Tanya BergerWolf How your nature photos can help protect wild animals TED

in 2014 i met sudam

the last male white northern white rhino

in the world

just four years later

sudan died

leaving two

both female northern white rhinos alive

the species are effectively extinct

and they’re not the only ones

we’re losing biodiversity at an

unprecedented scale

we’re in the middle of what is termed

the sixth mass extinction a biodiversity

crisis

and we don’t even have the scientific

and technological solution to keep up

knowing what we’re losing and how fast

the international unit for conservation

of nature

red list

is the official international

organization that keeps track of the

biodiversity of the world

and of the 130 000 species that they

track out of the millions that are out

there

majority

have their conservation status

as data deficient

or

their population trend as unknown

and these are iconic species like killer

whales

and polar bears

we don’t know how well they’re doing we

can’t make policy decisions

we can’t put the right resources to

protect them

how many african elephants are there how

fast are they lost to poaching how far

do the whales go and how many juvenile

turtles survive to adulthood we don’t

know

and these data are critical to

conservation decisions

so how do we get those data

there are not enough scientists to track

every animal everywhere all over the

world and not enough

collars and satellite tags to track them

besides to put a satellite tag or a

collar on an animal you have to actually

capture them

tranquilize them

have a vet

present to monitor the vital signs of

the animal and even if everything goes

right

the collar can get snagged on a branch

or the satellite tag can get infected so

this can be dangerous to the animals

today

images are the most abundant readily

available source of information about

anything from what you had for lunch

to what animals you saw in your backyard

or in a safari tour

coming from scientists field assistants

camera traps as well as drones and

tourists going on safaris and whale

watching tours there are millions of

images out there

if you could only take those millions of

images

and extract the information about

wildlife

well

artificial intelligence to the rescue

we designed algorithms and created a

platform

wildbook that uses modern artificial

intelligence machine learning

and computer vision to take these

millions of images

find the ones that contain animals

find where the animals are in those

pictures including that baby elephant

hiding behind its mom

and figure out not only species

but down to individual animal

recognizing zippy the zebra and joe the

giraffe and terry the turtle and willy

the whale

using the unique markings on an animal

body like a fingerprint

a body print if you wish

the stripes spots wrinkles notches as

well as the shape of a whale’s fluke or

the dorsal fin of a dolphin these are

unique as every animal is

and with information on when and where

the image was taken

we can now use

pictures

instead of colors and tags to track

animals

count them and even figure out their

social network who is whose animal’s

friend

this is an example page from a wild book

for whales and dolphins fluke book

and this is pinchy

the most sighted animal in that wild but

pinchy is a celebrity she’s a ham she

likes getting her picture taken

she has more than 600 sightings around

dominica she lives there she hangs out

there

and flipbook the wild book for wells and

dolphins

contains

more than a million images

of

almost 46

000

identified individuals

providing the basis for science and

conservation

we even

developed an artificial intelligence

agent

who scours social media publicly posted

images and

videos and finds the ones that contain

animals sends them off to this machine

learning back engine

for identification and adding to the

appropriate page of the right wild book

and then posting back in the comments of

the social media saying hey at two

minutes 46 seconds we found this whale

shark in your video here’s everything we

know about it and people respond

wow this is amazing how can i help

that

how can i help

we engage people

right where they are using turning their

vacation videos into data for science

and conservation with the help of

artificial intelligence

the wild book for whale sharks contains

data now for more than twelve thousand

individual whale sharks

from photographs brought in by

almost 9 000 citizen scientists 200 plus

conservation and science projects and

one very intelligent

agent

from social media

that together that is now the foundation

for the iucn red list entry

for the species

providing not only data for the global

population size

but determining its conservation status

and changing it from vulnerable to

endangered

and the population trend from stable to

decreasing not because the species are

doing any worse

but because we now know better

we can make better decisions we can

create better policy we can put the

right resources to

to support it

we have wild books for 53 species

from marine to terrestrial spanning the

entire globe and growing

the technology in wild bug

was also used for the first ever full

senses

of the entire species the endangered

gravy zebra

using photographs from ordinary people

just

taking pictures for two days

for the first time in january 2016

hundreds of people were driving around

kenya the country containing 95 of this

endangered species

from rangers

and school kids to tourists with

telephoto cameras they took more than 40

000 images and the machine learning

technology of wild book identified all

the animals providing the most accurate

count of the species so much so that

kenya wildlife service said this how

we’re going to track the species from

now on and do this every two years

with the event known as the great gravis

rally

and so we repeated it in 2018

with more than thousand people and also

in 2020

and that data

became the basis for the iucn red list

entry for the gravy zebra

as well as for the conservation policy

the endangered species management

for kenya wildlife service

artificial intelligence

democratizes science

it connects people

bringing together the pixels of

individual cameras into the global view

of biodiversity

ai

helps

create conservation policy

science and engage people at large scale

and high resolution

and it takes the incredible team

of wild me the non-profit home of wild

book as well as thousands of people all

over the world

who take pictures

annotate them and make them ready for ai

create technology

and use it for conservation as well as

all the people who work out there in the

field

protecting the biodiversity of the

planet

and i hope you join us

you

[Music]

