List all segment IDs given precomputed segmentation volume

[jakobtroidl]

Is there a way to use cloud volume to generate a list of all segment IDs given a precomputed segmentation volume? Here's a pseudocode example of what I am looking for.

from cloudvolume import CloudVolume

vol_path = "https://storage.googleapis.com/neuroglancer/drosophila_v0/seg_190410_FAFB_v02_ws_size_threshold_200"
vol = CloudVolume(vol_path, parallel=True, progress=True)

vol.all_ids() # list all segmentation IDs in the volume as np.array

Thanks so much for your help.

[william-silversmith]

Hi Jakob,

There are two answers to this question.

1. uniq = cv.image.unique(cv.bounds, cv.mip) will download the entire dataset and efficiently decode the downloaded image chunk files without fully decompressing them. This will take a while... It's probably something like 300-400 GB at highest resolution.

2. By lucky chance, I've been doing compression experiments at mip 1 with this volume and can readily produce a file with this list pretty much on demand. Mip 0 is trickier just because it's so big.

[jakobtroidl]

Thanks for your swift response, Will. I will try (1), but if you could share the file for mip 1 for the above volume, I'd be super, super grateful.

[william-silversmith]

I can get you that list with a turnaround time of probably 24-48 hrs.