Science & Technology

IIT Gandhinagar Develops AI Based Framework for Automated Pollen Classification

Researchers at the Indian Institute of Technology Gandhinagar (IITGN) have developed an AI-driven framework for automated classification of pollen grain images using scanning electron microscopy (SEM) and machine learning. The study, published in Botany Letters, introduces a scalable and reproducible solution for pollen analysis with wide-ranging scientific applications.

By : Ritika 2026-02-12 07:35:52

Gandhinagar - Researchers from the Indian Institute of Technology Gandhinagar (IITGN) have developed an artificial intelligence-based framework that enables automated classification of pollen grains using scanning electron microscopy (SEM) images and advanced computer vision models. The study combines microscopy, machine learning, and a web-based database to improve the speed and accuracy of pollen identification while reducing reliance on manual annotation.

The team created MPalyn (Medicinal Pollen and Palynology SEM Database), an open-access web application that organizes pollen images and associated species data. The dataset includes SEM images from 28 medicinal plant species collected from the IITGN campus, with 269 images used for segmentation and 5,842 images for classification. The researchers further implemented the YOLOv11n model for high-resolution image extraction and found that the Vision Transformer (ViT) model achieved the highest classification accuracy.

The research was led by Jaidev Sanjay Khalane, a final-year undergraduate at IITGN, under the guidance of Dr. Subramanian Sankaranarayanan, Assistant Professor in the Department of Biological Sciences and Engineering and Principal Investigator at the Plant Molecular & Developmental Cell Biology (PMDCB) Laboratory. Dr. Nilesh Gawande and Dr. Shanmuganathan Raman also contributed to the interdisciplinary study, integrating biological sciences with computer science and engineering.

The AI-powered framework has broad applications in agriculture, allergen source identification, biodiversity monitoring, plant taxonomy, paleoecology, archaeology, and climate studies. The researchers noted that improved algorithms can help reduce errors in identifying morphologically similar pollen species, enabling large-scale automated analysis and advancing research in ecological and medical sciences.

Click Here for More Science & Technology

Tags: # Science & Technology Gujrat