2014 年,我遇到了世界

上最后一头雄性北方白犀牛 sudam,

仅仅四年后,

苏丹就死了,

只剩下

两只雌性北方白犀牛活着。

这个物种实际上已经灭绝,

而且它们并不是唯一

我们正在以前所未有的速度失去生物多样性的物种

规模

我们正处于被

称为第六次大规模灭绝的中间 生物多样性

危机 我们甚至没有科学

和技术解决方案来

跟上我们正在失去的东西以及

国际自然保护单位

红色的速度 list

是官方的国际

组织,负责跟踪

世界

的生物多样性以及他们

在数以百万计的物种中追踪的 130 000 个物种,

其中

大多数物种的保护状态

为数据不足

或种群趋势未知

,这些都是标志性的 虎鲸和北极熊等物种

我们不知道它们的表现如何 我们

无法做出政策决定

我们无法做出正确的决定

保护它们的来源

那里有多少非洲象

它们因偷猎而流失的速度有多快

鲸鱼能走多远以及有多少幼

龟能存活到成年 我们不

知道这些数据对于

保护决策至关重要

那么我们如何获得这些数据 数据

没有足够的科学家来追踪

世界各地的每一种动物 没有足够的

项圈和卫星标签来追踪它们

除了在动物身上贴上卫星标签或

项圈 你必须实际

捕捉它们

让它们镇静

兽医 监测动物的生命

体征,即使一切

正常,项圈也可能被树枝卡住,

或者卫星标签可能被感染,因此

这对今天的动物来说可能是危险的,

图像是最丰富的现成

信息

来源 你午餐

吃了什么,你在后院看到了什么动物,

或者在一次野生动物园之旅中

,来自科学家现场助理的

相机陷阱以及 无人机和

游客进行狩猎和观鲸之旅

如果您只能拍摄这数百万张

图像

并提取有关野生动物的信息,那么那里就有数百万张图像

以及

人工智能进行救援

我们设计了算法并创建了

一个使用现代人工

智能机器学习

和计算机视觉来获取这

数百万张图像

找到包含动物的图像

找到动物在这些

照片中的位置,包括

躲在妈妈身后的小象

,不仅要找出物种,

还要找出个体动物

识别斑马和 joe the

giraffe and terry the turtle 和 willy

the wil

使用动物身体上的独特标记(

如指纹)

如果您

希望条纹斑点皱纹凹口

以及鲸鱼的吸虫

或海豚背鳍的形状这些

每只动物都是独一无二的,

并且有关于

图像何时何地的信息 采取

我们现在可以使用

图片

而不是颜色和标签来跟踪

动物

计数它们甚至找出他们的

社交网络谁是谁的动物的

朋友

这是来自

鲸鱼和海豚侥幸书

的野生书籍的示例页面这是

最有眼光的 那个狂野的动物,但又

矮又小是名人 她是火腿

她喜欢拍照

她在多米尼克周围有 600 多次

目击 她住在那里 她在那里闲逛

并翻书 这本关于水井和海豚的野书

包含

超过一百万张

几乎 46

000 个已

识别的个体

为科学和保护提供了基础,

我们甚至

开发了一种人工智能

代理

,它可以搜索社交媒体公开发布的

图像和

视频,并找到包含动物的图像和视频,

并将它们发送到这个机器

学习反向引擎

进行识别并添加到

适当的 正确的野书页面

,然后在社交媒体的评论中发回

说 嘿,在 2

分 46 秒时,我们

在您的视频中发现了这条鲸鲨,这是我们所

知道的一切,人们对此做出了回应,

哇,这太神奇了,我如何能提供帮助

我如何帮助

我们

在人们使用的地方吸引他们将他们的

假期视频转化为数据

在人工智能的帮助下进行科学和保护

鲸鲨野生手册包含

来自

近 9 000 名公民科学家带来的照片中超过 12,000 条鲸鲨个体的数据 200 多个

保护和科学项目以及

来自社会的一名非常聪明的代理人 媒体

认为,这现在是该物种

进入 IUCN 红色名录的基础,

不仅提供全球

种群规模的数据,

而且确定其保护状态

并将其从易危变为

濒危

,种群趋势从稳定变为

减少,这不是因为该物种

做得更糟,

但因为我们现在知道得更好,

我们可以做出更好的决定 我们可以

制定更好的政策 我们可以投入

适当的资源

来支持它

我们拥有涵盖全球 53 种从海洋到陆地的野生动物的野书,

并且

在野生昆虫中生长的技术

也被用于人类第一次完整的

感官 物种濒临灭绝的

肉汁斑马

使用普通人

拍摄的照片

2016 年 1 月第一次拍摄了两天

超过 40

000 张图像和 wild book 的机器学习

技术识别出所有

提供最准确

物种计数的动物,以至于

肯尼亚野生动物服务机构这样说,

我们将从现在开始跟踪物种

并做到这一点 两年来

,这场被称为“伟大的重力

拉力赛”的活动

,所以我们在 2018 年与 1000 多人重复了它,

并且 同样

在 2020 年

,该数据

成为 IUCN 红色名录

条目的肉汁斑马

以及保护政策

基础 肯尼亚野生动物服务

局的濒危物种管理 人工智能

使科学民主化

它将人们连接

起来,将单个相机的像素汇集

生物多样性的全球视野

AI

有助于

创建保护政策

科学并以大规模和高分辨率吸引人们

,它需要

Wild me 这个令人难以置信的团队(野生书的非营利之家)

以及世界各地成千上万的

拍照者进行

注释 他们并让他们为人工智能做好准备,

创造技术

并将其用于保护以及

所有在

保护地球生物多样性领域工作的人

,我希望你能加入我们

[音乐]