What amount of data makes up a tensor?

I am just getting into making my own ML functions from scratch and I am having trouble understanding what exactly a tensor is. My current understanding is that it is a multidimensional matrix that is a representation of the data you want to process, but I am confused on how exactly that works.

If I have a dataset of images, is each image its own tensor? Each section of an image? Does the whole image set become one tensor? And …

current data deeplearning functions making matrix multidimensional process representation scratch tensor understanding