Missing description in Web Detection feature of Google Cloud Vision API?

Question

I'm using the [Web Detection][1] feature of Google Cloud Vision API. However, for some images, the JSON response I receive don't have description parameters for some entities. On looking further, I found that description is missing for the entities whose id start with "/t/" and description is present for most of the entities whose id starts with "/m/". Can anyone suggest how should I go about this? Is this a bug or is this supposed to behave like this only? Also, is there any way where I can get some more details on the entities id and their syntax?

Here is the sample web detection JSON output with entity id starting with "/t" & "/m" having no description.

{
    "webEntities": [
      {
        "entityId": "/m/013_1c",
        "score": 0.608,
        "description": "Statue"
      },
      {
        "entityId": "/t/21mxcct4492j5",
        "score": 0.6404
      },
      {
        "entityId": "/m/0jg24",
        "score": 0.5815,
        "description": "Image"
      },
      {
        "entityId": "/g/11b77b4nf8",
        "score": 0.4837,
        "description": "2018"
      },
      {
        "entityId": "/t/24mypdx4svpvn",
        "score": 0.3909
      },
      {
        "entityId": "/m/0svqtrf",
        "score": 0.3664
      },
      {
        "entityId": "/t/2cvnxcsw4b8sf",
        "score": 0.3552
      },
]

People love to help whenever it's quick (e.g. they don't have to click on external links). Is there a way that you include some of the data from the screenshot into your question as code listing instead? — Cherusker, Jan 21 '19 at 17:56

score 0 · Answer 1 · answered Aug 22 '19 at 19:50

Remember that The Vision API requires images to be a sufficient size so that important features within the request can be easily distinguished. Sizes smaller or larger than these recommended sizes may work. However, smaller sizes may result in lower accuracy, while larger sizes may increase processing time and bandwidth usage without providing comparable benefits in accuracy. You can validate the Supported Images to determine if this could be the root cause of your issue.

I suggest you to try with your inputs in the Vision API web page; additionally, the Web detection tutorial could be useful to compare your results, the source code it's available in Github.

Missing description in Web Detection feature of Google Cloud Vision API?

1 Answers1