How to access data in a data set from notebook

Question

How to access data in a data set from notebook

Oren Koler 0

Hi

I'm trying to get image with and height that is stored in a dataset

with Image.open(image_path) as img:
                width, height = img.size

The URI is something like this:

azureml://subscriptions/7373be5f-9f84-4dab-a88e/resourcegroups/******/workspaces/azure-ml-X/datastores/workspaceblobstore/paths/LocalUpload/d1f07cd11d239601c6cf24/XXX/images/20.png

I get an error:
No such file or directory: azureml://subscriptions/7373be5f-9f84-4dab-a88e/resourcegroups/******/workspaces/azure-ml-X/datastores/workspaceblobstore/paths/LocalUpload/d1f07cd11d239601c6cf24/XXX/images/20.png

YutongTie-9091 54,031 Reputation points Moderator

2024-08-28T21:48:13.1733333+00:00

Hello Oren,

Thanks for reaching out to us, could you please share the steps or the document you are referring to so that we reproduce this issue?

Regards,

Yutong

Oren Koler 0

In the notebook:

from jsonl_converter import ConverterTest

converter = ConverterTest(URI_path)

In json_converter.py:

class ConverterTest():
   
    def __init__(self,     image_path):
		with Image.open(image_path) as img:
                width, height = img.size

Oren Koler 0

Also, if I use Azure SDK in the notebook:

# Initialize the file system
fs = AzureMachineLearningFileSystem(uri)

# List files in the directory
paths = fs.glob('UI/images/*')
print("Files in directory:")
for path in paths:
    with fs.open(path, 'r') as file:
        print(f"Contents of {path}:")

Then i works great.
The python file (json_converter.py) has no imports of azure.

Maybe that has somtheing to do with it

Oren Koler 0

This code works fine in notebook:

fs = AzureMachineLearningFileSystem(uri)

# List files in the directory
paths = fs.glob('UI/...../images/*')
print("Files in directory:")
for path in paths:
    with fs.open(path, 'r') as file:
        image_data = file.read()
        # Use BytesIO to handle the image data
        with BytesIO(image_data) as img_file:
            # Open the image with PIL
            with Image.open(img_file) as img:
                # Get image width
                width, height = img.size
                print(f"Image: {path} | Width: {width}, Height: {height}")

Your answer

YutongTie-9091 54,031 Reputation points Moderator

2024-08-28T21:48:13.1733333+00:00

Hello Oren,

Thanks for reaching out to us, could you please share the steps or the document you are referring to so that we reproduce this issue?

Regards,

Yutong
Oren Koler 0 Reputation points

2024-08-29T10:19:28.18+00:00

In the notebook:

from jsonl_converter import ConverterTest converter = ConverterTest(URI_path)

In json_converter.py:

class ConverterTest(): def __init__(self, image_path): with Image.open(image_path) as img: width, height = img.size
Oren Koler 0 Reputation points

2024-08-29T10:22:22.43+00:00

Also, if I use Azure SDK in the notebook:

# Initialize the file system fs = AzureMachineLearningFileSystem(uri) # List files in the directory paths = fs.glob('UI/images/*') print("Files in directory:") for path in paths: with fs.open(path, 'r') as file: print(f"Contents of {path}:")

Then i works great.
The python file (json_converter.py) has no imports of azure.

Maybe that has somtheing to do with it
Oren Koler 0 Reputation points

2024-08-29T10:26:54.1366667+00:00

This code works fine in notebook:

fs = AzureMachineLearningFileSystem(uri) # List files in the directory paths = fs.glob('UI/...../images/*') print("Files in directory:") for path in paths: with fs.open(path, 'r') as file: image_data = file.read() # Use BytesIO to handle the image data with BytesIO(image_data) as img_file: # Open the image with PIL with Image.open(img_file) as img: # Get image width width, height = img.size print(f"Image: {path} | Width: {width}, Height: {height}")