iOS11 displays a mini-screen shot to the lower left of the ; Apple could presumably invoke a high level camera response in the system software anytime the camera is accessed that shows a mini camera view on top of whatever the active app (maybe in the lower left so it doesn't obstruct an app) to show that that the camera is active and which one is in use (forward or rear). This would be similar to the view of yourself when doing a FaceTime call.
If the user sees window and did not want their picture/video captured , they could immediately close the malicious app.
Additionally, Apple could allow the user to touch the displayed mini-image (which could be capturing/sending live data to the malicious app) and when the displayed mini-image is touched, the mini-image grows to almost completely cover the screen and includes system-level software buttons, one of which says "turn off camera"? If pressed, the cameras turn off (in the active app) and the user can continue to stay in the app (assuming they still get value from the App but don't like the camera capturing them without their explicit consent).
Another software button in the camera mini-system-app might says "use camera, close this window"; to this, the app would continue running the camera but get rid of the visual obstruction of the mini-camera window.
The user could be offered to let the app use the camera (1) only when in the app, or (2) in the background (allowing the user to go to a different app but let the camera continue capturing info if the user and app developer want that experience.