Limit google vision api text detection to specific region

Question

I am trying to use google vision API to detect text from camera preview. However, I want the detected text to be within a specific region/rectangle in the camera review.

Answer 1

Till date, there is no such feature in GVision. But I handled a similar situation by filtering at the backend of my application, as follows.

1. Get the GVision text using FULL_TEXT_DETECTION
2. Get the co-ordinates of user action, i.e., user touch, or mouse click
3. Get the application length & width of the mobile device
4. Make a proportion or ratio function from points 2 & 3
5. Using the value from point 4, get extract the text of based on GVision bounds of the blocks

Update: 4/8/2018

{
  "requests": [
    {
      "imageContext": {
        "latLongRect": {
          "minLatLng": {
            "latitude": 0,
            "longitude": 0
          },
          "maxLatLng": {
            "latitude": 0,
            "longitude": 0
          }
        }
      }
    }
  ]
}

minLatLng is TopLeft, maxLatLng is BottomRight of the your region

For more check: https://cloud.google.com/vision/docs/reference/rest/v1/images/annotate#ImageContext