New Dataset from Folder

Hi,

How do you actually visualize a dataset that is not acquired via untar_data? I have downloaded a dataset manually but for some reason I couldn’t visualize it nor work on it. The code couldn’t simply identify the path. I get this error: FileNotFoundError: [Errno 2] No such file or directory: ‘/users/downloads/cars/ss’. No need to mention that the path is correct. I tried the ‘old’ method used in the 2018 course and it worked fine.

To exemplify it: when i run the code on my machine (not on GCP) using the classic approach of fastai-cpu, and using the old notebooks of 2018, the code identifies my path normally and allows me to work on my own datasets. However, when I work on the cloud and the 2019 course the code only allows me to work on datasets acquired from untar_data. It won’t identify my local datasets, saying that the path doesn’t exist. Which is absurd.

Did anybody encounter the same issue?

Thanks!

Hi !

When you’re working on the cloud you have to think of the cloud as a computer in and of itself, that you’re accessing through your browser. This cloud computer can’t access your local files. You have to upload the images to the cloud computer to access them with it.
How to do that ? You have several ways :

  • If you downloaded the dataset from the internet by clicking on a link, there’s this super handy chrome extension called CurlWget that provides you with a command you can just copy and paste in the terminal of your cloud computer. Just click on the link to download the dataset, and then on the extension icon, and copy the text.
  • If the files you have are only on your local computer, you can upload them to a cloud storage provider like Dropbox or Google Drive and then apply the first method, with CurlWget.
  • If you don’t want to do that or can’t, you can use the secure copy command. Please see here for example on how to do that.
1 Like

Hi Pierre,

Thank you for your quick response. Once I download the dataset via my terminal (GCP), where can I then locate it to use it on the notebook?

Thanks!

Basically I tried to download a dataset online - I retrieved the link and pasted it on my Google Cloud Terminal, however, I don’t know where the dataset is located. I am struggling to find the link which would allow me to use the untar_data function.,