Today, we welcome to our blog Eric DE-OLIVEIRA, Engineer-Researcher of EDF.
Eric, welcome. We are delighted to have this short interview with you today. EDF team is responsible of the French demo. D-HYDROFELX tools will be tested in two different pilot sites: one site is in Dorogne River (Mauzac HPP) and one is in Selune river. Your demo focuses on the protection of fish species in relation to optimize the hydropower production.
Q: But please, before discussing on the project, would you like to introduce yourself to get to know you?
A: In 2000-2003, I did a PhD on fish stock assessment and especially using the geostatistical approaches. After I was involved in an Intereg Project at IFREMER for which I worked on building method and model to predict seabed habitat distribution (Postdoc). Before my Phd, from 1996 to 1998, I was in subantarctic island (Crozet Archipelago), in charge of the field activities of the project “Biodiversity evaluation and ecosystem transformation in a subantarctic island”.
I joined EDF R&D in 2005 as engineer-researcher in fish biology. I’m involved in 2 domains; the first deals with the fish passage at hydropower plant and the second one concerns the impact of the power plant on the marine environment. With other department form EDF Lab., we develop new environmental monitoring system, such as remote sensing for seabed habitat or using acoustic methods to detect and quantify the fish passage in the river.
Q: Could you describe your company’s role on the project?A: As part of the D-Hydroflex project, EDF will work on optimising hydroelectric production regarding environmental regulations. The final goal is to produce a tool to detect and identify fish species in the river to protect them from the turbine passage. It will develop a digital solution allowing long-term and non-invasive fish monitoring in the vicinity of HPPs. EDF will therefore be in charge of a demonstration site where acoustic cameras will be installed, enabling a river section to be scanned and the passage of fish to be recorded continuously. A software will be developed to automatically detect, identify and count fish using computer vision methods. Industrialisation of the software will enable it to be used under real acquisition conditions and provide knowledge on the passage of species of interest.
Q: We would like to know more on acoustic cameras.
A: Acoustic cameras are increasingly used for fish monitoring because of their numerous advantages compared to other monitoring methods. It is a non-invasive tool that can record continuously with low dependence to environmental conditions, providing video-like data. Hence, acoustic cameras are high-frequency multiple beams sonars, meaning that they detect echoes in a large volume of water and project them as 2D images. They provide high-resolution acoustic videos on which fish can be visualised as they swim in the camera field of view. Morphological characteristics, such as length, and swimming behaviour can be analysed accurately, providing useful information for fish species identification and counting.
Q: What do you expect to achieve in D-HYDROFLEX through the fish monitoring?
A: The main challenge is to extract, in near-real time, the information of interest available from acoustic videos while handling the high number of videos. This requires the development of a method for automatically analysing these videos that can be recorded from different types of acoustic cameras. Their analysis is challenged by the quality, the resolution, the level of details of the images. Hence, while the identification of fish species from images is widely investigated in the optical field, it is more preliminary for acoustic videos where only the overall morphology of the fish and the deformation of its body during swimming can be visualised. The methodological challenge is therefore to extract this information and use it as input for a classification model. To achieve this, computer vision methods will be used, in particular deformation models, generally used in 2D animation, to track the deformation of the fish’s body and extract repeating and discriminating patterns.
This also posed an operational challenge, in order to record videos with the best acquisition quality while trying to cover as large a section of river as possible. The aim is to have the most representative monitoring and counting possible.
Q: What do you expect from the project?
A: The information that can be collected thanks to the proposed solution will increase knowledge on the migration dynamics of fish species (i.e., migration season, environmental triggers of the migration). This will enable the optimisation of the protection measure, such as turbine management or additional bypass. In addition to reduce the environmental impact of HPPs, such optimisation will enhance their performance, regarding hydropower production.
Eric, thank you for being with us today. It is really interesting how D-HYDROFLEX incorporates the biodiversity aspect in their research and development.
For information on the fish monitoring, please visit the following article:
Le Quinio A, De Oliveira E, Girard A, Guillard J, Roussel J-M, Zaoui F, et al. (2023) Automatic detection, identification and counting of anguilliform fish using in situ acoustic camera data: Development of a cross-camera morphological analysis approach. PLoS ONE 18(2): e0273588. https://doi.org/10.1371/journal.pone.0273588
Another # meet the D-HYDROFLEX team blog story is completed. Stay tuned to learn more on our workforce team!
Learn more on D-HYDROFLEX project here. Please subscribe here to our newsletter to keep updated on our activities.
