Computer Vision News - May 2020

2 Summary Deep Learning Library 6 Running this code, we obtain the plot below showing a bunch of images before and after the augmentation pipeline is applied. In the last row the corresponding altered masks are displayed. Conclusions The purpose of this article is to give an insight into the world of a comprehensive python library through two main applications that could be tried out and applied in several projects. While the latter experiment represents an easy and flexible way to solve the problem of applying the same transformation to image and mask (a common deep learning/imaging requirement), the former can be an initial step to explore model uncertainty. A little taste of the powerful things that we can do with deep learning models and of something else that we can visualise other than the predictions themselves i.e. the degree of confidence in those predictions. Many other options are offered by imgaug and worth exploring but that’s it for now!