Pls see discussion at https://github.com/fastai/fastai/issues/1687
i thought i remember there’s this ubiquitous image called “lenna.png” in older lib such as openCV, and i suspect it is included in their dist.
I m proposing we do that here. The use case is for quick call up of images, and test data augmentation on them, and help communicate/reproducing issues and unit testing. Granted there are others ways to grab data (by URL) at runtime, this may not be a P1 feature. But would like to feedback.
Also, such an image could be useful for testing simple training by using lot of data aug, one may be possible to do a classifier of presence/absence of whatever prominent object shown in the template. But this is a more remote case.