Given a partial object scan, X
, the goal is to recover the complete scan,
Y
.
This challenge uses the 3DObjectTex dataset.
Provided:
- A set of complete scans,
Y
.
Not provided:
- The partial scans,
X
. They are generated by the participants.
Provided:
- A set of partial scans,
X
.
Not provided:
- The ground-truth scans,
Y
. (They will be released after the competition.)
Considered for the evaluation:
- the shape,
- the texture.
See evaluation for the definition of the metric.
The predicted complete mesh, Y'
, should be in .obj
or .npz
(see
formats).
The mesh colour information should be either:
- a single texture atlas,
- or RGB colour stored as vertex attributes.
A mixture of vertex colour attribute and texture mapping is not allowed.
The mesh geometry may be different than the input, X
.
The submission must be a .zip
archive that unfolds in this directory
structure:
eval/
<scan_name>/
<scan_name>-completed.(npz|obj)
.../
where <scan_name>
correponds to the name of the input scan.
The data files are arranged in the following directory structure:
train/
<scan_name>/
<scan_name>.obj
<scan_name>.mtl
<scan_name>_0.png
.../
test/
<scan_name>/
<scan_name>.obj
<scan_name>.mtl
<scan_name>_0.png
.../
eval/
<scan_name>/
<scan_name>.obj
<scan_name>.mtl
<scan_name>_0.png
<scan_name>-partial.obj
<scan_name>-partial.mtl
<scan_name>-partial.png
.../
For each scan, there is one subdirectory with a unique <scan_name>
,
e.g. model_1
, model_1208
...
The files are:
<scan_name>(.obj|.mtl|_0.png)
=Y
: The ground-truth mesh, i.e. the raw scan.<scan_name>-partial(.obj|.mtl|.png)
=X
: Partial scan generated fromY
.
See formats for the mesh format.