A simple and quick approach to estimate soil parameters from hyperspectral images

About

This repository refers to the challenge proposed by the International Telecommunication Union (ITU), titled "GeoAI Challenge Estimating Soil Parameters from Hyperspectral Images by ITU". The objective of the competition is to estimate the quantity of potassium (K), phosphorus pentoxide ($P_2O_5$), magnesium (Mg), and pH present in the soil from hyperspectral images captured in agricultural areas in Poland.

Files

In the "report" folder, a detailed explanation of the methodology used during the competition is provided. The report is available in both English and Portuguese.

In the "src" folder, the corresponding source code for the methodology described in the report is available.

Finally, in the "submission" folder, the final submission file corresponds to the predictions made by the model for the test set.

Execution Environment

The provided code was executed in the Kaggle environment. To run it locally or in another environment, it is necessary to modify the file paths in the "main.ipynb" file and also download the dataset.