Science

AI-enabled critter camera only records targeted behaviors

AI-enabled critter camera only records targeted behaviors
A literal "bird's eye view" captured by the artificially intelligent bio-logger
A literal "bird's eye view" captured by the artificially intelligent bio-logger
View 2 Images
The AI system is initially trained using videos in which it has captured the desired behavior
1/2
The AI system is initially trained using videos in which it has captured the desired behavior
A literal "bird's eye view" captured by the artificially intelligent bio-logger
2/2
A literal "bird's eye view" captured by the artificially intelligent bio-logger

In order to observe specific behaviors, scientists will often rig wild animals up with tiny cameras or other "bio-logging" devices. A new technology could help these tools gather more data, by only fully powering up when needed.

Because they're frequently mounted on small animals, bio-loggers themselves have to be quite compact. This means that they have correspondingly tiny batteries, that don't allow for much runtime. Sometimes, this limitation is addressed by programming the device to record for just a few scheduled periods each day. The problem is, any behaviors that occur between those recording sessions are missed.

One alternative is to simply leave the device recording continuously, but then its battery will run out quite quickly – conceivably before the behavior that's the target of the study takes place.

Seeking a solution, scientists at Osaka University have developed an artificial intelligence-based bio-logger that incorporates an accelerometer and a GPS unit. Based on the output of these two sensors (which consume relatively little power) the bio-logger is able to determine when the animal is engaging in the targeted behavior. It then activates the more power-hungry camera, which only keeps recording as long as the behavior lasts.

The AI system is initially trained using videos in which it has captured the desired behavior
The AI system is initially trained using videos in which it has captured the desired behavior

Already, the technology has been successfully tested on seagulls off the coast of Japan.

"The new method improved the detection of foraging behaviors in the black-tailed gulls 15-fold compared with the random sampling method," says lead author of the study, Joseph Korpela.

The research is described in a paper that was recently published in the journal Communications Biology.

Gull's-view footage captured using the system can be seen in the following video.

Source: Osaka University

【AI on Animals】青森の海鳥が仲間の魚を横取り!AIを用いたバイオロギングによる海鳥の生態観測

No comments
0 comments
There are no comments. Be the first!