Toolset for Run-time Dataset Collection of Deep-scene Information
Gustav Aaro, Daniel Roos and Niklas Carlsson
Paper:
Gustav Aaro, Daniel Roos and Niklas Carlsson.
"Toolset for Run-time Dataset Collection of Deep-scene Information",
Proc. IEEE MASCOTS workshop,
Nice, France, Nov. 2020.
(pdf)
Abstract:
Virtual reality (VR) provides many exciting new application opportunities,
but also present new challenges. In contrast to 360-degree
videos that only allow a user to select its viewing direction,
in fully immersive VR, users can also move around
and interact with objects in the virtual world.
To most effectively deliver such services it is therefore important to
understand how users move around in relation to such objects.
In this paper, we present a methodology and software tool
for generating run-time datasets capturing a user's interactions with
such 3D environments, evaluate and compare different object identification
methods that we implement within the tool, and use
datasets collected with the tool to demonstrate example uses.
The tool was developed in Unity,
easily integrates with existing Unity applications through the use of
periodic calls that extracts information about the environment using
different ray-casting methods. The software tool and example
datasets are made available
with this paper.
Software and datasets
To help build upon our work, below, we make available code and example datasets.
-
Sotware:
The software and code used in our paper is made available here:
-
2018 scripts: unity,
python
-
2020 scripts: Plan to add soon.
-
Datasets:
Links to example datasets are/will be added here:
-
Performance test: performance
-
Example use-case traces: viking 1, 2, 3, ... (to be added)
Note: If you use or build on our datafiles, code, or ideas in your research,
please include a reference to our paper
(pdf).