load data as numpy array

[JunMa11]

Thanks for sharing the awesome work.

I'm not familiar with the zarr format. Would it be possible for you to share a demo on how to load the data as NumPy array?

[bentaculum]

You can write a numpy array to a persistent zarr array and then use the rest of the pipeline as is, here is an example:

import numpy as np
import zarr
from numcodecs import Blosc

raw = np.random.randint(0, 256, (256, 256, 256), dtype=np.uint8)
labels = np.random.randint(0, 2, (256, 256, 256), dtype=np.uint32)

x = zarr.open(
    store="test.zarr",
    path="volumes/raw",
    mode="w-",
    shape=raw.shape,
    chunks=(128,128,128),
    dtype=raw.dtype,
    compressor=Blosc(cname="zlib", clevel=3),
)
x.attrs.put({"offset": [0, 0, 0], "resolution": [5, 5, 5]})
x[:] = raw

y = zarr.open(
    store="test.zarr",
    path="volumes/labels/somestructure",
    mode="w-",
    shape=labels.shape,
    chunks=(128,128,128),
    dtype=labels.dtype,
    compressor=Blosc(cname="zlib", clevel=3),
)
y.attrs.put({"offset": [0, 0, 0], "resolution": [5, 5, 5]})
y[:] = labels

[JunMa11]

Hi @bentaculum ,

Thanks for your help very much.

How can I write the zarr file to numpy? since I'd like to build my own pipeline based on numpy format data.

[bentaculum]

Simply by slicing, here is a nice tutorial: https://zarr.readthedocs.io/en/stable/tutorial.html.