Use Google Colab and Kaggle Data with bonus: fastai2


I was just running through this process and thought it might be helpful for others:

What to do in Kaggle:
Step 1. Go to your Kaggle account next to Discussion and click the … and then Account.
Step 2. Scroll to API and click Create New API Token and a kaggle.json file will download.

What to do in Google Colab:
Step 3. Open a new Colab file

Step 4:
! pip install -q kaggle

Step 5:
add the following code:
from google.colab import files

Click choose files and upload the kaggle.json file

Step 6:
add the following code:
! mkdir ~/.kaggle
! cp kaggle.json ~/.kaggle/

Step 7:
add the following code:
! chmod 600 ~/.kaggle/kaggle.json

Step 8:
I prefer to actually mount Google Drive to put the files in the folders I want so I use this command:

from google.colab import drive
drive.mount(‘/content/gdrive’, forceremount=True) rootdir = “/content/gdrive/My Drive/”

You will have to accept and copy the auth code and paste that back into your notebook.

Step 9:
I prefer to CD into the folder I want to put my data into:

%cd ./gdrive/’My Drive’/[add drive here]

(this will confirm the directory you are in)

Step 10:
Download data by going to the competition you are working on and go to data. There you will see a line of code that says API. Like below:
! kaggle competitions download -c ‘name-of-competition’

I’m entering the Wheat one so this is the line (add ! and space before it):
! kaggle competitions download -c global-wheat-detection

For fastai2 in colab:
Add the following code to import:

!pip install fastai2

For vision run:
from import *


Please enter your comment!
Please enter your name here