This is more a general question since I am wondering that there are no standard tools out there (found actually many approaches via google, but nothing serious).
**The application is as follows**:
- I generate training data in simulation and store everything into rosbags.
- Further, I want to import the data to python to run some deep learning framework (like TF or keras) on it.
Using the rosbag api to interface the bags directly is, from my point of view, no option since it is comparable slow when loading the files.
Thus, I wrote my own scripts to convert, in this case, images to npz files.
But this would mean that for any other file format, I have to write my own converter which is a bit exhaustive.
Are there any general or preferable approaches/tools/scripts out there for rosbag conversion to npz or h5?
To give a little overview, I'd like to start a list with approaches I found.
Will also extend this list from time to time:
- https://github.com/strawlab/bag2hdf5
- https://gitlab.com/rosbag-development-team/rosbag-hdf5
- https://gist.github.com/bigsnarfdude/eeb156dc7b4caca69f5b31037da54708
↧