The purpose of our investigation surrounded the specific question of how to classify land cover types using a neural network with at least eighty six percent accuracy. Using ground truthed data and spectral signatures we were able to validate Landsat band data. We then set out to prove that we could train a neural network to classify two land cover types using the spectral signatures in a multilayer perceptron neural network. Our team was able to successfully train a neural network to use our collected data samples to classify two different land cover types with a hundred percent accuracy.

Access our entire data set collection below
View Data Set