Siamese Capsule Networks
This addresses face verification challenges for computer vision applications, but is incremental as it adapts existing capsule networks to a new task.
The paper tackled face verification in controlled and uncontrolled settings, where entities have complex internal representations and few instances per class, by introducing Siamese Capsule Networks, which performed well against strong baselines, yielding best results in few-shot learning with unseen subjects.
Capsule Networks have shown encouraging results on \textit{defacto} benchmark computer vision datasets such as MNIST, CIFAR and smallNORB. Although, they are yet to be tested on tasks where (1) the entities detected inherently have more complex internal representations and (2) there are very few instances per class to learn from and (3) where point-wise classification is not suitable. Hence, this paper carries out experiments on face verification in both controlled and uncontrolled settings that together address these points. In doing so we introduce \textit{Siamese Capsule Networks}, a new variant that can be used for pairwise learning tasks. The model is trained using contrastive loss with $\ell_2$-normalized capsule encoded pose features. We find that \textit{Siamese Capsule Networks} perform well against strong baselines on both pairwise learning datasets, yielding best results in the few-shot learning setting where image pairs in the test set contain unseen subjects.