this is more like a brainstorming topic, any suggestion will be welcome. Suppose there are two drone flying over a sea area. There is a big group of ships (following picture 1). each drone will capture the area only one time from different angle and the picture may not contain all the ships(picture 2 and picture 3).
Figure 2. the picture captured by drone 1
Figure 3. the picture captured by drone 2
I am thinking a way (no matter the traditional way or deep learning) to associate those spotted ships, i.e. ship4 and ship6.
PS: I have tried to use Siamese Network from another stackoverflow topic or Traditional techniques like SIFT SURF etc.
I hope there is a robust end2end way to find out the same objects from crowds.