Whoa! It looks like your Crowbox is ready to move on to Phase Two. If you have a number of Crows coming to eat regularly, as shown in your photos, they are probably ready to start learning Phase Two interaction.
You'll definitely need to deal with that squirrel!
As for identifying individual birds, it's very tricky, as Josh pointed out. I collected hundreds of photos of various Stellers Jays visiting our prototype Crowboxes and starting looking (by hand) for distinguishing features that might have helped in developing a computer-vision solution for identifying them. Many of our Jays had what appeared to be distinguishing facial features:
What I learned over watching these individuals for several months is that these facial features come and go. Feathers fall out and grow back. Small bumps, scratches, and wounds appear and then heal- it was very frustrating.
The closest I got to being able to positively identifying individuals visually involved the feet and legs- each bird appeared to have a unique pattern of reticulation across the bare section of their legs, and on the feet and 'toes'. Unlike the facial features, these features of the feet seemed to remain constant. This method brings with it a bunch of challenges, though- You need a good camera and good lighting conditions to view the black legs of a Crow well enough to distinguish these features in the skin. Also, the corvids I observed didn't exhibit consistency in the way they landed or stood upon the Crowbox, so I could not rely on any individual bird to present itself consistently to the camera. Sometimes we'd have a good view of the left leg, other times the right leg, and at other times, a rear view of both legs. It was pretty easy to do the identification work by hand/eye, but I never got very excited about trying to develop a computer vision solution for this problem.