Hojun Son
***- ***- **** ac26qw@r.postjobfree.com
**** ******** **, ********* *** Arbor, MI 48105- 2414 Objective:
Seeking
a
Computer
Vision
engineer
for
2018
internship
and
full
time Education
University
of
Michigan,
Ann
Arbor,
MI,
United
states
Expected:
May
2018 Master’s
Degree:
Electrical
Engineering
and
Computer
Science
(Computer
Vision)
GPA:
3.80/4.0 Courses: EECS 504, EECS 542, ROB599(Self- driving car), EECS 545 Inha
University,
Incheon,
Korea
Graduated:
Feb
2016 Bachelor
Degree:
Computer
Science
GPA:
4.0/4.5,
Major
GPA:
4.2/4.5 Graduate
thesis:
Architecture
of
decision
making
and
prediction
model
for
autonomous
vehicles
at
unprotected
left
turn
traffic signals
Illinois
Institute
of
Technology,
Chicago,
IL
Aug
2014
-
Feb
2015 Exchange
program:
Information
Technology Industry
Experiences
CrossRealms,
Chicago,
United
States
Dec
2014
-
Feb
2015 Intern
of
software
engineering
● Participated
in
non- profit
project
by
developing
recorder
and
MP3
player
with
Arduino
● Built
file
management
class
to
manage
recorded
files
in
Arduino Samsung
Software
Membership,
Seoul,
Korea
Jan
2012
-
Jul
2014 Software
Engineer
● Made
a
visualizer
to
show
metric
information
of
computer
programs
and
built
a
database
structure
with
large
data
● Implemented
a
home
network
simulator
and
viewer
with
Model
View
Controller
pattern
● Ran
Network
Simulator
2
tools
comparing
three
types
of
mac
protocols
and
build
a
visualizer
tool
for
the
result
● Designed
a
book
to
send
which
a
page
is
opened
and
which
an
event
is
touched
on
the
paper
book
to
connected
computer
● Developed
an
Android
application
for
a
road
based
social
network
service
using
GPS
data Academy
Experiences
Navigate
Visually
Impaired
Person,
Ann
Arbor,
University
of
Michigan
May
2017
-
Present Research
Assistant
● Detected
and
classified
crosswalk
signs
using
customized
VGG16
and
Adaboost
algorithm
in
Tensorflow
and
OpenCV
on smart
glasses.
● Implemented
a
tracker
to
return
original
coordinates
of
detected
crosswalk
sign
from
zoomed
images
based
on
optical
flow.
● Focusing
on
using
inertial
data
for
head
tracking
for
robustness
of
tracking
and
staying
of
crosswalk
signs
in
a
view.
● Training
end- to- end
deep
networks
similar
with
SSD
and
Yolo
combined
recurrent
rolling
convolution. Independent
Study,
Ann
Arbor,
University
of
Michigan
Sep
2017
-
Present Research
Assistant
● Implementing
deep
networks
based
on
GOTURN
to
track
objects
such
as
pedestrians
in
Tensorflow.
● Building
a
network
which
is
able
to
detect
cars
and
pedestrians
in
Tensorflow.
● Localization
and
matching
for
detected
objects
on
the
3D
coordinates.
(Not
sure) COVE,
Ann
Arbor,
University
of
Michigan
Oct
2016
–
Aug
2017 Research
Assistant
● Modified
tracker
source
files
to
make
tracker
benchmark
tools
on
Ubuntu
● Annotated
key
points
in
3D
models
to
classify
occluded
classes
in
images
● Implemented
specific
trackers
which
are
using
particle
filter,
color
distribution,
and
mean
shift
in
blob
tracking.