My team is starting a project that will use computer vision to identify animals. This is not a simple computer vision problem like: 'is it a frog?: yes|no`. The idea is that a camera takes photos of frogs in the wild based on motion detection, every time it senses movement it snaps a shot. So at the end of the day you might have 40 pictures of frogs in all sorts of variety. The computer vision app needs to be able to look through the images and uniquely identify the frogs, detecting that these three photos are of the same frog (frog A), these two photos are of a different frog (flog B), and so on.
The boss wants to stream line the project by using Amazon Rekognition, he thinks it can do this job. However, all my research tells me that it cannot. It could easily tell me that the photo is of a frog but I don't believe it could match frogs between photos.
Does anyone with experience on this know if Amazon Rekognintion can do the job, or possibly another web service? Possibly this is a lot of custom deep learning code and training.....hope not.