Error: all input arrays must have the same shape about docarray

[hanxiao]

when following your steps,

could you elaborate it by copy-paste your code here, thanks

[einfachluki]

[hanxiao]

based on the part of your code, I'm guessing your input images are not in the same shape, instead they are arbitrary size images. And if I'm correct about it, you need to add a preprocessing step to resize the images. You can do it by:

d = Document()
d.set_image_blob_shape(shape=(128, 128))

If you are working with DocumentArray as in README, then

from docarray import Document

def preproc(d: Document):
    return (d.load_uri_to_image_blob()  # load
             .set_image_blob_shape(shape=(128, 128))  # <----- ADD to unify the shape, 128 is just an example, use whatever makes sense to you
             .set_image_blob_normalization()  # normalize color 
             .set_image_blob_channel_axis(-1, 0))  # switch color axis for the pytorch model later

[einfachluki]

Thank you!