Behaviour Recognition Base on Smartphone's Sensor
-
Upload
ngo-duy-kien -
Category
Documents
-
view
221 -
download
0
Transcript of Behaviour Recognition Base on Smartphone's Sensor
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 1/20
Click to edit Master subtitle style
5/6/12
THESIS REPORT
Behaviour recognition base on
smartphone's sensor
Student: Ngo Duy KienClass: K53CA
Supervisor: Bui The Duy
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 2/20
p y
5/6/12
Content
I. Falling detection problem
II. Principle and method for system
III. Research progressIV. Format of the thesis
V. Research plan
22
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 3/20
5/6/12
FALLING DETECTIONproblem
33
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 4/20
5/6/12
Personal EmergencyResponse (PER) System
44
1
Distinguishingbetween fallsand other types of normalbody
movement
2
Detecting a falland notifyingan emergencyserviceautomatically
3
Include acurrent locationfor cell phoneusing GPS or WIFI assisted
positioningsystem
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 5/20
5/6/12
cce erome ersensor
- X-axis haspositive directiontoward the right side
of the device
- Y-axis haspositive direction
toward the top of the device
- Z-axis has
positive direction
55
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 6/20
5/6/12
Principle and method for system
66
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 7/20
5/6/12
Study of falls
77
q Falling Definition
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 8/20
5/6/12
Study of falls
q Falling Types
88
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 9/20
5/6/12
Study of falls
q Falling Types
99
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 10/20
5/6/12
Study of non-falls
Non-falls
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 11/20
5/6/12
Related Work
Real-Time Fall Detection based onShape and Motion Features
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 12/20
5/6/12
Falling detection based onaccelerometer
q Analytical Method
§ Definition
§ Threshold
12
12
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 13/20
5/6/12
Falling detection based onaccelerometer q Machine Learning
§ Identify natural clusters or groups of data contains
classified data and predicts the classes of unseen
data
§ Supervised Learning
§
Unsupervised Learning
13
13
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 14/20
5/6/12
Machine Learning
Using a sliding window to transform thestream of acceleration data into instances for machine learning. The following attributeswere derived from the data within each 0.8second windows:
The average length of the acceleration vector within the window.
The average acceleration along the x, y,zaxes.
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 15/20
5/6/12
RESEARCH PROGRESS
1515
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 16/20
5/6/12
Research progress
10/2011-11/2011:◦ Understood the problem
◦ Getting data and understood how
analysis data
12/2011- Now :
◦ Classification types of falls (oractivities daily living) with libsvm onweka
◦ 91%
1616
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 17/20
5/6/12
Format of the thesis
1717
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 18/20
5/6/12
Format of the thesis
1. Behavior problem and somepopular solution
2. Classification data with Machine
Learning
3. Implement system
4. Result and experiment
1818
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 19/20
5/6/12
Research plan
Finishing the classification withmachine learning
Compare the results with other
method to process data
Implement the system onSmartphone
Improving the implementation
1919
8/3/2019 Behaviour Recognition Base on Smartphone's Sensor
http://slidepdf.com/reader/full/behaviour-recognition-base-on-smartphones-sensor 20/20
5/6/12
Thank you forlistening
2020