Student assistant – Tree species detection using deep-learning

at  ETH Zrich

20%-30%, ZURICH, FIXED-TERM

The chair of Forest Resources Management (FORM) is seeking a motivated student to assist in the development of Deep Learning approaches for tree species identification. This project focuses on mapping individual trees and quantifying their species, which are critical tasks for forestry management. Using aerial RGB imagery, we aim to create a cost-effective, automated system for detecting and identifying tree species, with broad applications in forest monitoring. In collaboration with colleagues from Austria, the successful approach will be implemented in a protected area in Austria, with the goal of quantifying tree species diversity.
More details about the project can be found here: Tree Species Identification Project.

JOB DESCRIPTION

You will contribute to a project focused on automating the detection of individual tree species in forests using deep learning. Specifically, you will:

• Focus on the application of deep learning techniques in Python to process spatial and aerial data.
• Integrate various datasets, such as tree species annotations, climate, and topography, into deep learning algorithms.
• Test deep learning models (Transformers and CNNs) for optimal accuracy using large datasets that include over 11,000 tree species annotations from across Switzerland, along with climatic, topographic, and lidar data.
• Test the best algorithm developed for the identification of tree species over a protected area in Austria using the most accurate deep learning model.

Please refer the Job description for details