Designing a Lightweight Deep Learning –based Technique for HSI Segmentation
Implementing Organization
Indian Institute of Technology (IIT) Indore https://www.iiti.ac.in/
Principal Investigator
Dr. Puneet Gupta
Indian Institute of Technology (IIT) Indore https://www.iiti.ac.in/
puneet@iiti.ac.in
CO-Principal Investigator
Shri Ankur
Garg,
SAC
Ahmedabad
Indian Institute of Technology (IIT) Indore https://www.iiti.ac.in/
Project Overview
Hyperspectral imaging (HSI) offers detailed spectral data for agriculture and medicine but faces challenges like noise and high computational costs. Traditional and deep learning methods often struggle with complexity or efficiency. This proposal introduces a lightweight DL-based segmentation technique using a novel cube-based framework. By dividing HSI data into smaller spectral-spatial cubes and employing a fusion mechanism, the model maintains accuracy and coherence without heavy model compression. Optimized for edge computing, this approach ensures low computational overhead, making it ideal for real-world, resource-constrained environments.
Achievements
The project being undertaken by ISRO in collaboration with academia is directly or indirectly beneficial to the society.