Instead of blindly trying to extract data from video (which can be expensive from a computational perspective), a DeviceMode provides:

• A context for information extraction.
• Feedback to ensure that the capture of raw data is optimized to enable the extraction of information.
• A defined workflow that provides the desired information in a compact and intelligent form.

For example, devices that have motion detection today simply generate a message every time motion is detected. The DeviceMode provides:

• A context in which to capture frames.
• A feedback path to enhance the capture of frames to suite the information that is to be extracted.
• A workflow to extract information and filter it down to an essential DeviceMark or set of DeviceMarks that describe the contents of the Device.