Experiment examples

The repository contains examples of how to use Malpolon for different scenarios. The examples are organized by the type of user and the type of dataset used.

I. Custom train scenario

I have a dataset of pre-extracted image patches or raster files (e.g. bioclimatic data, satellite rasters…) of my own and I want to train a deep image neural network model on it. I want to be able to easily customize the training process and the model architecture.

Drop and play : I have a file (.csv) of Presence/Absence (PA) or Presence Only (PO) observations and I want to train a model on different environmental variables (rasters, satellite imagery) without having to worry about the data loading and on-the-fly extraction.
Custom dataset : I have my own dataset consisting of pre-extracted image patches and/or rasters and I want to train a model on it.

Sentinel-2A

See Sentinel-2a (training) GitHub README 🔗

MicroGeoLifeCLEF2022

See MicroGeoLifeCLEF2022 (training) GitHub README 🔗

II. Inference scenario

I have an observations file (.csv) and I want to predict the presence of species on a given area using a model I trained previously and a selected dataset or a shapefile I would provide.

III. Benchmarks scenario

I want to compare the performance of different models on a given known dataset; or I am a potential kaggle participant on the GeoLifeClef challenge. I want to train a model on the provided datasets without having to worry about the data loading, starting from a plug-and-play example.

GeoLifeClef2022 : contains a fully functionnal example of a model training on the GeoLifeClef2022 dataset, from data download, to training and prediction. Comes with datasets and datamodule.
GeoLifeClef2023 : contains dataloaders for the GeoLifeClef2023 dataset (different from the GLC2022 dataloaders). The training and prediction scripts are not provided.
GeoLifeClef2024_pre_extracted: contains a fully functionnal example of a model training on the GeoLifeClef2024 pre-extracted dataset, from data download, to training and prediction. Comes with datasets and datamodule for species prediction as well as habitats prediction using a direct approach (raster-based).