This presentation contains a case study on the project WildME which is based on computer vision and is helping in detecting aminals all across the world.
2. HISTORY OF COLLECTING DATA ON
WILDLIFE
• Wildlife biology used to depend on mark-recapture technique
(eg . ear tag ,leg band).
• Presence or absence were recorded manually.
• Physical tagging use to hurt or scare animals , that use to
make data less useful.
3. PROBLEM DESCRIPTION:
• Matching photos by eyes makes analysis less
accurate as well as tedious job..
• To Monitor population of endangered ,vulnerable,
extinct species is difficult.
IMAGINE LOOKING THROUGH THE SAME 5,000
IMAGES EVERY TIME YOU GET THE ONE,SO AS TO
IDENTIFY THE PATTERN OF SPOTS AND TAG
THEM…..
4. INNOVATION HISTORY
The year 2002 experience change
his life and lead to development of
AI ENABLE APPROACH TO
CONSERVATION.
He learnt plastic tags to track whale
shark which was highly in effective.
He developed a COMPUTER VISION
ALGORITHM to recognize the
unique pattern of spots on different
sharks to solve this problem.
As no such platform exist this gave
the birth to WILDME..
JASON HOLMBERG
(Director of engineering for wild
me and chief information of wild
me)
5. OBJECTIVE:
To find out:
What is the local wildlife population trajectory?
Where do animals go-and why?
Are recent conservation measures reversing
observed declines?
6. SOLUTION
• WILDBOOK
It helps us to identity and
protect animal population
using wildlife + citizen
science + AI+ computer
vision
NEWSPAPER CLIPPINGS:
2017-08-16
Check out the newest Wild book:. Giraffe spotter might be our best looking Wild book yet
and represents a consortium Giraffe spotter of collaborative researchers studying
endangered giraffes in Africa.
7. TECHNOLOGY AND METHDOLOGY
INTERFACE
DESIGN
COMPUTER VISION
ALGORITHM
ARTIFICIAL
INTELLIGENCE(ML)
COLLECTION
OF DATA
UPLOADING
DATA
• CITIZEN
SCIENCE
BASED
APPROACH
• RESEARCH
SCIENTIST
MACHINE
LEARNING(TEACHING
MACHINE)
DETECTION AND
IDENTIFICATION
DATA MANAGNEMENT
LAYER
8. PROJECTS:
• Manta Matcher:
• Whale Sharks:
Whale sharks
Step1: Data starts with single record of interaction with
animal(i.e.when someone takes picture of whale shark)
9. Step2: Data is encounter in wild book and than images are analysis with the
help of computer vision and ML (as machine is already being taught
about how to differentiate b/w species through various algorithm).
Fig.3 photo identification by
matching segments in dolphin
Fig.1 & Fig.2
explains pattern
recognization
10. Step3 : RESULT
• Finally we can
distinguish b/w 2 or
more animals of same
species.
• Through Fig.4 we can
look at the collected
encounters of a group
or individual and see
their movement,or look
at the no. of individual
over time to track birth
and death. Fig.4
11. EFFECTS
PRO’S:
• Look through images in
matter of seconds to find a
match.
• Wildbook can create data
record,link them in intutive
ways
• Reduce harm to animals
and saving researchers
countless hours.
CON’S:
• Security issues (poachers).
• Separation of some data
gets very complicated.
• Engaging citizen scientist is
sometimes difficult
12. FURTHER POSSIBILITY AND RELATED WORK
• The technology is progressing, but there is still a significant
lack of distinguishing all species right now .Researchers are
still working on developing a general technique to
differentiate all.
RELATED WORK : CLUNE
• Clune 's team developed deep neural networks of images that
mimic an animal's sensorial abilities. The process, however, is
not an easy feat as it requires an extensive amount of data.