Augmented reality

A method comprising, by a computing system:

• sampling a plurality of frames of media content with a sampling frame rate, each frame corresponding to at least part of an image from a scene in a video captured by one or more cameras;
• automatically detecting one or more objects in at least one of the frames;
• for each of the detected objects, automatically associating an annotation corresponding to that object for display with the object, wherein each annotation comprises one or more graphical elements;
• determining whether there is global motion of the scene by detecting global changes in the frames corresponding to movement of the one or more cameras;
• determining whether there is local object motion by detecting movement of the one or more objects in the frames;
• and
• updating each annotation in successive frames at a new sampling frame rate based on the determination that there is global motion and/or local object motion.

One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

• sample a plurality of frames of media content with a sampling frame rate, each frame corresponding to at least part of an image from a scene in a video captured by one or more cameras;
• automatically detect one or more objects in at least one of the frames;
• for each of the detected objects, automatically associate an annotation corresponding to that object for display with the object, wherein each annotation comprises one or more graphical elements;
• determine whether there is global motion of the scene by detecting global changes in the frames corresponding to movement of the one or more cameras;
• determine whether there is local object motion by detecting movement of the one or more objects in the frames;
• and
• update each annotation in successive frames at a new sampling frame rate based on the determination that there is global motion and/or local object motion.