Using multiple input images in Tensorflow for object recognition

I'm new to Tensorflow and machine learning in general so please forgive my ignorance.

I'm designing a mechanical process that will separate various objects and take photos/webcam stream of each at different angles (e.g. by rotating them).

I can find many tutorials around object detection and classification, but they all seem to be centred on a single image or snapshot from a webcam. I can't find anything that uses multiple photos of the same object, e.g. at different angles, to improve the recognition process.

To justify my approsch - certain objects might look the same from one angle, but if you rotate them they can be separately identified - in the same way you might look at something in real life and rotate it in your hand.

Can anyone point to tutorials that take multiple image inputs?

Many thanks!