Performance Analysis of CNN, AlexNet and VGGNet Models for Drought Prediction using Satellite Images
Shilpa Chaudhari, Vandana Sardar, Dhananjay Rahul, M. C. Chandan, Mahantesh Shivan Shivakale, K. Harini
Abstract
Drought is a naturally occurring event in particular geographical area. Droughts are classified as one of the major naturally occurring hazards causing severe impact on the entire environment as well as economy of the countries throughout the globe. Droughts are being aassociated with weather conditions that cannot be monitored only using weather data, strictly because these obtained data are likely to be incomplete, infrequent and ill-timed. Naturally predicting these kinds of events is not preferable and effective. Taking deep learning and artificial intelligence advancement into an account, we have analysed and compared the CNN, AlexNet and VGGNet using satellite images and indices calculated from those images of a particular geographical area and defined accuracy and performance of each models towards the data and for each type of drought indices like Normalized Difference Vegetation Index (NDVI), Soil Adjusted Vegetation Index(SAVi), Enhanced Vegetation Index(EVI) and Atmospherically Resistant Vegetation Index(ARVI), the models performance has been demonstrated in this paper.