Abstract: A method and a system described herein provide optimizing image and/or video compression for machine perception. According to an aspect, the method comprises receiving a raw image frame from a camera sensor; detecting a predefined object in the raw image frame and marking a region around the predefined object within the raw image frame as ROI. Based on the ROI, a partitioning scheme, a prediction mode, and quantization parameter are determined for improving coding efficiency. Machine perception efficiency is improved by selecting a quantization parameter table used for compressing and encoding the raw image or video frame based on a selected machine vision task. The selection of the quantization parameter table is based on training of the selected machine vision task using cost function optimization.

Abstract: A computer-implemented method and corresponding system for processing sensor data associated with a vehicle is provided. The sensor data may be compressed or encoded with a dictionary according to sparse approximation theory, resulting in a sparse representation of the sensor data. Processing may further comprise detecting an event associated with the vehicle, wherein an event may be an accident recorded by sensors of the vehicle providing the sensor data. The detection of the event may be based on processing of the sparse representation of the sensor data alone without decoding the sparse representation. The detection of the event may further employ machine learning methods trained to the detection of an event from the sparse representation of the sensor data, or a combination of sparse representations of sensor data originating from a plurality of vehicles or sensors.

Abstract: A device and computer-executable method is provided for adaptively determining a sampling scheme to be applied at a first sensor from among a plurality of sensors for sampling sensor data values corresponding to a signal. A sparsifying transform for a subsequent sampling time window of the first sensor is predicted, wherein the sparsifying transform is determined based on a predictive model of the sparsity of the signal. Moreover, a subsampling parameter for the subsequent sampling time window is determined. The subsampling parameter corresponds to a number of sensor data values to be acquired within the sampling time window. This subsampling parameter is determined based on the predicted sparsifying transform. Further determined is a compressive sampling scheme for the subsequent sampling time window of the first sensor. The compressive sampling scheme is determined based on the predicted sparsifying transform.

Abstract: A computer-executable method and system for querying sensor data of a plurality of sensors is provided. Sensor data is received that comprising sensor data values sampled by a first sensor of said plurality of sensors according to a first compressive sampling scheme. The first compressive sampling scheme can be applied by the first sensor within a sampling time window and the received sensor data corresponds to samples of a signal within the sampling time window. The sensor data is stored in a first database. A frequency decomposition of the signal is computed based on a sparsifying transform associated with the first compressive sampling scheme and the received sensor data. The frequency decomposition comprises one or more frequency components. The one or more frequency components are stored in a second database. A query is received from a client. The query specifies an event that indicates a critical signal condition of a signal.