How to estimate pixel size of images using deep NN?

Hi, this question is not specific to PyTorch but I was wondering if there is a way to estimate the pixel size or “image resolution” using deep learning. So, the goal would be to give an image to a model and have it estimate the size of each pixel (in millimeters). I imagine there should be works where the training was done in order to predict the resolution of images based a dataset of images with their resolutions as class variable. So, I looked up the literature but could not find anything so I am not sure whether this is possible or if I am just using the wrong keywords?

Interesting idea, but wouldn’t that depends on the depth as well?