Is there a list of all possible AWS Rekognition “scene” labels?

Question

Problem:

I want to separate out "objects" from "scenes" in AWS Rekognition API responses . Currently, the JSON responses lump "scenes" and "objects" all-together, eg "dog" and "beach" are side-by-side which is undesirable for my purposes.

Here's an example of a Rekognition DetectObjects API response exhibiting this behavior. Notice that "Indoors" (a concept, closer to a setting) and "Human" (an object) are lumped together:

# AWS Rekognition DetectFaces API Response JSON:
{
    "Labels": [
        {
            "Name": "Interior Design",
            "Confidence": 97.04463958740234,
            "Instances": [],
            "Parents": [
                {
                    "Name": "Indoors"
                }
            ]
        },
        {
            "Name": "Indoors",
            "Confidence": 97.04463958740234,
            "Instances": [],
            "Parents": []
        },
        {
            "Name": "Human",
            "Confidence": 92.32145690917969,
            "Instances": [],
            "Parents": []
        },
        {
            "Name": "Person",
            "Confidence": 92.32145690917969,
            "Instances": [
                {
                    "BoundingBox": {
                        "Width": 0.7570410966873169,
                        "Height": 0.8664773106575012,
                        "Left": 0.00028589036082848907,
                        "Top": 0.10754377394914627
                    },
                    "Confidence": 92.32145690917969
                }
            ],
            "Parents": []
        },
        {
            "Name": "Costume",
            "Confidence": 92.03617858886719,
            "Instances": [],
            "Parents": []
        },
        {
            "Name": "Clothing",
            "Confidence": 87.49675750732422,
            "Instances": [],
            "Parents": []
        },
        {
            "Name": "Apparel",
            "Confidence": 87.49675750732422,
            "Instances": [],
            "Parents": []
        }
    ],
    "LabelModelVersion": "2.0",
    "ResponseMetadata": {
        "RequestId": "6ddc0be3-afa5-49d9-8984-008e86fec1ae",
        "HTTPStatusCode": 200,
        "HTTPHeaders": {
            "content-type": "application/x-amz-json-1.1",
            "date": "Wed, 28 Oct 2020 01: 49: 19 GMT",
   "x-amzn-requestid": "6ddc0be3-afa5-49d9-8984-008e86fec1ae",
            "content-length": "766",
            "connection": "keep-alive"
        },
        "RetryAttempts": 0
    }
}

What I'm hoping for:

Somewhere there is a complete list of all possible "scenes" and "concepts" that can appear in a Rekognition API response , as separated from all possible "objects". Someone out there can tell me where to find that list.

What I've tried:

Searching for that list. Also manually combing through many examples of responses and compiling a set of possible "scenes" by hand. This is a fairly broke approach, the tedium of which prompted this question.

Thank you!

Answer 1

Amazon just released an update for Rekognition:

AWS customers can now view all the labels supported by Amazon Rekognition

Customers can download the list of supported labels and object bounding boxes from our documentation page or from the 'Label detection' tab of the Amazon Rekognition Console . In addition, on the Rekognition console, customers can use a search bar to easily check whether their label is already supported or not. Using the same interface, customers can request new labels that they would like Amazon Rekognition to support, or provide any other product feedback.