The existence of dark matter in galactic halos is inferred from the velocity dispersion of stars in a galaxy (sigma). It’s important to realize that this is a very rough kind of estimate because every aspect of it (as in the case of the M-sigma relation) is roughly estimated. Virial mass is inferred from average temperatures of gas by way of detexting X-rays. Central black hole mass is inferred through the M-sigma relation, and stellar masses are inferred through luminosity. Every so often a new paper comes out revising the overall estimates, but the proportion of DM to luminous matter stays pretty constant.

Now, in a sense this makes a “selfie” easier, because you just need a series of snapshots which help you make all of these inferences. On the other hand because we’re in the disk of our galaxy it is difficult to image because we have to deal with all of that hot interstellar gas and dust, and the extremely bright galactic core. I’m not sure how much, when you take into account a long period of time and many wavelengths can be observed, but I’d say we can observe the majority of our galaxy in at least one wavelength. Still, we’re not talking about “selfies” here, just really well educated guesses based on observations of our own galaxy and similar spiral galaxies.

So as far as inferring the existence of a dark matter halo in our own galaxy we’re pretty certain, and it seems to make up about 95% of the galactic mass. It’s based on something a lot more tenuous and complex than a “selfie” though. Once you’ve done your best to estimate the total luminous mass of the galaxy and it’s rough distribution though, and ruled out MACHOs with lending surveys all that’s left in the remaining narrow window is dark matter. The only other theory with any chance at all is MOND, but it has to thread crushingly narrow passages between observation and tests of GR.

Isn't it to do with the rotational velocity? If it isn't what you'd expect from the matter that you see, you need DM for it to add up.