ISIC-Archive-Downloader | ISIC Archive of lesion images | Download Utils library

You can download a subset of the archive by specifying how many images you would like. python download_archive.py --num-images 1000 If this option isn't present, the program will download all the available images. You can start downloading images from an offset. python download_archive.py --offset 100 This is useful for example if you would like to append upon a prior download. You can choose to download either only benign or malignant images. python download_archive.py --filter benign Note: If you would like k benign images instead of all the benign images, you could do python download_archive.py --num-images k --filter benign. You can choose to download the segmentation of the images python download_archive.py -s and the directory which they will be downloaded to. python download_archive.py -s --seg-dir /Data/Segmentations Some images have multiple segmentations offered, made with different skill level. You can choose a preferred skill level (e.g expert). python download_archive.py -s --seg-level novice That means that, when available, the script will download a segmentation with the preferred skill level. If no preference was given, the first available segmentation will be downloaded. Note: It has been suggested that sometimes segmentations tagged as 'novice' skill are more accurate than there 'expert' alternative. So perhaps relying the the 'expert' segmentations are always better can be incorrect. You can choose not to download the lesion images. python download_archive.py --no-images This might be useful if you would like to download only the descriptions of segmentation images. You can change the default directories the images and the descriptions will be downloaded into. python download_archive.py --images-dir /Data/Images --descs-dir /Data/Descriptions. You can also change the default amount of processes that will work in parallel to download the archive. python download_archive.py --p 16 But if you have no knowledge about this one, the default will be fine.
You can download a subset of the archive by specifying how many images you would like. python download_archive.py --num-images 1000 If this option isn't present, the program will download all the available images.
You can start downloading images from an offset. python download_archive.py --offset 100 This is useful for example if you would like to append upon a prior download.
You can choose to download either only benign or malignant images. python download_archive.py --filter benign Note: If you would like k benign images instead of all the benign images, you could do python download_archive.py --num-images k --filter benign
You can choose to download the segmentation of the images python download_archive.py -s and the directory which they will be downloaded to. python download_archive.py -s --seg-dir /Data/Segmentations Some images have multiple segmentations offered, made with different skill level. You can choose a preferred skill level (e.g expert). python download_archive.py -s --seg-level novice That means that, when available, the script will download a segmentation with the preferred skill level. If no preference was given, the first available segmentation will be downloaded. Note: It has been suggested that sometimes segmentations tagged as 'novice' skill are more accurate than there 'expert' alternative. So perhaps relying the the 'expert' segmentations are always better can be incorrect.
You can choose not to download the lesion images. python download_archive.py --no-images This might be useful if you would like to download only the descriptions of segmentation images.
You can change the default directories the images and the descriptions will be downloaded into. python download_archive.py --images-dir /Data/Images --descs-dir /Data/Descriptions
You can also change the default amount of processes that will work in parallel to download the archive. python download_archive.py --p 16 But if you have no knowledge about this one, the default will be fine.