Developed GRANet-Lite, a lightweight ghost-residual attention CNN for accurate and explainable breast cancer classification, optimized with histopathological preprocessing strategies and validated with Grad-CAM++ and Score-CAM.

Proposed CBAM-Xception, an attention-guided deep learning model for accurate and interpretable skin cancer classification.

A comprehensive framework for classification using optimized MRI and fusion.

A multi-function robotic system for remote monitoring and control.

Using structural MRI and DL to detect neurodegenerative diseases.

Developed a stacking ensemble model using VGG16 and machine learning classifiers on color-normalized PBS images for accurate early detection and subtype classification of ALL.

Proposed ResSplit-KAN, a lightweight and explainable deep learning model for ALL classification, integrating multi-scale residual learning with KAN to achieve high accuracy and interpretable results.

Proposed DermFusionNet, an attention-guided ensemble model combining multiple pretrained backbones with CBAM, achieving highly accurate and interpretable skin lesion classification.

Developed SmartBin, a real-time waste classification system using MobileNetV3Small integrated with Arduino-based hardware for efficient and automated waste sorting.

Developed an explainable skin cancer diagnosis framework combining Xception, CBAM, and Grad-CAM++ with a fine-tuned vision-language model for accurate and interpretable predictions.