Upload
anisha-nayak
View
219
Download
0
Embed Size (px)
Citation preview
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 1/18
Adaptation of
Lecture Videos to Mobile Device
Presented by:
Anisha Nayak & Akanksha Patil
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 2/18
Introduction
•Mobiles and PDAs are becoming ubiquitous,
specially amongst students and there’s a need for
having all the applications of our real world into our
pocket!
•One such application is viewing lecture videos on
your mobile or PDAs.
•These videos are typically meant for the
computers and not hand held devices like mobile
phones, hence our proposal is to disprove this and
bring the above said need to reality!!
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 3/18
Problem Statement
Lecture videos are traditionally designed for
desktop computers, these require high video bit-
rates, due to which viewing them on low network
bandwidth connections like GPRS incurs long delays
and high costs. So the challenge is to come up with
some innovation in order to map these high BW
videos onto the available BW.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 4/18
CHALLENGES IN IMPLEMENTATION
There are two main challenges or problems that areencountered in this adaptation process:
1. Network Bandwidth and Cost - Network connections (likeGPRS) that are used by mobile phones have a very lownetwork bandwidth (40kbps). This is very low as compared tothe bit rate of a video which is in the range of 400-1200kbps
or even higher. Higher the bandwidth, higher you need to paythe service provider.
2. Usability of the content - Lecture videos usually have
written material, either as presentation slides or in thewritings of the instructor. Hence, the adaptation process mustensure that understanding and visibility of the content mustnot be compromised.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 5/18
Technical terms :•VIDEO TRANSCODING:
Video transcoding is the process of converting the video
from one compression format to another. In the process,video parameters like bit rate, frame rate and resolution can
be changed to meet the target device requirements.
2. CONTENT AWARE VIDEO ADAPTATION:
Content aware video adaptation is the process of adapting
the video based on its content, so that the content is visible
clearly, even at lower bit rates.But, compressing to very low bit rates degrades the
quality for any compression format, and can make the
content incomprehensible. Hence the above said process is a
trade off between compression and clarity of data.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 6/18
Current work in the field :1. There have been attempts, to identify important segments of a video
and important objects or regions in a video frame sequence.
These are then encoded at a high quality and other
portions are encoded at a low quality.These methods achieve a descent video quality (i.e.
that looks as good as the original one) at moderate bit rates of 100kbps for non-lecture videos, but at very low bit rates of 40kbpsand in the context of lecture videos, the quality and visibility of
written content is doubtful, as the video quality might degradefurther.
2. In the second method, non-changing portions of a lecture videoare identified and one image is extracted from each such region.
The output is a slide-show of such images alongwith the audio stream of the whole lecture. This method basicallyaims to achieve reduction in bandwidth by eliminatingredundancies in the video.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 7/18
Solution for the problem
We employ the same concept of
displaying slide-show of images extracted
from the video, as done in previous workbut in a different way. We identify Study
Elements within the video.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 8/18
BUT WHAT ARE STUDY ELEMENTS?
STUDY ELEMENTS
• “Study Element” is the portion of video showing any medium of instruction.
• Medium of instruction could be a ‘slide of presentation’ or ‘anexplanation written by a instructor on a white paper’ or the instructorherself explaining something.
• We then extract images from each Study-Element at a differentintervals; for example one image every 5seconds.
• The output is the slide-show of the extracted images sent to the client
according to the interval at which they were extracted from theelement, along with the audio.
• The user can specify the user experience desired for each Study-Element and has the flexibility for choosing the network bandwidthavailable as input.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 9/18
System architecture• The adaptation methodology accepts the
video as input.
•It further accepts the desired user experienceand the network bandwidth available, asparameters from the user.
• The desired user experience value for everyStudy-Element is accepted.
• For example, the desired user experience
values for presentation element andinstructor element are taken as input.
• The output consists of a set of images,sending interval for each Study-Element andcontinuous audio.
• Images are sent to the client device according
to the sending• interval. For example, if the sending interval is
five seconds, then one image is sent every fiveseconds.
• The audio is continuously streamed to theclient.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 10/18
We have defined three types of Study-Elements
1) Presentation Element - Portion of video
that shows one slide of a presentation2) White Paper Element - Portion of video
that shows white paper on which
instructor is writing something
3) Instructor Element - Portion of videothat shows instructor talking Sample
images of the elements are shown in
Figure below.
Study elements are portions of videothat are different from one another in
terms of the following properties:
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 11/18
User Experience
• We define “User Experience” as a value between 0 and 1that
represents the desired level of user experience. This is given
as input by the user to the system.
• since images are seen with some delay, the user would be
more satisfied if that delay is lesser. Hence, a zero value
means that the delay experienced is large (to be chosen by
system administrator),and a value of one means that the
delay experienced is equal to one second.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 12/18
Calculating system parameters
• Calculating Sending Intervals
• User Experience of Study-Elements
•
Calculating network bandwidth and size• Finding Output Sending Interval
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 13/18
User Experience of Study-Elements
1) Presentation Element: 2) White Paper Element:
Delay Experienced(D2) = Time that user
sees the image of the slide−
Time that theslide actually started in the original video
User Experience(U2) = 1 Sec
Delay Experienced
Delay Experienced(D1) = Sending interval
User Experience(U1) = 1 Sec Delay Experienced
3) Instructor Element
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 14/18
Calculating network bandwidth and size
Network Overhead(NO) = Image Size
Sending RateSize Overhead(SO) = Total Size Of The Images Extracted From The
element
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 15/18
Video Size Comparison
Comparison of Size Reduction
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 16/18
CONCLUSION AND FUTURE WORK
Slide show based adaptation achieves far greater bandwidth
reduction than other techniques.
It can be used to adapt lecture videos so that they can be viewed onlow bandwidth network connections.
It should be observed that we assume that video of slides is shown
because this is important for offline viewing of the lecture.
Currently, for our method, tagging the boundaries of Study-Elements is assumed done manually typically at the time of production
of video.
We feel that this process could be automated by using a technique
called shot detection, that identifies the places in the video, where
major changes occur.
We also plan to implement the idea as a working system,
and take the user experience of actual users. This could be
then compared with the values obtained from the metrics.
7/30/2019 Ppt Sildes
http://slidepdf.com/reader/full/ppt-sildes 17/18
REFERENCES
[1] H.264. ‘‘http://ati.amd.com/products/pdf/h264 whitepaper.pdf’’.
[2] Shih-Fu Chang, Di Zhong, and Raj Kumar. Real-time content-based
adaptive streaming of sports videos. In CBAIVL ’01: Proceedings of the
IEEE Workshop on Content-based Access of Image and Video Libraries
(CBAIVL’01), page 139, Washington, DC, USA, 2001. IEEE Computer
Society.
[3] Ming-Ho Hsiao, Yi-Wen Chen, Hua-Tsung Chen, Kuan-Hung Chou,
and Suh-Yin Lee. Content-aware video adaptation under low-bitrateconstraint. EURASIP J. Adv. Signal Process, 2007(2):27 – 27, 2007.
[4] W. Tavanapong. A characteristics-based bandwidth reduction
technique
for pre-recorded videos. In IEEE International Conference on Multimediaand Exposition, page WP11, 2000.
[5] J. Xin and C-W Sun M-T. Lin. Digital video transcoding. Technical
report, Mitsubishi Electric Research Laboratories, January 2005. Digital
Video Transcoding, Proceedings of the IEEE, Vol. 93, Issue 1, pp. 84-97,
January 2005, IEEE Xplore (http://ieeexplore.ieee.org/Xplore