Research Interests
- Artificial Intelligence
- Machine Learning and Deep Learning
- Distributed and Edge Computing
- Security, Privacy, and Trustworthy Computing
- Federated and Privacy-Preserving Learning
- Computer Vision and Image Analysis
- Remote Sensing and Hyperspectral Imaging
- Medical Imaging and Healthcare AI
- Secure Multimedia Processing
Research Projects
Competitive research projects funded by CeRRF, Deakin University, Australia and the Institute of Research and Training (IRT), HSTU, focusing on hyperspectral imagery, agricultural land identification, and information security.
Emerging Rock Detection Using Deep Learning
Funded by: Centre for Regional and Rural Futures (CeRRF), Deakin University, Australia
Position: Research Assistance (RA)
Duration: 06(Six) Months
This project focuses on detecting emerging rocks in drone imagery using the Deep Learning (DL) paradigm.
A Hybrid Spectral–Spatial Dimensionality Reduction Method to Extract Agriculture Fields
Funded by: Institute of Research and Training (IRT), HSTU, Dinajpur
Position: Principal Investigator (PI)
Fiscal Year: 2022–2023
Develops a hybrid spectral–spatial dimensionality reduction pipeline for satellite hyperspectral imagery, enabling accurate extraction of agricultural fields while keeping the computational cost manageable for large-scale remote sensing analysis.
Extracting Agriculture Lands from Hyperspectral Imagery Using Deep CNNs
Funded by: Institute of Research and Training (IRT), HSTU, Dinajpur
Position: Principal Investigator (PI)
Fiscal Year: 2021–2022
Explores deep convolutional neural network architectures for precise segmentation and classification of agricultural land from hyperspectral satellite data, supporting data-driven decision making in agriculture and land management.
Dimensionality Reduction for Identifying Agricultural Land from Hyperspectral Images
Funded by: Institute of Research and Training (IRT), HSTU, Dinajpur
Position: Principal Investigator (PI)
Fiscal Year: 2020–2021
Proposes an efficient dimensionality reduction strategy tailored to agricultural land classification, balancing discriminative power with computational efficiency for large-area monitoring.
Improved LSB Watermarking Using Logistic Map and AES Encryption
Funded by: Institute of Research and Training (IRT), HSTU
Position: Co-Principal Investigator (Co-PI)
Fiscal Year: 2019–2020
Security-focused research on robust image watermarking that combines LSB techniques with a one-dimensional logistic map and AES encryption to strengthen resistance against steganalysis and unauthorized tampering.