Candidate:
Yiqin
Huang
Title:
Object
and
Text
Detectors
under
the
Binary
Classification
Scenario
Date:
January
10,
2022
Time:
13:00
Place:
online
Supervisor(s):
Tripunitara,
Mahesh
Abstract:
Hate
speech
can
be
expressed
through
Internet
memes,
which
combine
images
and
texts
to
create
complicated
and
subtle
sentiments.
This
kind
of
hate
speech
is
easier
to
spread
and
more
difficult
to
be
detected
automatically.
Thus,
object
detection,
especially
its
subfield
text
detection,
is
an
essential
phase
of
the
pipeline
to
filter
out
those
multimodal
hate
speech
from
a
massive
quantity
of
images.
In
this
seminar,
I
will
introduce
the
development
of
object
and
text
detection
frameworks
in
the
deep
learning
era,
and
present
my
empirical
results
of
evaluating
two
object
detectors
and
two
text
detectors
under
the
binary
classification
scenario.