I worked on an image recognition tablet app last year where there were real symbols placed around a historic attraction. Users were required to explore this attraction and 'scan' the symbols using the camera in a tablet app.
Due to technical constraints, the users had to be a specific distance from the symbols for them to be recognised by the app, and hold the tablet still.
We included an on-boarding process to help them with this, which went as follows:
- Ensure that the user is positioned in front of a 'tutorial' symbol when they first start the 'experience'.
- Explain that the experience involves searching the attraction for symbols, which they must scan with their tablet.
- Display the camera screen, with tutorial text (and audio) which reads as follows:
- "Look at the symbol in front of you."
- "Hold your tablet up to the symbol."
- "Now position the symbol in the frame on your screen to scan it."
- Also on the camera screen is a semi-transparent frame, pictured below, which shows the frame in which the user must position the symbol.
- It uses four corners as the frame.
- Anything outside the frame is made darker to accentuate the inside of the frame.
In practice it worked, but it wasn't as intuitive to users as I would have hoped.
The question: Is image recognition such as this inherently unintuitive, or is there something we could have done differently (within our technical constraints)?