Updating the App¶
Instructions vary depending on how you installed the application. Please locate the appropriate instructions below that fit your specific needs.
Installed using conda¶
If you want to make absolutely sure that everything works as expected, you can always re-run the installation instructions to create an entirely new environment. This may also bear the benefit of security that your old version is still available if needed. Sometimes, as was the case from v1.0.x to v1.1.x, changes may be breaking changes. Meaning that data you created previously may not transfer 1:1 to the new version:
conda create -n track -c trexing trex
In the case mentioned the change was from Tensorflow 1 to Tensorflow 2 with a new network architecture. This changes the way that visual identification weights are saved. You were not able to load old-format weights (from v1.0.x) anymore in the new version (>1.1).
Simply open up an anaconda shell and activate your environment, e.g.:
conda activate tracking
and now all you need to do is:
conda update -c trexing trex
and follow the instructions on screen.
Older versions of TRex may use different python versions, which can cause the
conda update process to fail since it does not automatically update python along with TRex. In such a case, the easiest way is to reinstall the application as described in Installation and thus replace your current conda environment with a new one, or to manually mention the required python version as part of the update command, e.g.:
conda update -c trexing trex python=3.7
If you compiled the software yourself, then you simply need to execute:
git pull cd Application/build cmake --build .
Make sure that - if you customized the source code in your version - your changes do not block git from updating the repository in step one.