High Impact Blow Inspection over a Reactive Mobile-Cloud Framework

companyscourgeAI and Robotics

Oct 19, 2013 (4 years and 22 days ago)

82 views

High Impact Blow
Inspection over a
Reactive Mobile
-
Cloud
Framework

Presentation by:

Eric L. Luster

Hong Wu

Project
Introduction

1.
The
state
-
of
-
the
-
art of this
Project


a)
Various
systems that are designed to evaluate correlations
between head acceleration measurements and concussions
are in the early stages of research and
development.


b)
Automatically
detect the impact between the athletes by using
online machine learning method
.


c)
Design a more effective solution for delivering textual and
imagery to mobile devices







Project
Introduction

1.
The Instrumented Football Helmet (IFH) is a standard regulation
football helmet that is equipped with sensors that measure, record,
and analyze impacts. Upon examining similar products and existing
patents, there are two areas of potential infringement.


2.
US Patent number 5978972, filed in June 11, 1997, which outlines
a system designed to measure and record in real time data relating
to translational and angular acceleration of an individual’s head
during normal sporting activity.


[1] A. Camp, A.
Boeckmann
, M. Olson, K. Hughes, ECE 477 Final Report − Fall 2008 Team 2 − PHI
-
Master


2009
-

Simbex
receives an NIH SBIR
Phase I award for
develop HitAlert™
-

high schools and
youth football
programs

2010
-

Simbex
receives an NIH SBIR
Phase II award to
continue
development of
HitAlert™ technology
to expand to enhance
Simbex's product
offerings in head
impact biomechanics.

2010
-

A team from
Arizona State
University work on
state of the art
wireless delivery
methods for
reporting in real
-
time head impacts
and concussions

Related
Research

News

By

Riddell

on Tuesday, August 10, 2010

Ruling Finds Schutt Infringed Riddell’s Concussion

Reduction Technology Patents


(CHICAGO, August 10, 2010)


A federal court jury in
Madison, Wis., has found that Schutt Sports Inc.’s DNA
and ION football helmets infringed the

concussion
reduction technology

features of the Riddell Revolution
family of football helmets. The jury awarded Riddell just
under
$30 million
in damages for Schutt’s infringing
activities.



Paper # 1

P. Viola and M. Jones, “Rapid Object Detection using a Boosted
Cascade of Simple Features,”

Hong Wu

Background


P. Viola and M. Jones, “Rapid Object Detection using a
Boosted Cascade of Simple Features,” IEEE Computer
Society Conference on Computer Vision and Pattern
Recognition, Los Alamitos, CA, USA: IEEE Computer
Society, 2001, p. 511.


Classical method in computer vision, cited over 4000 times


Motivation


Relation between computer vision, machine learning and
mobile computing

Reduce the labor work of marking the samples.

Reduce the time used in training.


Method

-

Data
-
driven training VS Intelligence
-
driven training

-

General feature

-

Adaboost

Demo


http://www.youtube.com/watch?v=0tSxMmAngs8&feature=player_embedded




Online VS Offline Adaboost


Online Training:

-

Get the sample one by one

-

Adaptive

-

Not accurate in all cases


Offline Training:

-

Get all the samples


at one time


Problem



Failure Case


http://www.youtube.com/watch?v=3AnWc5J9968&NR=1

Relation with the project


Automatically detect the impact with less supervision.


Assume that the athlete was tracked by a camera and
an impact is a true alarm if the athlete is running and
then fall down.


The concept of online machine learning can be used in
other applications such as training an accelerator
sensor to detect the gesture of a person by using heart
rate sensor.


Training: Accelerator sensor + heart rate sensor

Testing: Accelerator sensor


Paper #2


Eric L. Luster


Support for Mobile Access to DICOM
Images Over Heterogeneous Radio
Networks



I.Maglogiannis
, G.
Kormentzas
, and T.
Pliakas
, Wavelet
-
Based
Compression With ROI Coding Support for Mobile Access to DICOM
Images Over Heterogeneous Radio Networks, IEEE Transaction on
Information Technology in Biomedicine, vol. 13, no., 4 July 2009

Paper Background


The visual quality of the medical images/scans is required to
be high, in order to ensure correct and efficient assessment
resulting in correct diagnosis.



In this context, a mobile device has to handle medical images
of significant sizes, while also taking into account its own
limitations concerning memory and processing resources.

DLWIC


Useful when a user browses medical images
using slow
-
bandwidth connections,



DLWIC uses the progressivism by stopping the
coding when the quality of the reconstruction
exceeds a threshold given as an input
parameter to the algorithm.

Relevance to our Project


Application enhances the viewing of the following types of images
on a mobile device:



Computed Tomography (CT) scans


Computed Radiography (CR) scans


Magnetic Resonance (MR) images


Stored in picture archiving and communication systems (PACS)


Hospital Information Systems (HIS)



Furthermore, the current medical image viewers do not take into
consideration the special requirements and needs of an
heterogeneous radio access environment composed of different
radio access technologies [e.g., GPRS/UMTS, WLAN, and DVB
-
H).

Back
-
up Slides

Original Master Schedule

Master Schedule

Tables & Figures