Iinvalid subscription key when using Microsoft Emotion API for Video in API Testing Console and in Python 2.7

Question

I only manage to use the Emotion API subscription key for pictures but never for videos. It makes no difference whether I use the API Testing Console or try to call the Emotion API by Pathon 2.7. In both cases I get a response status 202 Accepted, however when opening the Operation-Location it says

{ "error": { "code": "Unauthorized", "message": "Access denied due to 
invalid subscription key. Make sure you are subscribed to an API you are 
trying to call and provide the right key." } }

On the Emotion API explanatory page it says that Response 202 means that

The service has accepted the request and will start the process later. In the response, there is a "Operation-Location" header. Client side should further query the operation status from the URL specified in this header.

Then there is Response 401, which is exactly what my Operation-Location contains. I do not understand why I'm getting a response 202 which looks like response 401.

I have tried to call the API with Python using at least three code versions that I found on the Internet that all amount to the same, I found the code here : Microsoft Emotion API for Python - upload video from memory python-upload-video-from-memory

   import httplib
   import urllib
   import base64
   import json
   import pandas as pd
   import numpy as np
   import requests

   _url = 'https://api.projectoxford.ai/emotion/v1.0/recognizeInVideo'
   _key = '**********************'
   _maxNumRetries = 10

   paramsPost = urllib.urlencode({'outputStyle' : 'perFrame', \
                           'file':'C:/path/to/file/file.mp4'})
   headersPost = dict()
   headersPost['Ocp-Apim-Subscription-Key'] = _key
   headersPost['content-type'] = 'application/octet-stream'
   jsonGet = {}
   headersGet = dict()
   headersGet['Ocp-Apim-Subscription-Key'] = _key
   paramsGet = urllib.urlencode({})

   responsePost = requests.request('post', _url + "?" + paramsPost, \
   data=open('C:/path/to/file/file.mp4','rb').read(), \
   headers = headersPost)

   print responsePost.status_code

   videoIDLocation = responsePost.headers['Operation-Location']
   print videoIDLocation

Note that changing _url = 'https://api.projectoxford.ai/emotion/v1.0/recognizeInVideo' to _url = 'https://westus.api.cognitive.microsoft.com/emotion/v1.0/recognizeInVideo' doesn't help.

However, afterwards I wait and run every half an hour:

   getResponse = requests.request('get', videoIDLocation, json = jsonGet,\
   data = None, headers = headersGet, params = paramsGet)

   print json.loads(getResponse.text)['status']

The outcome has been 'Running' for hours and my video is only about half an hour long.

Here is what my Testing Console looks like Testing Console for Emotion API, Emotion Recognition in Video Here I used another video that is about 5 minutes long and available on the internet. I found the video in a different usage example

 https://benheubl.github.io/data%20analysis/fr/

that uses a very similar code, which again gets me a response status 202 Accepted and when opening the Operation-Location the subscription key is wrong

Here the code:

import httplib
import urllib
import base64
import json
import pandas as pd
import numpy as np
import requests



# you have to sign up for an API key, which has some allowances. Check the 
API documentation for further details:
_url = 'https://api.projectoxford.ai/emotion/v1.0/recognizeinvideo'
_key = '*********************' #Here you have to paste your 
primary key
_maxNumRetries = 10

# URL direction: I hosted this on my domain
urlVideo = 'http://datacandy.co.uk/blog2.mp4'

# Computer Vision parameters
paramsPost = { 'outputStyle' : 'perFrame'}

headersPost = dict()
headersPost['Ocp-Apim-Subscription-Key'] = _key
headersPost['Content-Type'] = 'application/json'

jsonPost = { 'url': urlVideo }

responsePost = requests.request( 'post', _url, json = jsonPost, data = None, 
headers = headersPost, params = paramsPost )
if responsePost.status_code == 202: # everything went well!
  videoIDLocation = responsePost.headers['Operation-Location']
  print videoIDLocation

There are further examples on the internet and they all seem to work but replicating any of them never worked for me. Does anyone have any idea what could be wrong?

Answer 1

The Video Feature of Emotion API retires October 30th, so maybe you should change your procedure to screenshots anyways.

But for your question: The API returns you an URL where your results are accessible. You cannot open this URL in your browser, this will give you the notice of "invalid key", instead you need to call over python again this URL including your key. I will post you my code how to get the score, I am using Python 3, so there might be some adjustments necessary. Only "tricky" point is getting the Operation ID, which is just the ID in the URL ( =location in my case) which leads to your request. Rest of the parameters like subscription key etc. is as before.

#extract operation ID from location-string
OID = location[67:]
bod = ""

try:
    conn = 
    http.client.HTTPSConnection('westus.api.cognitive.microsoft.com')
conn.request("GET", "/emotion/v1.0/operations/"+OID+"?%s" %params, bod, headers)
    response = conn.getresponse()
    data = response.read()
    print(data)
    conn.close()
except Exception as e:
print("[Errno {0}] {1}".format(e.errno, e.strerror))