CapsNets work very well on MNIST but there is a lot of research work to be done in terms of understanding if the same impressive results can be achieved on other datasets such as CIFAR, or on more general collections of images. If you are interested to know more please have a look at the following:

● Google's AI Wizard Unveils a new twist on neural networks: https://www.wired.com/story/googles-ai-wizard-unveils-a-new-twist-on-neural-networks/

● Google Researchers Have a New Alternative to Traditional Neural Networks: https://www.technologyreview.com/the-download/609297/google-researchers-have-a-new-alternative-to-traditional-neural-networks/

● Keras-CapsNet is a Keras implementation available at https://github.com/XifengGuo/CapsNet-Keras

● Geoffrey Hinton talks about what is wrong with convolutional neural nets: https://www.youtube.com/watch?v=rTawFwUvnLE&feature=youtu.be