Habitats labelled data

This dataset has been created within the EU H2020 Natural Intelligence project (ID  101016970).

The dataset contains labeled pictures of typical and early warning species of four different habitats, divided into four subfolders.

Each subfolder contains:

1) species data images (.jpg files)

2) .txt files for each image containing the labelling information. The first value indicates the species (defined in the .yaml file), the remaining values describe the box vertices.

3) a single .yaml file containing the parameter used in the labelling and detection AI training code:

                  - path to the dataset root directory

                  - path to the the train images directory

                  - path to the validation images directory

                  - path to the test images directory

                  - number of classes (species)

The four subfolders are named after the four habitats, i.e.,

-              Dunes. This refers to EU habitats 2110 and 2120 in Italian dunes. The included species are:

                            o   Achillea maritima – typical species

                            o   Calamagrotis arenaria – typical species

                            o   Carpobrotus acinaciformis – invasive species

                            o   Eryngium maritimum – typical species

                            o   Pancratium maritimum – typical species

                            o   Thinopyrum junceum – typical species

-              Grasslands. This refers to EU habitat 6210* in the Italian Central Apennines. The included species are:

                            o   Asphodelus macrocarpus – early warning species

                            o   Dactylorhiza sambucina – typical species

                            o   Orchis morio – typical species

-              Forests. This refers to EU habitat 9210* in the Italian Apennines. The included species are:

                            o   Anemonoides nemorosa – typical species

                            o   Corydalis cava – typical species

                            o   Doronicum columnae – early warning species

                            o   Anemonoides ranunculoides – typical species

-              Screes. This refers to EU habitat 8110 and 8120 in the Italian Alps. The included species are:

                            o   Cerastium sp. – typical species

                            o   Luzula alpino-pilosa – early warning species

                            o   Saxifraga – typical species

                            o   Ranunculus glacialis – typical species

                            o   Geum reptans – typical species

                            o   Papaver alpinum – typical species

Researchers from a variety of disciplines can benefit from using this dataset because of its multidisciplinary scope. Botanists could evaluate the accuracy of this data as well as the habitat's conditions using the plant images that the robot captured, robotic engineers could test their AI algorithms for identifying and classifying different species using these data. The code used to label the images can be found here: https://github.com/ivangrov/ModifiedOpenLabelling