search

nmi-lab / decolle-public

GNU General Public License v3.0
41 stars 22 forks source link

Tutorial? #18

Open GiorgosAlexakis opened 2 years ago

GiorgosAlexakis commented 2 years ago

Is there any tutorial notebook? It would be really helpful . Can I try another DVS dataset(with the same file format) easily or do I need to implement the conversion on my own? For example the action recognition dataset from here uses the aedat format. https://github.com/CrystalMiaoshu/PAFBenchmark

weinman commented 2 years ago

(This issue should probably go in the torchneuromorphic repo, but anyway...)

By way of analogy with the DVSGestureDataset It looks like you'd be able to create a subclass of NeuromorphicDataset and implement all the same methods in a very similar way.

It seems as though the repo you sent has data in a slightly different format, because the bitmasks it uses for the fields are narrower than those in the IBM DVS data (i.e., contrast this with that given by IBM here, and implemented here.

Of course, the other thing you'll have to worry about is labeling the data streams if you're doing supervised learning.

Long story short, there's a bit of conversion to do; it won't be turnkey, but you've got two templates to guide you from one (PAFBenchmark) to the other (DVSGestures in torchneuromorphic). Good luck!