A while back I painstakingly created an EQ preset for my own everyday headphones (ATH-M50) based on published frequency response curves. These are the settings if anybody's interested:
Freq Gain Q
45Hz -4dB 3
120Hz -5dB 3
3KHz +5dB 8
9KHz -4dB 10
14.5KHz (high shelf) +4dB 0.7
However, after going to that trouble I don't actually use these corrections anymore. They don't make that much difference, and even if they did I simply won't make EQ decisions in headphones. Doing so only yields a mix that sounds good in those headphones, but probably won't translate well to other systems.
Now, I realize that some folks have no choice but to mix with headphones. They live in an apartment, like to work late at night when the significant other is sleeping, can't afford good speakers, or have a severely acoustically-challenged room. Maybe all of the above. For them, it makes sense to try and improve their headphones to the extent that it's possible.
However, I would suggest that when you demo this software be sure to perform a blind A/B test to verify that the difference is significant enough to justify the hassle. By hassle, I mean when you inevitably forget to bypass the correction before exporting and then wonder why your tune sounds so weird in the car. Don't forget to volume-adjust your A/B test, since the correction will almost certainly involve some boosts.