Azure Computer Vision APIs Returns default values - microsoft-cognitive

Question

I'm currently using the Computer Vision API to get image dimensions and also to detect color & adult content.
The API returns back the correct metadata containing the width, height, and format, but all the values for Color or Adult Content come back as default values (zeros/null/false/"None").
If it makes any difference, I'm calling the service in US West which is listed as having these features enabled. Also, I specify that I want these features in the request, which is why I receive the objects for these two features.
How do I get this information to be populated accordingly?
Example request I just used with today's Bing image. I'm using the Microsoft.Azure.CognitiveServices.Vision.ComputerVision package from nuget which contains the AnalyzeImageAsync method.
POST https://westus.api.cognitive.microsoft.com/vision/v1.0/analyze?
visualFeatures=Adult%2CColor HTTP/1.1
Ocp-Apim-Subscription-Key: <REMOVED>
User-Agent: FxVersion/4.6.26515.07 Microsoft.Azure.CognitiveServices.Vision.ComputerVision.ComputerVisionAPI/1.1.0.0
x-ms-request-root-id: 5e7823bb-4cac1f594af57fcf
x-ms-request-id: |5e7823bb-4cac1f594af57fcf.2.
Request-Id: |5e7823bb-4cac1f594af57fcf.2.
Content-Type: application/json; charset=utf-8
Content-Length: 129
Host: westus.api.cognitive.microsoft.com
{
"url": "https://outerpulsemedia.blob.core.windows.net/uploads/images/0be893af-9edc-416a-a779-929d2e2d1428-1529100079.jpg"
}
Response:
{
"adult": {
"isAdultContent": false,
"adultScore": 0.0,
"isRacyContent": false,
"racyScore": 0.0
},
"color": {
"dominantColorForeground": "None",
"dominantColorBackground": "None",
"dominantColors": [],
"accentColor": "000000",
"isBwImg": false
},
"requestId": "5b57186d-9976-4af7-97fd-e06b33fb0ecb",
"metadata": {
"height": 1200,
"width": 1920,
"format": "Jpeg"
}
}

Answer 1

The JPEG image is not, strictly speaking, correctly encoded. Specifically the EOI (end of image) marker is missing. The image should end with FF D9, whereas your image is ending simply in FF. This is why the image classifier is returning default values instead of true values.
While a case could be made that either the (a) return response is deceptive, or (b) the API should have returned a 400 Bad Request, it is probably more beneficial for you to try and understand how the image got truncated by a byte.