Man, worst case scenarios for sure on those RHW images! I just now checked Recent Observations of Red-Headed Woodpecker; in a sample of 23 images, 3 were not immediately identifiable. Not sure if they are (or will be) Research Grade. This is going to be far worse for birds and mammals and other harder-to-image groups. So based on a terribly small sample I’d guess the “bad photo” rate is under 10%. Then at worst you’d expect 1 or 2 out of every 10 images to be something frustrating where the best you can do is guess. That wouldn’t be a terrible experience unless you are a perfectionist or a thin skinned person who wants a high score as validation of your expertise.
Important caveat for any teachers however: game is a practice tool only and should never be used as a summative assessment or “test.” For that you want to use hand-picked images and can use any of several quizzing or test apps such as Quizizz or your favorite LMS or a piece of paper and a projector.