PEER-REVIEWED CONFERENCE PAPERS & AI RESEARCH.
Published scientific research focusing on Explainable Deep Learning (Grad-CAM), Multi-Task Convolutional Networks, and Parameter-Efficient Vision Transformers (ViTs) with registered global DOIs.
FruitQ-GradeX: Determining Fruit Quality and Grading with Explainable Deep Learning
This paper presents a novel multi-task deep learning framework for simultaneous fruit classification and quality assessment using a multi-headed Convolutional Neural Network (CNN). The proposed model achieves state-of-the-art performance on a curated dataset of four Indian fruits (apple, banana, guava, and orange) with two quality classes (good and bad), achieving 98% accuracy in fruit classification and 99% accuracy in quality detection with Grad-CAM visual heatmaps.
Hyperspectral Fruit and Vegetable Classification Using Convolutional Neural Networks with EfficientNetB3
Deep learning has revolutionized the fruits and vegetables classification in the food and farming sectors. This study presents an EfficientNetB3 model trained on a comprehensive Kaggle dataset of 36 distinct agricultural classes, achieving 99.29% training accuracy and 97.21% test accuracy. Using transfer learning and real-time predictions with Streamlit and OpenCV, the model enables automated sorting and quality inspection in supply chain logistics.
Cropsense: Explainable Deep Learning Framework for Accurate Quality Detection in Solanaceous Crops
This paper presents a novel multi-headed convolutional neural network (CNN) architecture for simultaneous classification of potato and tomato crops and their quality assessment (healthy vs. diseased). The model achieves 99.9% accuracy in crop classification and 98.5% accuracy in quality assessment, with Grad-CAM visualizations providing visual interpretability for agricultural diagnostic decisions.
An Explainable Deep Learning Approach for Quality Assessment in Solanaceous Crops
This paper presents a multi-headed hybrid CNN-ViT architecture for simultaneous crop classification and quality assessment in Solanaceous crops (potato and tomato). Utilizing a shared Vision Transformer backbone, the model reduces parameters by 30% while achieving 98.5% quality detection accuracy, supported by Grad-CAM interpretability visualizations.