Abstract:

In the last few years, Agricultural research has been developed faster from different computational technologies that last resources we can have the convenience of how agriculture can be grown. For the classification of lands, we have used land satellite images from which we have trained with images like Forests lands, Agriculture lands, Urban lands, and range performance of these classifiers are compared. There are only some studies with various training samples of some remote sensing images. Mostly the Sentinel-2 multispectral imager. Using sentinel-2 image data we have trained and compared the working of RF, KNN, and SVM classifiers for land. Along with the t algorithm, we have also compared deep learning highest like CNN. In Vietnam around the red river delta in the area of 30*30KM^2 land covers can be classified using 14 training sample sizes which include appropriate and inappropriate around 50 to till 1250 pixels. The high accuracy can be observed through all declassification from 90% to 95%. CNN produced the nation’s overall dead-end training sample sizes. According to this, the next high accuracy produced after the CNN algorithm is SVM. Given the training sample sizes, this yielded the best accuracy with the least sensitivity. both machine learning and deep learning algorithms are used and compared. Comparing the deep learning algorithms produced high accuracy.

Description:

.