Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/froukje/ai4eo-hyperview

ML challenge to predict soil parameters from hyperspectral images.
https://github.com/froukje/ai4eo-hyperview

challenge earth-observation hyperview machine-learning random-forest

Last synced: about 1 month ago
JSON representation

ML challenge to predict soil parameters from hyperspectral images.

Host: GitHub
URL: https://github.com/froukje/ai4eo-hyperview
Owner: froukje
Created: 2022-09-26T19:09:28.000Z (about 2 years ago)
Default Branch: main
Last Pushed: 2022-10-17T21:13:12.000Z (about 2 years ago)
Last Synced: 2024-01-27T01:38:13.693Z (11 months ago)
Topics: challenge, earth-observation, hyperview, machine-learning, random-forest
Language: Jupyter Notebook
Homepage:
Size: 22.1 MB
Stars: 1
Watchers: 1
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

        # AI4EO Hyperview

This repository contains the contribution of the team EagleEyes to the [AI4EO Hyperview Machine Learning Challenge](https://platform.ai4eo.eu/seeing-beyond-the-visible)

**Overview**

The objective of the AI4EO HYPERVIEW challenge is to predict agriculturally relevant soil pa-

rameters (K, Mg, P2O5, pH) from airborne hyperspectral images. We present a hybrid model fusing

Random Forest and K-nearest neighbor regressors that exploit the average spectral reflectance, as

well as derived features such as gradients, wavelet coeﬀicients, and Fourier transforms. The solution

is computationally lightweight and improves upon the challenge baseline by 21%.

**This Repository contains the following:**

* A jupyter notebook containing the final solution can be found in [final-submission](final-submission).

* Conference paper: Kuzu, R.S., Albrecht, F., Arnold, C., Kamath, R.,Konen, K. (2022), [Predicting Soil Properties from Hyperspectral Satellite Images](hyperview_for_ICIP_camera_ready_eagleeyes.pdf), in 29th IEEE International Conference on Image Processing (IEEE ICIP 2022), Bordeaux, France.

* The folder [notebooks](notebooks) contains some jupyter notebooks for data exploration and first simple models

* We explored several other approaches, which can be found in [hyperview](hyperview)

    * Different Neural Network architectures (based on keras) [NN keras](hyperview/keras])

    * [PSELTAE model](https://github.com/VSainteuf/pytorch-psetae) (based on pytorch-lightning) [PSELTAE](hyperview/pytorch_lightning) 

    * [Random Forest and XGBoost models](random-forest)

![hyperview_award.png](hyperview_award.png)

A more detailed README can be found [here](https://github.com/ridvansalihkuzu/hyperview_eagleeyes)