Borda / kaggle_image-classify

Various Kaggle image classification challenges solutions
https://borda.github.io/kaggle_plant-pathology
MIT License
41 stars 12 forks source link

Question about dataset #10

Closed albertoceballos closed 3 years ago

albertoceballos commented 3 years ago

Is the plant pathology 2021 dataset used in the Pytorch Lightning notebook a resized version of the dataset? There are some commands that are commented out:

# ! apt-get install -qq -y imagemagick # ! mogrify -resize 960 train_images/*.jpg

Were these commands used to resize the dataset before running the notebook?

edgarriba commented 3 years ago

Yes, we generated a downscaled version of the dataset offline due to a bottleneck issue with loading large images with pil.

Borda commented 3 years ago

yes, you can also download it here from Kaggle - https://www.kaggle.com/jirkaborovec/plant-pathology-2021-fgvc8-960px