DeepTrees Symposium: Deep Learning and Remote Sensing in Tree Monitoring

Contribution List

7. Challenges in Crown Spread and DBH Estimation from Segmented Tree Crown Polygons

Session 1

Deeptrees updates

Tree traits such as crown area (CA), crown radius (CR), and diameter at breast height (DBH) provide insight into a tree’s growth dynamics, biomass, and the overall biophysical and biochemical processes of ecosystems. Accurately measuring these traits is essential for carbon estimation, urban tree management, and biodiversity assessments. The allometric relationships between CA, CR, and DBH are...

9. Characterizing the high latitudinal treeline ecotone using multi-source remote sensing data

Session 3

Potential use cases

The term treeline refers to the climatic limit of tree growth, occurring where seasonalmean temperatures decline to around 6 °C with a growing season of at least 3months (Körner 2020). Thereby, the treeline defines the boundary of the tree as a lifeform. However, treelines are shaped by various environmental factors and appear indifferent forms, typically classified as diffuse, abrupt, islands...

1. deadtrees.earth - A Database of Centimeter-Scale Aerial Imagery as Reference Data for Mapping Global Tree Mortality

Session 2

Results from the community

Today, excessive tree mortality is a global concern. Accurate monitoring for ecological insights and environmental predictions is urgently required. Earth observation data, combined with supervised machine learning, offer a promising approach to map tree mortality over time.
However, global-scale machine learning requires broad training data covering a wide range of conditions and forest...

6. DeepTrees: Deep-Learning based spatiotemporal tree inventorying from public orthoimages

Session 1

Deeptrees updates

Field-based methods for tree inventorying are time-consuming, labour-intensive, and often expensive. With the advent of remote sensing technologies and deep learning, it is now possible to leverage digital public orthoimages and artificial intelligence to automate and enhance the tree inventorying process [1].

The DeepTrees project aims to develop and implement deep learning models for...

3. Fine-tuning and evaluating deep learning models for tree crown segmentation

Session 2

Deeptrees updates

The DeepTree project uses deep learning models to segment tree crowns in orthoimages. Building on pretrained models [1], we fine tune them with labeled orthoimages from the city of Halle (Saale), Germany. We present a deep learning model that is optimized for segmenting tree crowns in orthoimages from this region.
The deep learning model is evaluated based on the city’s registry of trees [2]...

5. Integrating UAV Imagery and LiDAR Data for Tree Structural Properties

Session 2

Results from the community

Forests cover approximately 31% of the global land area and are essential for biodiversity, supporting 80% of terrestrial species. However, these ecosystems are vulnerable to human-driven climate change, necessitating automated systems to monitor structural changes at the individual tree level and assess forest responses to climate anomalies. UAV-based tree canopy detection is valuable for...

8. Seeing the trees in the forest (and other places)

Session 1

Opening Remarks + Keynote

Why do we want to map trees and what are the obstacles on the road? And what is a tree by the way? This introductory talk will discuss some of the challenges, opportunities and applications of tree detection and delineation in very high resolution image data.

2. SELF-SUPERVISED BACKBONES FOR FOREST REMOTE SENSING

Session 2

Results from the community

Self-Supervised Learning (SSL) presents an opportunity to leverage large volumes of unlabeled data to improve outcomes for tasks for which training data is limited. This is the case in forest remote sensing where high quality annotated datasets are often too small to fully take advantage of data hungry deep learning models. Recognizing this, we have pre-trained two classes of neural network...

10. The UFZ forest monitor

Session 2

Results from the community

The UFZ Forest Condition Monitor is a project funded by the Helmholtz Association within the [Land Cover & Dynamics Working Group][1] of the [Remote Sensing Department][2] at the [Helmholtz Centre for Environmental Research - UFZ][3] in Leipzig. The UFZ Forest Condition Monitor provides (remote sensing-based) comprehensive annual/seasonal derivations of forest condition continuously since...

4. Using AI to detect and identify tree species from satellite images in Hawaii

Session 3

Potential use cases

The application of artificial intelligence (AI) in ecology presents significant opportunities for enhancing biodiversity monitoring and informing conservation efforts, particularly in tropical regions. This project aims to apply the DeepTrees pipelines to a tropical setting at two spatial scales.

First, at a large scale, we aim to identify key species of interest for restoration and...