Art manager

AI helps accelerate ecological studies


[ad_1]

Environmentalists have monitored seal populations for decades, creating vast libraries of aerial photos. Counting the number of seals in these photos requires hours of painstaking work to manually identify the animals in each image.

Today, an interdisciplinary team of researchers including Jeroen Hoekendijk, doctoral student at Wageningen University and Research (WUR) and employed by the Royal Netherlands Institute for Sea Research (NIOZ), and Devis Tuia, professor Associate and Head of the Environmental Institute The computation and earth observation laboratory at EPFL Valais has developed a more efficient approach to counting objects in ecological studies. In their study, Posted in Scientific reports, they use a deep learning model to count the number of seals in archived photos. Their method could go through 100 images in less than a minute, compared to an hour for a human expert.

No labeling required

“In ecology, the most commonly used deep learning models are first trained to detect individual objects, after which the detected objects are counted. This type of model requires a lot of annotations of individual objects during training, ”explains Hoekendijk. However, the method applied by the research team eliminates the need to label individual seals beforehand, significantly speeding up the procedure since only the total number of animals in the photo is needed. In addition, their method can be used to count any individual item or animal, and thus potentially help process not only new photos, but also those that could not be analyzed due to lack of time. This represents decades of photos that could provide important information on how the size of the population has changed over time.

From macroscopic to microscopic

The way seals appear in aerial photos can vary widely from batch to batch, depending on the elevation and angle at which the photo was taken. The research team therefore assessed the robustness to such a variation. Additionally, to demonstrate the potential of their deep learning model, the scientists tested their approach on a fundamentally different dataset, on a much smaller scale: images of microscopic growth rings in fishbones. called otoliths. These otoliths, or auditory stones, are hard structures of calcium carbonate located directly behind a fish’s brain. The scientists trained their model to count the daily growth rings visible in the images, which are used to estimate the age of the fish. These growth rings are known to be extremely difficult to label individually. The research team found that their model had roughly the same margin of error as the manual methods, but could process 100 images in less than a minute, whereas it would take three hours for an expert.

The next step

The next step will be to apply similar approaches to satellite images of inaccessible Arctic regions where several seal populations live that are on the Red List of Threatened Species compiled by the International Union for the Conservation of Nature. “We plan to use this approach to study endangered species in this remote part of the world, where temperatures are rising twice as fast as elsewhere on the planet,” explains Tuia. Knowing where animals are concentrated is essential to protect these often endangered species.

– This press release was originally published on the The website of the Federal Polytechnic of Lausanne

[ad_2]