In my understanding, facial recognition is more about measuring distances between “landmarks”, such as eye-to-eye, nose-to-eyes, eyes-to-ears, etc.
Style transfer mostly changes the texture of strokes/fill, so it shouldn’t affect face recognition much.