VisionQL is a node app without a web server.

It is a wrapper around Google Vision API's face detect and label calls.

The result of the API call is in json and it cast to TypeScript interfaces.
With those types it is pretty easy to write queries.
The query has to be be written in TypeScript, they are canned queries.

They are methods of SimpleQuery class:

• faceCount()
• happyFaceCount()
• hasCat()
• hasDog()
• isMeme()

It should work without too much problem and well enough to experiment with.