Overview

ISIC

ISIC datasets

http://isic-archive.com/

Currently using:

• ISIC_MSK-2_1

identified as possibly usable:

• ISIC_MSK-1_1
• ISIC_MSK-1_2
• ISIC_MSK-2_1
• ISIC_MSK-4_1
• ISIC_UDA-1_1
• ISIC_UDA-2_1

metadata and information: view ISIC/csv_metadata/

Running it all

To run the scripts and notebooks, follow these steps:

1. read and execute ISIC/notebooks_metadata_and_images/get_3_images.ipynb
2. sort files by e.g. using script ISIC/dataordering.py or split with ISIC/0_datasplitting.ipynb
3. update ISIC/preparing.sh
4. run notebooks in (numbered) order

Working with a Spot Instance

After setting up a spot instance, copy the IP, paste it in a new tab in your browser, log in (deep_learning) and do the following:

1. click New -> Terminal
2. Copy, paste and execute the following lines:

wget https://raw.githubusercontent.com/linoba/melanoma-classification/master/ISIC/preparing.sh
chmod +x preparing.sh
./preparing.sh
pwd

Continue with:

3. Copy the path that appeared after last line
4. Open the notebook you want to run
5. Make sure that the path to your datafolder is set to the path you copied above with the suffix "/data/"
6. Run your tests

Weight files

Our best result with file 4_110.h5 can be found here: https://files.fm/f/n69jjnpu

Cats_and_dogs

Code taken from tutorial.