Detects and labels scenes in images or frames, returning timecodes and scene metadata for video annotation.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.