• Oops!
    Something went wrong.
    Please try again later.

Understanding sperm whales through AI

  • Oops!
    Something went wrong.
    Please try again later.

[Marine Biologist Lisa Steiner]

"I'm probably one of the original sperm whale geeks basically."

Marine biologist Lisa Steiner has made it her life’s work

to help people understand sperm whales better through machine learning -

a type of artificial intelligence - in the sea around the Azores in the North Atlantic.

"My name is Lisa Steiner. I'm sitting on my back porch out in the Azores."

It’s been a painstaking task that she has carried out manually and single-handedly, for the last 35 years.

Since the 1980s Steiner has been photographing the tails - or flukes - of the endangered species

trying to match them with other sightings of the same whale to work out its migration routes and population sizes.

"We started with black and white film in the cameras that we would develop ourselves in the darkroom and then print contact sheets and then from the contact sheets you pick out the specific fluke pictures that you want to print and then you printed those pictures and then you match them."

Thankfully, the task is now being carried out faster and more accurately by a computer program - developed by Amazon Web Services and Capgemini.

The Fluketracker recognizes the unique shape of each whale's fluke, which, much like a fingerprint, is unique to the individual.

It’s already helped Steiner identify more than 200 new individual animals.

"It finds matches that my old program didn't find and that looking by eye I would never find. The machine learning for me is, they've trained it to know what a tail is, what a tail looks like and then they can train the computers to recognize the contours at the edge and then it goes from there and it sees more and more tails that I've provided. They had tails from 2005 to 2018. All the information is in a database so we can see where they've been, who they've been hanging out with, kind of what the social structure of what the groups are."

Steiner says the end goal is to make the platform open source for citizen scientists to upload their marine life photographs for fun and to help scientific research.

"There's nothing I'd rather be doing on a nice day, nice weather with nice whales although when there's bad weather and the whales I'd rather be in bed."

Video Transcript

LISA STEINER: I told you it looked a bit like a tree trunk, and there it is just floating at the surface. I'm probably one of the original sperm whale geeks, basically.

[CHUCKLES]

- Marine biologist, Lisa Steiner, has made it her life's work to help people understand sperm whales better through machine learning, a type of artificial intelligence in the sea around the Azores in the North Atlantic.

LISA STEINER: OK there goes the arch, So, next time, it's going to dive. Everybody ready? And there goes.

My name is Lisa Steiner. I'm sitting on my back porch out in the Azores.

- It's been a painstaking task that she's carried out manually, and single-handedly, for the last 35 years. Since the 1980s, Steiner has been photographing the tails, or flukes, of the endangered species, trying to match them with other sightings of the same whale to work out its migration routes and population sizes.

LISA STEINER: We started with black and white film in the cameras that we would develop ourselves in the darkroom, and then print contact sheets. And then from the contact sheets you [INAUDIBLE] pick out the specific fluke pictures that you want to print, and then you printed those pictures, and then you match them.

- Thankfully, the task is now being carried out faster and more accurately by a computer program, developed by Amazon Web Services and Capgemini. The Fluketracker recognizes the unique shape of each whale's fluke, which, much like a fingerprint, is unique to the individual. It's already helped Steiner identify more than 200 new individual animals.

LISA STEINER: It finds matches that my old program didn't find, and that looking by eye I would never find. The machine learning, for me, is that it's learned-- they've trained it to know what a tail is, what a tail looks like, and then they can train the computers to recognize the contours of the edge, and then it goes from there, and it sees more and more tails that I've provided. They had tails from 2005 to 2018.

All the information is in a database. So we can see where they've been, who they've been hanging out with, kind of what the social structure of the groups are.

- Steiner says the end goal is to make the platform open source for citizen scientists to upload their marine life photographs for fun and to help scientific research.

LISA STEINER: There's nothing I'd rather be doing on a nice day, nice weather with nice whales. Although, when there's bad weather and no whales, I'd rather be in bed.

[LAUGHS]