Explainable few-shot learning workflow for detecting invasive and exotic tree species

This is the supporting dataset of research work: Explainable few-shot learning workflow for detecting invasive and exotic tree species. (Link to be added after the publication) In this research, we presents a workflow that tackles both challenges by proposing an explainable few-shot learning workflow for detecting invasive and exotic tree species in the Atlantic Forest of Brazil using Unmanned Aerial Vehicle (UAV) images. By integrating a Siamese network with explainable AI (XAI), the workflow enables the classification of tree species with minimal labeled data while providing visual, case-based explanations for the predictions.

The workflow is accessible in this GitHub repository (Link to be added after the publication). The required dataset of this workflow in provided in this Zenodo repository.

This dataset repository has the following contents

• uav_img.zip: the UAV orthomosaic image (.tif) of the study area used in this research, with related metadata

• tree_labels.zip: the labels of trees created by expert, available in .shp and .gpkg

• cutouts.zip: tree cutouts used in this study. They are two sub-directories:

  • all_cutouts: all the candidated cutouts from three sources. See the README.md file insisde this folder for more information
  • selected cutout: the manually selected cutouts from all cutouts used for training.

• training_pairs_20000.zarr.zip: training data created for base network traning. It is created by pairing the selected cutouts.

• netflora.zip: Netflora workflow prediction results
• optimized_models.zip: Optimized base models (shallow and deep) and refined models with different shots/fold setup.
• n_fold_x_validation.zip: data pairs for refinement traing, with n fold and x valiation setup.