Tiny AI-based biologgers that reveal the interesting moments of a bird’s day

Have you ever wondered what wild animals do all day? The documentaries offer a glimpse into their lives, but the animals under the watchful eye do nothing interesting. The true essence of their behaviors remains elusive. Now, researchers in Japan have developed a camera that allows us to capture these behaviors.

In a recent study published in PNAS NexusResearchers at Osaka University have created a small sensor-based data logger (called a biologger) that automatically detects and records video of unusual behaviors in wild seabirds without supervision from researchers.

Uncommon behaviors, such as diving into water for food, can lead to new insights or even new directions in research. But observing enough behaviors to infer results is difficult, especially when these behaviors take place in an environment that is not hospitable to humans, such as the open ocean. As a result, the detailed behaviors of these animals remain largely unknown.

“Video cameras placed on the animal are an excellent way to observe behavior,” says Kei Tanigaki, lead author of the study. However, video cameras consume a lot of power and this leads to a trade-off. “Either the video only records until the battery runs out, in which case you might miss the strange behavior, or you use a larger, heavier battery, which is not suitable for the animal.”

Credit: Osaka University

To avoid having to make this decision for the wild seabirds under study, the team uses low-power sensors, such as accelerometers, to determine when unusual behavior is occurring. The camera is then turned on, the behavior is recorded, and turned off until the next time. This biologger is the first to use artificial intelligence to perform this task.

“We used a method called isolation forest,” says Takuya Maekawa, lead author. “This method detects outlier events well, but like many other AI algorithms, it is computationally complex. This means that, like video cameras, it consumes a lot of power.” For the biologgers, the researchers needed a lightweight algorithm, so they trained the original isolation forest on their data and then used it as a “master” to train a smaller “student” outlier detector installed on the biologger.

The final biologger is 23 g, which is less than 5% of the body weight of the striped shearwater birds under study. Eighteen biological loggers were deployed, a total of 205 hours of low-power sensor data were collected, and 76 5-minute videos were collected. The researchers were able to collect enough data to reveal novel aspects of the birds’ head-shaking, foraging behaviors.

This approach, which overcomes the battery life limitation of most biologgers, will help us understand the behaviors of wildlife that venture into human-inhabited areas. It will also allow animals to be observed in extreme environments that are inaccessible to humans. This means that many other rare behaviors can be studied in the future, from Japanese monkeys washing sweet potatoes to penguins eating jellyfish.