Pre-requisites

In order to be able to run a video prediction, it is necessary to have a video which has already been uploaded into the Cloud Service.

Use the video media_id to send the PUT request to the endpoint predict/v2/videos/YOUR_VIDEO_MEDIA_ID/predict

Example of a successful response

When the request is sent, the response should a 200. status code and return the video media_id. This means the video prediction has been started. After a few minutes the prediction should be done and the prediction results should be ready.

{
  "media_id": "0d1aeafa-d360-47c2-afaf-c6149281c202"
}

To see how to poll the video prediction status and see video prediction results, please refer to GET /media endpoint.

Frame Rate

Note that, upon processing, all videos are transformed into a frame rate of 24 FPS (frames-per-second).

The Neurons API standardizes all video uploads to a frame rate of 24 frames per second (fps) for analysis purposes. This is a standard practice, meaning that regardless of the original video frame rate—whether it’s 25 fps, 30 fps, or any other—the API processes and analyzes it at 24 fps consistently.

Note: Videos with a width or height above 1024px will be downscaled to 1024px.

Additional considerations

Please note that, after video prediction has been triggered, it will take a few seconds for the media prediction status to switch from init into started. Also, re-calling the video prediction trigger endpoint when the status is started will not have any specific effect in the prediction currently being executed. A 202 response will be yielded with the message:

{
    "detail": "Prediction already started"
}