ValueError: Image dimensions must be powers of 2

[Shivansh20128]

Hi @SaashaJoshi, there seems to be a problem with the first example given in the repository readme file. The following code gave a value error - "ValueError: Image dimensions must be powers of 2." Please look into this.

# INEQR Encoding Method
import torch.utils.data
from piqture.data_loader.mnist_data_loader import load_mnist_dataset
from piqture.embeddings.image_embeddings.ineqr import INEQR

# Load MNIST dataset
train_dataset, test_dataset = load_mnist_dataset()

# Retrieve a single image from the dataset
image, label = train_dataset[0]
image_size = tuple(image.squeeze().size())

# Change pixel values from float to integer
pixel_vals = (image * 255).round().to(torch.uint8)
pixel_vals = pixel_vals.tolist()

embedding = INEQR(image_size, pixel_vals).ineqr()

# Display circuit.
embedding.draw()

ValueError                                Traceback (most recent call last)
Cell In[1], line 17
     14 pixel_vals = (image * 255).round().to(torch.uint8)
     15 pixel_vals = pixel_vals.tolist()
---> 17 embedding = INEQR(image_size, pixel_vals).ineqr()
     19 # Display circuit.
     20 embedding.draw()

File /Users/Shared/Drive/Qiskit/piQture/piqture/embeddings/image_embeddings/ineqr.py:43, in INEQR.__init__(self, img_dims, pixel_vals, max_color_intensity)
     37 def __init__(
     38     self,
     39     img_dims: tuple[int, int],
     40     pixel_vals: list[list[list]],
     41     max_color_intensity: int = 255,
     42 ):
---> 43     NEQR.__init__(self, img_dims, pixel_vals, max_color_intensity)
     45     # Determine number of qubits for position embedding
     46     self.x_coord = int(math.log(img_dims[0], 2))

File /Users/Shared/Drive/Qiskit/piQture/piqture/embeddings/image_embeddings/neqr.py:33, in NEQR.__init__(self, img_dims, pixel_vals, max_color_intensity)
     27 def __init__(
     28     self,
     29     img_dims: tuple[int, int],
     30     pixel_vals: list[list],
     31     max_color_intensity: int = 255,
     32 ):
---> 33     ImageEmbedding.__init__(self, img_dims, pixel_vals)
     35     if max_color_intensity < 0 or max_color_intensity > 255:
     36         raise ValueError(
     37             \"Maximum color intensity cannot be less than 0 or greater than 255.\"
     38         )

File /Users/Shared/Drive/Qiskit/piQture/piqture/embeddings/image_embedding.py:44, in ImageEmbedding.__init__(self, img_dims, pixel_vals, color_channels)
     38 if (
     39     not all((isinstance(dims, int) for dims in img_dims))
     40     or all((isinstance(dims, bool) for dims in img_dims))
     41     or not isinstance(img_dims, tuple)
     42 ):
     43     raise TypeError(\"Input img_dims must be of the type tuple[int, ...].\")
---> 44 self.validate_image_dimensions(img_dims)
     45 self.img_dims = img_dims
     47 self.color_channels = color_channels

File /Users/Shared/Drive/Qiskit/piQture/piqture/embeddings/image_embeddings/ineqr.py:72, in INEQR.validate_image_dimensions(self, img_dims)
     70 for dim in img_dims:
     71     if math.ceil(math.log(dim, 2)) != math.floor(math.log(dim, 2)):
---> 72         raise ValueError(\"Image dimensions must be powers of 2.\")

ValueError: Image dimensions must be powers of 2."
}

[Shivansh20128]

Maybe we can change the image size while loading the dataset.

# Load MNIST dataset
train_dataset, test_dataset = load_mnist_dataset(img_size=(8, 8))

[Shivansh20128]

I also noticed that the project is not documented that well, as It only contains the installation steps and some details about the dataset loader. Do you have any document where I could read what the project is about, what it can do, and what the goals of the project are? If not, do you want to work together on making a roadmap for this project? Because I think this could be so much bigger...Do you have a community channel where we could discuss this?
Thanks

[SaashaJoshi]

Hi @Shivansh20128,
Thanks for pointing out the lack of documentation. It would be great to sit and plan out something to correct this. Let me know if you will be interested in helping in creating this documentation plan! Also, you may want to refer the issue #116

[Shivansh20128]

Great @SaashaJoshi ! Let's get to it then. Let me know a communication channel we could use for this. Meanwhile, I will make a PR for the above issue.
I have seen the issue #116, but the problem is there is a lack of information about the project in the documentation. We'll need to add some of the documentation ourselves before asking the open-source community to do it.

[SaashaJoshi]

Let me see how to setup a discord channel for piQture. I will get back to you.