I have a dataset of two different type of images. Say, I have images of a person and his all 10 fingerprints. I want to create a relation between them to predict one from another. How I can do that and which architecture is suitable for this problem or similar type of problem.
I would try a pair of separate deep image embeddings with a contrastive loss. The idea is similar to the Siamese network architecture. In Siamese networks the pairs of images are of the same type - so both input images are fed through the copy of the same network. In your case the images are of different kinds, so I would just have separate nets for person images and fingerprints.