The EcoVision Lab in the Department of Mathematical Modeling and Machine Learning (DM3L) at University of Zurich is seeking applications for a Doctoral Candidate in computer vision and machine learning for developing novel deep learning methods for plant and animal species identification from camera-trap and crowd-sourced imagery.
We offer an exciting and stimulating environment to study and work in. The University of Zurich has several internationally recognized research groups dedicated to data science, machine learning, remote sensing, biodiversity, and more broadly ecology. We also collaborate with several other institutions and companies in the fields of computer vision, machine learning and earth observation, in Switzerland and abroad. The EcoVision Lab is a member of, the ETH AI Center, the UZH Digital Society Initiative, the UN-ETH partnership, and the ETH for Development Center (ETH4D).
PhD Position in Deep Learning for Biodiversity Monitoring (EU Horizon Project NextBON)
We are looking for a highly motivated PhD candidate to join a large European research initiative aimed at transforming how biodiversity is monitored across Europe. The position is part of NextBON, a Horizon Europe project coordinated by the University of Copenhagen and involving more than 20 leading research institutions across Europe. The project's goal is ambitious: to build a harmonised, policy-relevant biodiversity observation network that can support environmental decision-making at national and European levels. While the PhD candidate will be based in Zurich, the candidate will be encouraged to do a research stay at NIBIO in Norway.
Why This Project Matters
Biodiversity monitoring technologies have advanced rapidly in recent years. Earth observation satellites, environmental DNA (eDNA) metabarcoding, automated sensor networks, and AI-based species identification now generate unprecedented amounts of ecological data. However, these methods are often used in isolation, with limited coordination or validation across countries. As a result, their full potential for informing environmental policy and management remains underutilized. NextBON aims to change this. The consortium is developing a validated and harmonised blueprint for large-scale biodiversity monitoring that can be adopted by EU Member States and research infrastructures beyond the project's lifetime.
The Scientific Vision
NextBON will establish a three-tier biodiversity observation network.
- Tier 1: Large-scale monitoring using satellite-based Earth observation
- Tier 2 & 3: Local, in-depth ecological validation through in situ observations at carefully-selected sites
A dedicated Multi-Criteria Decision Analysis toolkit will guide where and how monitoring sites are selected, ensuring ecological and geographic representativeness across Europe. All monitoring workflows are explicitly linked to policy requirements and are co-developed with major international partners such as GBIF, LifeWatch, and BioAgora to ensure long-term operational impact.
Doctoral Candidate in computer vision and machine learning for developing novel deep learning methods for plant and animal species identification from camera-trap and crowd-sourced imagery
Your responsibilities
Within NextBON, the EcoVision Lab (in close collaboration with NIBIO in Norway, group of Stefano Puliti) will focus on advancing biodiversity monitoring at the most detailed level - Tier 3 sites. As a PhD candidate, you will develop novel deep learning and computer vision methods to transform large-scale photo and video datasets into Essential Biodiversity Variables (EBVs).
Your research will include:
- Developing deep learning models for species detection and identification
- Estimating species abundance and phenological stages
- Producing calibrated uncertainty estimates for ecological predictions
- Training models on heterogeneous data sources (, camera traps, GBIF, LUCAS, NFI records)
- Exploring multimodal fusion with environmental DNA, passive acoustics, and satellite data
The ultimate goal is to generate spatially explicit, policy-relevant biodiversity indicators grounded in robust machine learning methodology.
Research Freedom & Methodological Innovation
The project offers significant freedom to explore impactful methodological directions in modern AI, including: self-supervised learning, multimodal learning, geospatial representation learning, uncertainty estimation, interpretability and explainability. We aim for high-impact publications both in machine learning venues (, CVPR, ICCV, ECCV, ICLR, NeurIPS) and leading interdisciplinary journals such as Remote Sensing of Environment, ISPRS Journal, and Nature Sustainability.
Your profile
Why Join?
This PhD offers:
- A j4id10110490a j4it0311a j4iy26a