Words have not been tested at all as to how well they play with Google's Vision API.

Examples:
Can the API reliably identify object X?
Are there words that will trigger a false positive, e.g. identifying a pot as a pan?