"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
Mobile based distribution system for managing abandoned or lost pets
1. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Mobile-Based Distributed System
for Managing Abandoned or Lost Pets
José-Luis Poza-Luján(1), Daniel Garrote-
Hildebrand(2), Juan-Luis Posadas-Yagüe(1) and José-
Enrique Simó-Ten(1)(1) University Institute of Control Systems and Industrial Computing (ai2)
(2) School of Engineering in Computer Science (ETSinf)
Universitat Politècnica de València (UPV)
Camino de vera, s/n. 46022 Valencia (Spain)
1
2. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Current problem
• In most countries, legislation penalizes animal
abuse and neglect. However, animal loss or
abandonment is a current problem in some
communities.
• Numerous associations work in communities to
help animals either welcoming in animal shelters or
finding them a new home.
• Animal Protection Societies (APS) lack common
information, databases and channels to receive
animal warnings and notify the news.
2
3. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
• Limited to specific
geographic areas.
• Different data
specification.
• Non-automated
systems.
Current solutions
3
HeLP Lost and
Adoptable Pets
Petfinder
GlobalPetAlert
City of Calgary
4. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
• APS transmit the notice through their own webs as
well as in social networks.
• Currently advises are sent without checking
whether lost animals have been located.
Besides, users receive messages without picking the
geographic area of interest as they can receive
advises of lost animals that are located far away
from the area to be searched.
• It is of special interest the use of the new
technologies to optimize the entire process.
Current solutions
4
5. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
• There has been carried out several studies about
protocols that must be followed when either a lost
animal or abandoned animal is found by someone.
• Most of these protocols insist specially on both the
importance of information management and the
huge possibilities of being applied into other
technology fields regarding animal researches.
• In order to follow these protocols, animal
protection societies usually use web pages and
mobile applications in which they can warn about a
lost animal or search one to adopt.
Proposed solution
5
6. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Proposed solution
6
• Client
I. Inform about a lost or
abandoned pet
1) By means a smart
mobile device
2) Through a Web
Application
II. Receive notices
• Server
I. Receives the
notification
1) Processes the data
2) Match information
3) Generates warnings
7. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Proposed solution
Virtual
draws
Picture
SoundForm
Geolocation
Date and
Time
DATAMETADATA
XML-based
message
creation
XML
File
Image /
Audio file Main data
extraction
CLIENTS SERVER
Data-
matching
Selective
Warings
Data
base
···
Data-
process
Manager
User
Animal specie, Size, Skin
tone and similar data
Time, Geolocation
and User
Smart
mobile
Computer Web
Server
- Warnings
- Data management
- Users management
- Security
Step 1
Step 3
Step 2
User
Step 5
Step 6
Step 4Step 7
Step 8
Abandoned or lost pet
7
8. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Client components
• Client Input
–Artificial methods:
generate data to be
processed directly.
• Form
• Virtual draw
–Natural methods:
generate data that
requires a pre-processed
• Picture
• Voice
8
• Client Input
–Metadata:
provided directly
by the smart
mobile device
• Date
• Time
• Geoloation
• User
9. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Server components
• Data extractor
–Data extractor
–Data matching
–Data process
–Warning generator
9
XML
File
Image /
Audio file Main data
extraction
SERVER
Data-
matching
Selective
Warings
Data
base···
Data-
process
Manager
User
Web
Server
- Warnings
- Data management
- Users management
- Security
10. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
System developed
• Client: mobile App
–Picture view
• Provides a captured directly
or uploaded from file.
• Photo must be processed
on the server
–Form View
• Provides information
directly into the database
format.
10
11. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
System developed
• Server: Web service and database
11
12. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Experiments and results
• Nowadays the system is being developed.
• First and second phase
–We are developing the mobile application and
the server-side applications.
–Some collation algorithms, with the selective
alarms, has been developed.
• First experiments
– Main data extraction phase
– Label success rate: Total percentage of the labels
extracted successful assigned to a pet.
12
13. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Experiments and results
• Performance tests
–Communications functions
–Data base insertion
• Main data extraction
–Data extractor from picture (white, dog)
• Image segmentation (5 dogs and two images per dog)
– Clustering method: dog: 56,7% white: 82,3% success
– Edge detection: dog: 63,6% white: 82,1% succes
• Main problem: The Label success rate depends on the
quality of the image (environment, mobile device
model, etc.)
13
14. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Conclusions and future work
• This project has several technological challenges
that should be achieved are mainly associated with
data collation and selective advises.
• Challenges refer to avoid duplicate data in
databases or even to provide a simple storage
method.
• System suggests a few social challenges.
– The main one consists of achieving a collation level that
permits to increase the lost animal location, as selective
messages allow delimiting the search up to the recent
places in which the animal has been to.
14
15. 10th International Symposium on Distributed Computing and Artificial Intelligence
Salamanca, 22nd-24th May, 2013
Mobile-Based Distributed System
for Managing Abandoned or Lost Pets
José-Luis Poza-Luján(1), Daniel Garrote-
Hildebrand(2), Juan-Luis Posadas-Yagüe(1) and José-
Enrique Simó-Ten(1)
15
Thanks for your attention