writing, there was no way for WebODM to
get GCP files (see below) into the
processing, but it is coming. Therefore,
datasets without EXIF data do not complete
in WebODM. Oddly enough, some datasets
that ran in regular ODM did not run in
Let’s start with a simple and small
dataset: caliterra. This is a construction
phase view of a planned community in
Dripping Springs, TX. First, use the terminal
to navigate to where you want to place the
files (or use the GitHub GUI client). Clone a
copy of the caliterra repository with git clone
DroneMap/ odm_data_caliterra.git or by
clicking the “Download ZIP” button in the
web interface. Open the folder and have a
This dataset has 77 images. Opening
one up and inspecting the EXIF data, we can learn a lot
about the image. We can see where it was taken, what
kind of camera was used, and other information about the
photograph (Figure 7).
You’ll notice in the example data chart (Figure 6) that
some of the datasets don’t have EXIF data associated with
the photos, and some have a GCP file. GCP stands for
ground control point. We’ll talk more about using GCPs
next month, but for now we just need to know that these
files contain the coordinates of known points of reference
on the ground and in multiple photos.
For datasets with no EXIF data, GCPs let us tie into the
real world coordinate system. When dealing with
coordinates, expect things to get messy as there are many
different reference datums, coordinate systems, and
misconceptions about converting. An error could result in
hundreds of meters of offset, which may or may not be
acceptable in the given application.
Processing Images with
Once you are connected to WebODM, you will see an
empty “First Project” and a few simple controls. Go ahead
and click the “Add Project” button and fill out the pertinent
information. Again, I’m using the caliterra dataset here.
SERVO 04.2017 49
Figure 5: There are nine sample datasets in the odm_data repository and one more
linked from the ODM documentation. You can clone or download each repository
individually. All together they are just over 13 GB.
Figure 6: I tried to process many of the example datasets with
both ODM and WebODM. Some would not stitch for unknown
reasons. It seems that ODM is more likely to succeed as WebODM
does not have as many options and is not as well tested yet.
Figure 7: EXIF data from an example photo shows many of the
camera parameters when the photo was snapped. More
information is available via the Canon and GPS tabs as well.