AWS Documentation » Amazon Rekognition » Developer Guide » Detecting Objects and Scenes » Detecting Labels in a Video

Detecting Labels in a Video

Amazon Rekognition Video can detect labels, and the time a label is detected, in a video. For an SDK code example, see Analyzing a Video Stored in an Amazon S3 Bucket with Java or Python (SDK). For an AWS CLI example, see Analyzing a Video with the AWS Command Line Interface.

Amazon Rekognition Video label detection is an asynchronous operation. To start the detection of labels in a video, call StartLabelDetection. Amazon Rekognition Video publishes the completion status of the video analysis to an Amazon Simple Notification Service topic. If the video analysis is succesful, call GetLabelDetection to get the detected labels. For information about calling the video recognition API operations, see Calling Amazon Rekognition Video Operations.

GetLabelDetection Operation Response

GetLabelDetection returns an array (Labels) that contains information about the labels detected in the video. The array can be sorted either by time or by the label detected by specifying the SortBy parameter.

Note

GetLabelDetection doesn't support the hierarchical taxonomy or bounding box information that the image operation DetectLabels supports. GetLabelDetection returns null for the Parents and Instances attributes.

The following example is the JSON response of the GetLabelDetection. In the response, note the following:

• Sort order – The array of labels returned is sorted by time. To sort by label, specify NAME in the SortBy input parameter for GetLabelDetection. If the label appears multiple times in the video, there will be multiples instances of the (LabelDetection) element.

• Label information – The LabelDetection array element contains a (Label) object which contains the label name and the confidence Amazon Rekognition has in the accuracy of the detected label. Timestamp is the time, in milliseconds from the start of the video, that the label was detected.

• Paging information – The example shows one page of label detection information. You can specify how many LabelDetection objects to return in the MaxResults input parameter for GetLabelDetection. If more results than MaxResults exist, GetLabelDetection returns a token (NextToken) used to get the next page of results. For more information, see Getting Amazon Rekognition Video Analysis Results.

• Video information – The response includes information about the video format (VideoMetadata) in each page of information returned by GetLabelDetection.

{
  "Labels": [{
    "Label": {
      "Confidence": 99.03720092773438,
      "Name": "Speech"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 71.6698989868164,
      "Name": "Pumpkin"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 71.6698989868164,
      "Name": "Squash"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 71.6698989868164,
      "Name": "Vegetable"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 71.44749450683594,
      "Name": "Clothing"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 71.44749450683594,
      "Name": "Overcoat"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 71.44749450683594,
      "Name": "Suit"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 58.84939956665039,
      "Name": "Jar"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 58.84939956665039,
      "Name": "Porcelain"
    },
    "Timestamp": 0
  }, {
    "Label": {
      "Confidence": 58.84939956665039,
      "Name": "Vase"
    },
    "Timestamp": 0
  }],
  "NextToken": "BgPXS37fNUY+Hrd7jYbBoTk1I5LevLm/MV+dhCwbXxoVOgMi0di6xYSeL6/Dztya/Pflx9xxxx==",
  "VideoMetadata": {
    "Codec": "h264",
    "DurationMillis": 67301,
    "Format": "mov,mp4,m4a,3gp,3g2,mj2",
    "FrameHeight": 1080,
    "FrameRate": 29.970029830932617,
    "FrameWidth": 1920
  }
}