Oracle Cloud Infrastructure Documentation

Stored Video Analysis

Vision lets you process stored videos to apply labels to individual frames, classify images, and detect objects and faces.

Three options are available for specifying a video:
  • Analyze a demo video
  • Upload a video from local storage
  • Upload a video from object storage
Supported features are:

You can use pretrained models or custom models with stored videos.

Label Detection

All the labels detected in the video are listed. Only the first ten are shown, click Show more to display all of them. Click a label to see when in the video time line it's applied. The labels detected at a specific time are listed separately. Click the left or right arrows by the time to change it and so see the labels at different times. The JSON request and response are available too.

Object Detection

All the objects detected in the video are listed. Only the first ten are shown, click Show more to display all of them. Click an object to see when in the video time line it's detected. The objects detected at a specific time are listed separately. Click the left or right arrows by the time to change it and so see the objects at different times. The JSON request and response are available too. Bounding boxes are placed round any object detected.

Text Detection

All the texts detected in the video are listed. Only the first ten are shown, click Show more to display all of them. Click a text to see when in the video time line it's detected. The texts detected at a specific time are listed separately. Click the left or right arrows by the time to change it and so see the texts at different times. The JSON request and response are available too. Bounding boxes are placed round any text detected.

Face Detection

All the faces detected in the video are listed. Only the first ten are shown, click Show more to display all of them. Click a face to see when in the video time line it's detected. The faces detected at a specific time are listed separately. Click the left or right arrows by the time to change it and so see the faces at different times. The JSON request and response are available too. Bounding boxes are placed round any face detected.