Mapping and Classification of Field Margin Vegetation using High Resolution Satellite Imagery and Deep Learning Models in a Tropical Landscape

This data set is a supplementary material to the publication ''Mapping and Classification of Field Margin Vegetation using High Resolution Satellite Imagery and Deep Learning Models in a Tropical Landscape" (Prakash et al., 2026). This dataset contains the training, validation, and testing data used for the deep-learning-based classification of Field Margin Vegetation (FMV) using high-resolution WorldView-3 satellite imagery. The dataset was prepared for mapping field margin vegetation in agricultural landscape. The original Worldview-3 satellite imagery was obtained under a commercial data license and therefore cannot be redistributed publicly. The datasets provided here contain only the derived image chips and corresponding reference labels that are permitted to be shared.

The data are organized in three subsets:
Training dataset: Used for training the deep learning model (U-Net and Deep LabV3+)
Validation dataset: Used for monitoring model performance and model selection during training.
Testing dataset: Used independently to evaluate the performance and generalisation capability of the trained model.

Data Preparation:
The input data were derived from WorldView-3 multispectral imagery and divided into smaller chips for model development. Corresponding reference masks were prepared for the classification of FMV.

The datasets were used together with the source code provided in the associated repository to ensure transparency and reproducibility of the model-development workflow.

For further information, please refer to the Readme.txt file.