Issues with Data Visualization and Model Evaluation

[IsmailAM1999]

Dear developers,

Thank you for your hard work and for providing such a comprehensive dataset. I have encountered a few issues after exporting records to frame-based representation. With data visualization and model evaluation, I have encountered some version issues. Should I stay on Python 2.7 or switch to a different version?

Additionally, I noticed that "dvs_frame_60x80" and "aps_frame_60x80" do not exist in the hdf5 files. Are they generated by the resize_int16 function in data visualization and then appended to the hdf5 file?

Thank you again for your interest in my request.

[tobidelbruck]

Thanks for your interest. Yes, unfortunately you should stick with 2.7 at least to generate your desired data files... some of the code uses some multiprocessing that is pretty tricky (the viewer in particular) and it is broken in python 3 as I recall. About the small DVS and APS frames, I don't know about them. I never used them. BTW, please check the issues, I recall that it is possible there is a pull for python 3 compatibility that we never got around to merging. If you do a fork we are happy to merge pulls if you find some improvements. Again, please check the issues because there are some useful threads in there as I recall. Tobi

[IsmailAM1999]

Thank you for your quick response.

Indeed, I noticed it during the export, but the export script works very well in the end.

Yes, I will be glad to propose a Python 3-compatible version later. For the moment, I am still focused on using the dataset and how to incorporate it into a training / evaluation model. If you have any ideas to propose for the frame-based exported version, it would help me a lot.

Once again, thank you for your efforts and for this rich dataset.