Fig. 3From: From compressive sampling to compressive tasking: retrieving semantics in compressed domain with low bandwidthFlowchart of the proposed joint SCI and SCV framework. SCI records the high-speed dynamic scenes and sends the coded measurements to the measurement domain SCV for fast semantic information retrieval. When fine semantic information is required for some regions of interest (ROI), the ROI selection signal will be generated and sent to the video domain SCV. Then, the video domain SCV will be awakened and conduct detailed semantic information retrieval after reconstructing the ROI videos. While providing semantic information to the terminal application, the SCV modules also send feedback to the SCI system to improve its self-adaptability and robustnessBack to article page