Missions

The preservation of biodiversity and its ecosystem services in urban areas is a major issue for territorial planning. Local authorities currently lack effective tools to implement viable solutions for promoting biodiversity.
In this context, analyzing urban landscapes and the functionality of green corridors is crucial to support the maintenance of urban flora and fauna. Many ecological connectivity studies rely on satellite or aerial imagery and land-use maps (e.g., BD TOPO), relying on 2D analyses. While effective, a two-dimensional approach does not account for the actual structure of vegetation or the ground conditions beneath the canopy—key parameters for assessing ecological connectivity quality. The Bio3DiverCity project aims to analyze urban ecological corridors in 3D.
With the advent of new data sources, such as the CO3D satellite and the IGN's upcoming 2025 high-density LiDAR campaigns, opportunities for 3D landscape monitoring and analysis are expanding. These datasets, combined with cost-effective photogrammetric or terrestrial LiDAR acquisitions, enable the study of landscape structures, particularly in urban environments and their interactions with vegetation.

Within our research consortium, recent efforts have led to the development of 3D shape and structure analysis tools (especially for LiDAR data), applied in various contexts (natural environments, urban infrastructure). However, two key challenges remain:
1. Characterizing large-scale 3D structures of vegetation and landscapes.
2. Integrating and calibrating multiple 3D data sources (terrestrial and aerial LiDAR).

This position is part of a work package within the Bio3DiverCity project, focusing on two research areas:
- Axis 1: 3D Data Acquisition and Calibration – Comparing terrestrial LiDAR (TLS) and aerial LiDAR (ALS) sensors.
- Axis 2: 3D Landscape Metrics Extraction – Analyzing habitat fragmentation and vegetation complexity using both morphological properties and unsupervised learning approaches (e.g., deep neural networks).

Activities

The researcher will collaborate closely with an interdisciplinary team to tackle the following challenges:
Axis 1: Consistency Between TLS and ALS Data
- Develop domain adaptation methods to harmonize data from terrestrial and aerial sensors.
- Evaluate the transferability of methodologies calibrated on TLS data to larger scales using ALS or high-density LiDAR.
Axis 2: 3D Descriptor Extraction and Analysis
- Adapt and train 3D neural networks (e.g., KPConv) for feature extraction from 3D data.
- Develop tools to compute metrics such as habitat fragmentation, vegetation complexity, and internal ecosystem structure.
- Design a user-friendly interface (e.g., a CloudCompare plugin) to make these tools accessible to researchers, urban planners, and businesses.

Skills

Essential Skills:
- Ph.D. in remote sensing, geomatics, computer science, or a related field.
- Strong expertise in 3D point cloud processing and LiDAR data manipulation.
- Proficiency in deep learning frameworks (e.g., TensorFlow, PyTorch).
- Solid programming skills in Python and/or C++ (experience with CloudCompare is a plus).
- Knowledge of domain adaptation or optimal transport methods would be advantageous.

Preferred Skills:
- Experience in user interface development or creating scientific software plugins.
- Understanding of urban ecosystems and environmental challenges in cities.
- Strong initiative, autonomy, and ability to work within a multidisciplinary team.

Work Context

The post-doctoral fellow will work at the Littoral - Environnement - Télédétection - Géomatique laboratory (UMR CNRS 6554 LETG) in Rennes. The LETG scientific field is environmental geography. This unit has expertise in human geography, physical geography and geomatics.