科学家们使用人工智能来帮助追踪鲸的迁徙

Scientists use AI to help track whale migration 
Scientists in California are using artificial intelligence to help match humpback whale tails that tell them where they have been and how healthy they are. NBC News' Anne Thompson spoke to one scientist about how the whale tracking has helped preserve the ocean.
加利福尼亚州的科学家们正在利用人工智能来帮助匹配座头鲸的尾鳍,这些尾鳍能告诉他们这些鲸鱼去过哪里以及它们的健康状况如何。NBC新闻频道的安妮·汤普森采访了一位科学家,了解这种鲸鱼追踪工作如何有助于保护海洋。
2024.10.27 NBC News
tags: 新闻 海洋 环境
内容
  • 英中
  • 中英
  • 盲听
  • 1

    [1] It's feeding time for humpback whales off of California's Central Coast. - Oh, I just saw a whale! - Really?

  • 2

    [2] Kate Spencer is our boat captain. - Yes, indeedy! Monterey Bay is like the best restaurant in a food court.

  • 3

    [3] Through the summer and fall, humpbacks dine on anchovies and krill to prepare for the trip to Mexico's warmer waters, where the whales breed.

  • 4

    [4] Oh, whales behind us! - Making this prime time for scientist Ted Cheeseman. - There they are! They can eat 10 to 20,000 pounds a day. so they're literally eating up to 20% of their body weight in a day!

  • 5

    [5] Before you see the whales, you can actually smell them, because after feasting on anchovies, their breath has a unique aroma. - An aroma that had seemingly vaporized a mere 50 years ago.

  • 6

    [6] Once hunted to near extinction, whales are now resurfacing here in Monterey Bay in dramatic numbers.

  • 7

    [7] Now, with just a camera, photos from scientists and the public, are helping researchers track whales from anywhere around the world.

  • 8

    [8] Thanks to artificial intelligence in real time, Cheeseman can match their tails or flukes, which tell stories of where the whales have been and how healthy they are.

  • 9

    [9] The pictures are uploaded and matched on Cheeseman's website, Happy Whale. - This photo identification has been around for 50 years, but what we've done is automated it so that we can just gather as much data as possible. - So, all of today's photos.

  • 10

    [10] Through AI, that image is matched to other whales already photographed, like this one called Oldtimer and Monae.

  • 11

    [11] Our last recorded sighting of Monae was back in February down in Baja, Mexico.

  • 12

    [12] In his lab, Cheeseman says it is facial recognition technology that makes this possible.

  • 13

    [13] The AI image recognition extracts all the signals in there. Like, if you imagine on our face, I have kind of a big nose, that's part of the signal, right? Or, all the different features that it can pull detail out of to look for a similar photo.

  • 14

    [14] Now, is it true that no two whale tails are alike? - Absolutely. like a fingerprint or a face, every humpback whale's tail is distinct.

  • 15

    [15] And AI makes that recognition fast. - How long does it take you per photo? - We don't measure the time, it's instantaneous.

  • 16

    [16] Over 100,000 humpbacks identified. Some migrate thousands of miles, navigating threats from fishing lines and waters warmed by climate change, making food harder to find.

  • 17

    [17] Warmer waters produce less food, and so there's less fish. Less fish, less food for the whales.

  • 18

    [18] But Cheeseman says the most valuable information gained may be human. - "You connect with one whale, it connects you with the whole ocean. To have that healthy ocean, we have to treat it like it's part of our home. It's like it's a place we care about."

  • 19

    [19] Using technology to connect and preserve the ocean for everyone and everything.

  • 20

    [20] Before AI, identifying the whales was a time-consuming process. Cheeseman actually had to look at photos with his own eyes and try to see if they matched. It was an imprecise science.

  • 21

    [21] Now, technology allows for much more precision, and Cheeseman says if you want to help out, the best pictures come from cameras with telephoto lenses. I'm Anne Thompson in New York. Now, back to you!

  • 1

    这是加州中部海岸座头鲸的进食时间。 - 噢,我刚看到一头鲸! - 真的吗?

  • 2

    凯特·斯宾塞(博物学家)是我们的船长。 - 是的,确实如此!(加州中部的)蒙特利湾就像美食广场上最好的餐厅(比喻观鲸位置绝佳)。

  • 3

    整个夏天和秋天,座头鲸以凤尾鱼和磷虾为食,为前往墨西哥温暖的水域做准备,那里是鲸繁殖的地方。

  • 4

    哦,鲸在我们后面! - 这是科学家泰德·吉士曼的黄金时间。 - 它们来了!它们一天能吃掉1到2万磅。所以它们每天吃掉的食物相当于自身体重的20%!

  • 5

    在你看到鲸之前,你可以闻到它们的气味,因为在吃了凤尾鱼之后,它们的口气有一种独特的芬芳。 - 一种似乎在50年前就消散了的气味。

  • 6

    曾经被捕杀到濒临灭绝的鲸现在又大量出现在蒙特利湾。

  • 7

    现在,只需要一台相机,来自科学家和公众的照片就可以帮助研究人员追踪来自世界各地的鲸。

  • 8

    多亏了实时人工智能,泰德·吉士曼可以匹配它们的尾梢或尾叶,这些信息可以告诉鲸去过哪里,它们有多健康的报告。

  • 9

    这些照片被上传到泰德·吉士曼的网站Happy Whale上并进行比对。 - 这种照片识别已经有50年的历史了,但我们所做的是自动化的,这样我们就可以收集尽可能多的数据。 - 今天所有的照片。

  • 10

    通过人工智能,这张照片与其它已经拍摄的鲸相匹配,比如这张照片叫“老前辈莫内”。

  • 11

    我们最后一次看到莫内是在二月份,在墨西哥的下加利福尼亚。

  • 12

    在他的实验室里,吉士曼说面部识别技术使这一切成为可能。

  • 13

    人工智能图像识别会提取其中的所有信号。比如,如果你想象一下我们的脸,我有一个大鼻子,这是信号的一部分,对吧?或者,它可以从所有不同的特征中提取细节来寻找相似的照片。

  • 14

    那么,有没有两条鲸的尾巴是一样的呢? - 当然有。就像指纹或脸一样,每条座头鲸的尾巴都是独特的。

  • 15

    而人工智能让这种识别变得很快。 - 每张照片识别要花多长时间? - 我们不测量时间,它是瞬间完成的。

  • 16

    已确认的座头鲸超过10万头。有些迁徙数千英里,在钓鱼线和因气候变化而变暖的水域的威胁中迁徙,这使得食物更难找到。

  • 17

    温暖的海水产生的食物更少,所以鱼也更少。鱼越来越少,鲸鱼的食物也越来越少。

  • 18

    但吉士曼表示,获得的最有价值的信息可能是关于人类自身的。-“你和一头鲸联系在一起,它就会让你与整个海洋建立联系。要拥有健康的海洋,我们必须把它当作家园的一部分来对待。就好像它是我们所关心的一个地方。”

  • 19

    利用科技为所有人和所有事物关联和保护海洋。

  • 20

    在AI出现之前,识别鲸是一个耗时的过程。比如吉士曼必须亲肉眼对比照片,看看它们是否匹配。这是一门不精确的科学。

  • 21

    现在,技术允许更高的精度,吉士曼说,如果你想帮忙,最好的照片来自长焦镜头的相机。我是安妮·汤普森,在纽约报道。现在,交给你(同事)。