BigPlantSens –  Assessing Synergies of Big Data and Deep Learning for Remote Sensing of Plant Species

Project summary

Accurate mapping of plant species distributions is vital for conservation, agriculture, and forestry. Utilizing high-resolution UAV orthoimages and advanced pattern-recognition methods like convolutional neural networks (CNNs) enables precise species segmentation. Traditionally, training these models requires extensive data from labor-intensive field surveys, but citizen science platforms offer a more efficient alternative with millions of crowd-sourced plant photos and species labels. Leveraging this data can significantly enhance our ability to monitor biodiversity and track hundreds of plant species across diverse landscapes over time. Hence, BigPlantSens assess whether a series of plant species can be mapped without generating new training data and by only using preexisting knowledge from citizen science.