TAPAS-360°: a Tool for the Design and Experimental Evaluation of 360° Video Streaming Systems

TAPAS-360° is a tool that allows rapid prototyping of Adaptive Streaming control algorithms and viewport-adaptive algorithms for omnidirectional video streaming.

If you are a developer and you want to design an test a new Adaptive Streaming control algorithm or a viewport adaptive algorithm, please check the documentation placed in the doc/ directory.

If you are a user and you want to experiment with the control algorithms made available by TAPAS-360° you can follow the instructions given below in the "Usage" section.

Installation

For Ubuntu-based Linux distributions install the following building dependencies (also collected in dependencies.txt):

apt-get install --no-install-recommends python3-twisted
apt-get install --no-install-recommends python3-twisted-bin
apt-get install --no-install-recommends python3-twisted-core
apt-get install --no-install-recommends python3-twisted-web
apt-get install --no-install-recommends gstreamer1.0-plugins-*
apt-get install --no-install-recommends gstreamer1.0-tools
apt-get install --no-install-recommends gstreamer1.0-libav
apt-get install --no-install-recommends libgstreamer1.0-dev
apt-get install --no-install-recommends python3-gst-1.0
apt-get install --no-install-recommends python3-gobject
apt-get install --no-install-recommends python3-numpy
apt-get install --no-install-recommends python3-scipy
apt-get install --no-install-recommends python3-psutil
apt-get install --no-install-recommends gir1.2-gtk-3.0

Building TAPAS-360° docker image

TAPAS-360° is shipped with easy-to-use scripts that help with building and running a docker image to perform experiments with the tool.

To build the tapas-360 docker image simply run:

./build_tapas360_dockerimage.sh

Then, use the helper script

./run_tapas360_docker.sh

to run TAPAS-360°. Notice that the last script assumes you can run docker without sudo (see the Docker documentation).

Usage

Play a default playlist:

$ python3 play.py

Play a default playlist with a "conventional" adaptive controller:

$ python3 play.py --controller conventional

Play a playlist specified by its URL:

$ python3 play.py --url http://mysite.com/myplaylist.m3u8

Play a sample MPEG-DASH video:

$ python3 play.py --url http://yt-dash-mse-test.commondatastorage.googleapis.com/media/car-20120827-manifest.mpd

Play an omnidirectional content: Play a MPEG-DASH omnidirectional content (DASH playlist is courtesy of Bitmovin): ::

$ python3 play.py --vr True --url http://bitmovin-a.akamaihd.net/content/playhouse-vr/mpds/105560.mpd

Play omnidirectional content using the provided "conventional" viewport adaptive strategy (HLS playlist is courtesy of Bitmovin):

$ python3 play.py --vr True --view_controller conventional --url http://bitmovin-a.akamaihd.net/content/playhouse-vr/m3u8s/105560.m3u8

Play a playlist for logs, without decoding video:

$ python3 play.py --media_engine nodec

Play a playlist with a fake player (emulated playout buffer and no decoding):

$ python3 play.py --media_engine fake

Play only the highest quality of the playlist:

$ python3 play.py --controller max

Player options:

$ python3 play.py --help

To Enable debug just define the variable DEBUG to a value lager than 1:

$ DEBUG=2 python3 play.py

Emulating HMD movement with public datasets

TAPAS-360° allows to emulate the user's head position by reading a Comma Separated Values (.csv) file containing the angular data jointly with a timestamp related to playback time. The format is simply

time,alpha,beta,gamma

where:

• time is the timestamp in seconds
• alpha is the yaw angle in degrees
• beta is the pitch angle in degrees
• gamma is the roll angle in degrees

A list of publicly available datasets that can be easily used with TAPAS-360° includes:

1. http://dash.ipv6.enstb.fr/headMovements/
2. https://www.interdigital.com/data_sets/salient-360-dataset
3. https://github.com/V-Sense/VR_user_behaviour

It is our plan to provide helper scripts to convert those dataset to a format that is understood by the default HMD emulator provided by TAPAS-360°. Of course, it is always possible modifying the HMD behavior by, f.i., inheriting from the HMDEmulator class.

Documentation

Documentation can be built using sphinx. To build the documentation in HTML format:

cd docs/sphinx
make html

The documentation will be made available in docs/sphinx/_build/html. Many other options are available (f.i., Latex, epub, etc), run make to discover all of them.

Features to be implemented soon

[ ] Saliency maps data integration: to allow the design of saliency-based viewport adaptive algorithms

You are encouraged to submit feature requests in the Issues section.

How to cite this work

If you find our tool useful and you use it in your research please cite us as follows:

@inproceedings{10.1145/3394171.3414541,
  author = {Ribezzo, Giuseppe and De Cicco, Luca and Palmisano, Vittorio and Mascolo, Saverio},
  title = {TAPAS-360°: A Tool for the Design and Experimental Evaluation of 360° Video Streaming Systems},
  year = {2020},
  isbn = {9781450379885},
  doi = {10.1145/3394171.3414541},
  booktitle = {Proceedings of the 28th ACM International Conference on Multimedia},
  pages = {4477–4480},
  series = {MM '20}
}

Acknowledgments

This work has been partially supported by the Italian Ministry of Economic Development (MISE) through the CLIPS project (no. F/050136/01/X32). Any opinions, findings, conclusions or recommendations expressed in this work are the authors’ and do not necessarily reflect the views of the funding agencies.