| Project ID |
BITS-SRIP/041127/2026 |
| Project Title |
Enhanced Neural Network for classification of Hyperspectral Imagery ensembled models. using |
| Project Description |
Project Objectives
Space-borne hyperspectral sensor dataset collection and preprocessing: To collect hyperspectral signatures of rooftop materials and water resources from space-borne hyperspectral sensors over the ROIs. To calibrate and map hyperspectral signatures from these sensors with the field/in-situ measurements. To create sensor-exclusive and integrated hyperspectral signature libraries.
Implementation of ensembled machine learning model-based classification techniques: To develop state-of-the-art DL –based ensembled NNs. To further develop foundation models (for feature representation) and classification models (for HIS classification). Additionally, preprocessing steps such as data augmentation, data normalization, and feature-space reduction will be integrated within the DNN framework.
Evaluation metrics and reporting: To evaluate the performance of DL-based ensembled NN models, performance indicators such as accuracy, precision, recall, F1-score, AUC-ROC, and DICE coefficient will be used. To prepare a report that highlights quantitative and qualitative results, and discussions drawn on different scenarios, dataset characteristics, and computational efficiency, etc. |
| Project Discipline |
Student must have clear understanding of Machine learning, Deep learning, Python. He/she may be preferred if their exposure to image processing is significant. Soft skills such as word processing, PPT presentation, and knowledge of Latex is helpful. Coding skills and working knowledge of ML/DL/Cv libraries in Python is also preferred. |
| Faculty Name |
Gopal Singh Phartiyal |
| Department |
Department of Computer Science & Information Systems |
