I have been looking through research and cannot find too much on this topic. I believe that I might be searching for the wrong terms.
What I am trying to find is any literature on classifying a group of correlated and different images into a single class. Such that each image might have its own probability of being in a class but an overall meta probability is calculated for the entire group of images.
Almost as though additional images were used to either increase or lower the probability of belonging to a class.