PlantTraits2023

Overview

This competition aims to predict plant properties - so called plant traits - from citizen science plant photographs. Why are plant traits currently so relevant? Plant traits are plant properties that are used to describe how plants function how they interact with the environment.  Due to global change, the environment and hence the biosphere is being transformed at accelerating pace. However, we can hardly project on a global scale how plant traits and as such entire ecosystems will react to climate change because we do not have sufficient data on plant trait. A data treasure in this regard may be the very growing availability of citizen science photographs. There are already more than 20 million plant photographs available covering all ecosystem types and continents. The aim of the competition is to predict the plant traits from the plant photographs using deep learning-based regression models (e.g. CNNs). Any prediction that helps us to better understand the global change on ecosystems is helpful; so please help us exploring data treasure and the distribution of plant traits around the world!

The dataset includes more than 30,000 plant images from the iNaturalist project (obtained from www.gbif.org). Each plant images comes together with ancillary information on climate variables. The photographs together with the climate information (x) shall be used to predict plant traits (y). Information of the plant traits was acquired from the TRY database (https://www.try-db.org/ ).

Competition

• Start Date: February 15

• End Date: May 24

• Kaggle URL: https://www.kaggle.com/competitions/planttraits2023 

Organizers

Teja Kattenborn

Remote Sensing for Earth System Research (RSC4Earth, Leipzig University, Germany)

https://rsc4earth.de/authors/tkattenborn/