Analysis of Machine Learning Algorithms for Crop Mapping on Satellite Image Data

Crop classification is main area of our planet for understanding the agricultural cover. Studies via satellite imagery are often limited to public data with low revisit rates and/or coarse spatial resolution. However, a recent surge in satellite data from new-aerospace companies provides daily imagery with relatively high spatial resolution. With high revisit rates in satellite image capture enable the incorporation of temporal information into crop classification schemes. With high cadence temporal information just now becoming available, there is plenty of room to explore the data and methods for classification [60].Crop mapping methodology is used for the monitoring of various crop types. These methodology is depend on a large space of satellite imagery and different time series data values which is use in supervised classifiers such as Support Vector Machines (SVMs) and Random Forest (RF)[1]. These classifiers are applied at three unique degrees of crop terminology order and compare the result with accuracy and execution time. SVM gives ideal execution and demonstrates essentially better than RF for the least level of the classification. The significance of information factors such as Near Infrared (NIR), vegetation red-edge, and Short-Wave Infrared (SWIR) multispectral groups, and the Normalized Difference Vegetation (NDVI) and Plant Senescence Reflectance (PSRI) are used during cutting edge crop phenology stages and crop mapping [2].