US 12,482,263 B2
Method for determining video segments to be transferred
Gregor Blott, Salzgitter (DE); and Jan Rexilius, Hannover (DE)
Assigned to ROBERT BOSCH GMBH, Stuttgart (DE)
Filed by Robert Bosch GmbH, Stuttgart (DE)
Filed on Sep. 13, 2022, as Appl. No. 17/931,660.
Claims priority of application No. 10 2021 210 337.5 (DE), filed on Sep. 17, 2021.
Prior Publication US 2023/0090806 A1, Mar. 23, 2023
Int. Cl. G06V 20/40 (2022.01); G06V 10/82 (2022.01); G06V 20/59 (2022.01)
CPC G06V 20/49 (2022.01) [G06V 10/82 (2022.01); G06V 20/41 (2022.01); G06V 20/59 (2022.01)] 16 Claims
OG exemplary drawing
 
1. A method for determining video segments, which are to be transferred, of a video, which is recorded in a vehicle using a camera in the vehicle, the method comprising:
analyzing, in the vehicle, multiple video segments of the video recorded in the vehicle with respect to a respective content of each of the video segments, and generating for each of the video segments, based on the analyzing, data relating to the respective content of the video segment, the data for each of the video segments including a semantic description of the respective content of the video segment and a respective identifier of the video segment;
conveying, from the vehicle, only the data generated for the video segments and not the video segments themselves to an extra-vehicular processing system;
receiving, by the vehicle from the extra-vehicular processing system in response to the conveyed data, information indicating which one or more of the video segments is to be transferred from the vehicle to the extra-vehicular processing system, the information including the respective identifier of each of the one or more of the video segments to be transferred; and
in response to receiving, from the extra-vehicular processing system, the information indicating which one or more of the video segments is to be transferred including the respective identifier of each of the one or more of the video segments to be transferred, transferring, from the vehicle, only the indicated one or more of the video segments and not others of the video segments, to the extra-vehicular processing system, wherein the vehicle transfers the indicated one or more of the video segments and not others of the video segments to the extra-vehicular processing system only after receiving, from the extra-vehicular processing system, the information indicating which one or more of the video segments is to be transferred;
wherein the analyzing of the multiple video segments includes analysis, using an artificial neural network in the vehicle, of what each of the video segments shows, and creating, for each of the video segments, using the artificial neural network, at least one feature vector by linking different pieces of information present in the video segment, and
wherein the at least one feature vector of each of the video segments is forwarded to a Long Short-Term Memory (LSTM) network in the vehicle, with which the semantic description of the respective content of each of the video segments is created or selected, so that a volume of data of the semantic description is less than a volume of data of the video segment, so as to provide data compression;
wherein the video which is recorded in the vehicle using the camera, includes (i) a video of a passenger compartment of the vehicle, or (ii) a video of surroundings of the vehicle.