Reading Data from .npy file


I have the data set with the .npy format of images with size 64x64 grayscale images. how to use this to create a image databunch object?


I think this conversation might help you:

1 Like

I am trying sign language classification problem.
i am facing errors.
here is the code

please help me to get solved.


We might need a little more context. What errors are you getting?

There is a .png version of sign-language mnist images on kaggle here if its helpful,

Also if you can confirm where you got your dataset from - then I can replicate it accurately.


Hi Kieran,

Thank you.

The data-set i am referring from kaggle.

By the way i am not using the create_cnn but i create a Linear layer on my own as demonstrated in lecture 5 of DL1 in course-v3.

I am getting errors when getting learning rate and when using slice in fit one cycle method.

The data set which you are referring by images, i can able to create a model with it.

Thanks again for your kind help.


I can get the notebook running if you change the floats to long.

I tried to run a fit_one_cycle and it was terrible, but I don’t really look into too much detail into how the rest of the notebook is set up.

Hope this helps you get back on track with it.