How to create localised media collection

Hi All, We have a requirement where each media attribute can have multiple assets as shown in below example. E.g. Product has attribute "Product Image", which can have images as "left image", "right image" etc. (this list may vary for different attributes) and all these images should be localised. i.e. EN can have different set of images for above defined types and DE can have different set.For multiple assets, we can use media Collection, and we tried localising the MediaCollection, but it dint work. Could you please provide the best approach for this requirement? Thanks in advance.