Our Deepfake Detection API identifies whether or not an image or video query is a deepfake. Similar to our other detection models, this API product locates faces in an image or frame of video. For each detected face, this model outputs a bounding box for its location, a classification, and accompanying confidence score.

The output object for Deepfake Detection lists the following information for every detected face:

  • The geometric description of the detected bounding box (vertices and dimensions).
  • The predicted class for the detection (yes_deepfake or no_deepfake).
  • The confidence score for the detection (on a scale of 0 to 1).
{
   "output":[
      {
         "time":0,
         "bounding_poly":[
            {
               "vertices":[
                  {
                     "x":296.89046664237975,
                     "y":58.84569320678711
                  },
                  {
                     "x":735.0210741043092,
                     "y":58.84569320678711
                  },
                  {
                     "x":735.0210741043092,
                     "y":702.5396911621093
                  },
                  {
                     "x":296.89046664237975,
                     "y":702.5396911621093
                  }
               ],
               "dimensions":{
                  "top":58.84569320678711,
                  "bottom":702.5396911621093,
                  "left":296.89046664237975,
                  "right":735.0210741043092
               },
               "classes":[
                  {
                     "class":"no_deepfake",
                     "score":0.000009033866263052914
                  },
                  {
                     "class":"yes_deepfake",
                     "score":0.9999909400939941
                  }
               ],
               "meta":{
                  "type":"face",
                  "score":0.9413483142852783,
                  "id":"5ef76840-51bb-11ed-8d15-713d6434ef35"
               }
            }
         ]
      }
   ]
}