Soumik Banerjee

Assistant Professor Trainee

soumikbanerjee@rvu.edu.in Website

About
Publication & Works
Research Summary
Awards & Achievements

With a strong foundation in data science and applied artificial intelligence, I have been deeply engaged in exploring the intersection of analytics, automation, and intelligent systems. My work has involved developing AI-driven solutions for domains ranging from demand forecasting and business optimization to generative AI and autonomous research systems.

Before transitioning into academia, I worked on several data-centric projects that combined research and industry practices — from building predictive models and deep learning pipelines to integrating large language models (LLMs) for automation and decision support. My academic background includes a Master’s in Data Science, where I focused on bridging theoretical understanding with real-world problem-solving across image recognition, natural language processing, and multimodal analytics.

My current areas of research interest include generative AI, machine learning, and intelligent automation, with an emphasis on creating adaptive systems that enhance human decision-making, interpretability, and creativity. I am also interested in explainable AI (XAI) and its role in ensuring transparency and trust in deep learning models.

I have co-authored a publication titled “PCNN-based grape leaf disease detection using MobileNetV2 and ViT with XAI,” presented at the 5th International Conference on Data Science, Computation, and Security (IDSCS 2024) and published in Lecture Notes in Networks and Systems (Springer, SCOPUS Q4).

Stay grounded and make it happen.

PCNN-Based Grape Leaf Disease Detection Using MobileNetV2 and ViT with XAI

This study presents an innovative approach to grape leaf disease detection by integrating a parallel Convolutional Neural Network (PCNN) architecture combining MobileNetV2 and Vision Transformer (ViT). Trained on a dataset of 4,639 images, the model achieved an impressive accuracy of 99.20%, with precision, recall, and F1-score values of 0.9957. To enhance model interpretability, Explainable AI (XAI) techniques such as Grad-CAM and LIME were employed, providing insights into the decision-making process. The research addresses the significant economic impact of grape diseases, estimated at $3 billion annually, by offering a scalable solution for early detection and efficient pesticide application, thereby promoting sustainable agricultural practices. Source link: https://link.springer.com/chapter/10.1007/978-981-96-4880-1_22

Developed a PCNN-based grape leaf disease detection framework combining MobileNetV2 and Vision Transformer (ViT).
Achieved 99.2% accuracy with precision, recall, and F1-score above 0.99 on real-world grape leaf datasets.
Integrated Explainable AI (XAI) techniques such as Grad-CAM and LIME for model interpretability and transparency.
Addressed real-world agricultural challenges, enabling early disease detection and efficient pesticide application.
Demonstrated a scalable AI solution for precision agriculture, bridging deep learning research with practical applications.
Research Interest
Deep Learning (DL): Image recognition, speech emotion analysis, and multimodal analytics.
AI Agents & Intelligent Automation: Developing AI-driven systems for predictive modeling, decision support, and workflow automation.
Large Language Models (LLMs) Integration: Leveraging LLMs for automation, research, and intelligent decision-making.

Generative AI: Exploring AI solutions for adaptive, creative, and practical applications across domains.

Achieved Batch Highest CGPA in Bachelor's: Achieved the highest CGPA of 9.72/10 in their Bachelor's degree, demonstrating academic dedication.

Recipient of National Scholarship Portal (NSP) Scholarship: Received the NSP Scholarship, placing me among 5000 science students nationwide to earn this recognition.