Post-doctoral position

Artificial intelligence for a
transnational acoustic biodiversity monitoring network

Application deadline: 15 Feb 2024        Location: Université de Toulon, La Garde

Start date: early 2024        Duration: 1.5 year (extendable up to 2.5y)

Supervisor: Prof. Ricard Marxer <ricard.marxer@lis-lab.fr>

Apply: HERE

Context

This post-doc position is funded by the EU Biodiversa+ project TABMON (Transnational Acoustic Biodiversity MOnitoring Network). Through a consortium of partners from Spain, France, Netherlands and Norway, TABMON will apply autonomous acoustic sensing to transnational biodiversity monitoring across a large latitudinal range in Europe. The collection, analysis and integration of acoustic biodiversity observations will be harmonised. This will provide a critical tool to fill existing gaps in reporting to EU directives and measuring progress towards policy targets in the EU Biodiversity Strategy for 2030.

Subject

The post-doc will participate in two main tasks of the project. The first one consists in deploying a network of passive acoustic monitoring stations in several natural national and regional parks in France. The goal of this action is to deploy the first transnational acoustic biodiversity monitoring network in Europe. We will use a state-of-the-art autonomous acoustic monitoring system, which has recently been validated in a national deployment across Norway [1]. Devices will record data continuously over a two-year period, uploading data over a mobile network connection to be analyzed in the cloud in real-time.

The second task is the development of AI methods to acoustically detect avian species and predict environmental biodiversity variables. This will consist in configuring and calibrating AI animal sound recognition to yield robust and balanced detection performance across many species. To achieve this, we will combine off-the-shelf AI [2] with newly-trained AI models [3] designed for our targeted species and habitats. We will develop a pre-processing and pre-filtering strategy to accelerate inference and reduce carbon footprint, and calibrate sound event detection sensitivity, based on in situ data. We will focus on ensuring that our algorithms cover all required taxonomic groups in our target habitats, and study coherence of inference across the wide geographic span of our field sites and bird migratory routes.

Profile

The candidate shall have the following profile:

• PhD in one of the following fields: machine learning, computer science, applied mathematics, statistics, signal processing
• Good English written and spoken language skills
• Programming skills, preferably in Python
• Experience in one of the main DL frameworks (e.g. PyTorch, Tensorflow)

• Notions in audio processing and/or object detection

Application procedure

In order to apply please complete the following application form: https://forms.gle/8oGjfypVBWk8QCng8 

For any questions regarding the position, please contact ricard.marxer@lis-lab.fr

Monthly Salary

Depends on the experience. The salary is compatible with the costs of living in Toulon (e.g., rent prices are 50% lower in Toulon than in Paris) (More info).

Environment

DYNI is a team of the LIS laboratory (UMR 7020 CNRS) at the University of Toulon. The team is composed of 6 faculty members including a chair in AI and 2 AI ANR JCJC laureats. The team also comprises 5 post-docs, 6 PhD students.

Ricard Marxer (supervisor of this position) leads the team (dyni.pages.lis-lab.fr) and is the director of a Erasmus Mundus Joint Master’s degree MIR in AI and robotics (www.master-mir.eu).

References

1. Sethi, S. S., Fossøy, F., Cretois, B., & Rosten, C. M. (2021). Management relevant applications of acoustic monitoring for Norwegian nature – The Sound of Norway. In 31. Norsk institutt for naturforskning (NINA). https://brage.nina.no/nina-xmlui/handle/11250/2832294
2. S. Kahl, C. M. Wood, M. Eibl, H. Klinck (2021). BirdNET: A deep learning solution for avian diversity monitoring, Ecological Informatics, Volume 61, 2021, 101236, ISSN 1574-9541, https://doi.org/10.1016/j.ecoinf.2021.101236
3. Best, P., Paris, S., Glotin, H., & Marxer, R. (2023). Deep audio embeddings for vocalisation clustering. Plos one, 18(7), e0283396. https://doi.org/10.1371/journal.pone.0283396