The School of Computer Science is pleased to present…
Unlocking the Potential of Simulated Hyperspectral Imaging in Agro Environmental Analysis: A Comprehensive Study of Algorithmic Approaches
MSc Thesis Proposal by: Ashwitha Basani
Date: Friday, 10 May 2024
Time: 10:00 AM
Location: Essex Hall, Room 122
Abstract:
This new study tackles the powdery mildew challenge in agriculture by focusing on identifying and evaluating the severity of disease in tomato plants. The research combines imaging and advanced deep learning methods to develop a technique that transforms RGB images into Simulated Hyperspectral Images (SHSI) and performs spectral and spatial analysis of the plants for plant health assessment. The process begins with image preprocessing and feature extraction using the VGG16 model, ResNet50 and EfficientNet-B7. Then, a neural network generator model is used to convert RGB images into SHSIs, providing insights into the spectrum. This method enables the image analysis to perform assessments from SHSIs for health classification using NDVI values, which is meticulously compared with accurate hyperspectral data using metrics like mean absolute error (MAE) and root mean squared error (RMSE). This strategy enhances precision farming, environmental monitoring, and remote sensing accuracy. ResNet50’s architecture offers a robust framework for the spectral and spatial analysis required in this study, making it a suitable choice over VGG16 and EfficientNet-B7 for transforming RGB images into SHSI and performing health assessments of tomato plants. It also offers a scalable and practical approach for assessing tomato crops vegetation health and disease severity. The goal of this project is to improve practices and ensure food security by promoting crop quality on a global scale.
Thesis Committee:
Internal Reader: Dr. Dan Wu
External Reader: Dr. Mohamed Belalia
Advisors: Dr. Boubakeur Boufama and Dr. Shafaq Khan
