentation. Our approach incorporates multi-scale attention mechanisms that adaptively focus on relevant anatomical structures while suppressing backgrou...
A Comprehensive Survey of Vision Transformers in Medical Imaging
ACM Computing Surveys, 2025
This survey provides a comprehensive review of Vision Transformer (ViT) architectures and their applications in medical image analysis. We systematically categorize over 200 recent works across modalities including X-ray, CT, MRI, and histopathology....
GAN-Based Synthetic Data Augmentation for Rare Disease Classification
Medical Image Analysis, 2024
We address the critical challenge of limited training data in rare disease classification by proposing a conditional GAN framework for generating high-fidelity synthetic medical images. Our approach generates diverse, clinically realistic images that...
Robust Face Recognition Under Occlusion Using Part-Based Feature Fusion
Pattern Recognition, 2024
This work proposes a part-based deep feature fusion method for face recognition under partial occlusion. By decomposing facial features into local part representations and employing an adaptive fusion strategy, our method achieves robust recognition ...
Multi-Spectral Satellite Image Classification Using Hybrid CNN-Transformer Architecture
Remote Sensing of Environment, 2024
We present a hybrid CNN-Transformer model for multi-spectral satellite image classification that combines the local feature extraction capability of CNNs with the global context modeling of transformers. Our method achieves 96.8% overall accuracy on ...