3results about How to "Processing small amount of data" patented technology

This application relates to a method, apparatus, and storage medium for determining the drivable area of ​​a residential parking area. The method includes: acquiring an obstacle occupancy grid map and a vehicle trajectory route for residential parking area mapping; generating reference key points on the vehicle trajectory route; searching along the perpendicular direction of the vehicle's heading from the reference key points to both sides of the vehicle trajectory route until a target point is found, where the target point is the obstacle point closest to the reference key point on a straight line along the search direction, determined by the obstacle occupancy grid map; and obtaining a drivable area outline in the residential parking area by sequentially connecting the target points on the same side of the vehicle trajectory route, and determining the drivable area in the residential parking area based on the drivable area outline. This method can reduce the amount of obstacle occupancy grid map data required for vehicle trajectory planning.

This invention discloses a lightweight traffic video processing and multimodal intelligent perception system and method, comprising: a streaming media preprocessing module for performing availability detection and parameter configuration on the accessed video stream and outputting a stable video stream; a video polling and frame extraction module for establishing a video polling queue, extracting key frames from the video stream, and splicing them into a lightweight video stream; a deep learning algorithm module for using a lightweight visual model to perform preliminary traffic event detection on the lightweight video stream; a multimodal large model service module for secondary confirmation of detected traffic events, extraction of key information, and generation of event reports; and a perception and analysis platform module for realizing the visualization, hierarchical dispatch, and closed-loop management of traffic events. Compared with existing technologies, this invention reduces the computational power requirements for video processing through video stream polling and key frame extraction mechanisms, enabling lightweight access and processing of large-scale road network videos.

The present disclosure relates to an image processing method and device, electronic equipment and storage medium, comprising: acquiring a plurality of images collected by a plurality of image acquisition modules in the electronic equipment; wherein the field of view of different image acquisition modules is different; in the image preview process, determining a first image from the plurality of images according to an editing operation acting on the display screen of the electronic equipment; cutting off the part of the first image that has no overlap with the content of the editing range acted on by the editing operation to obtain a first image region, and displaying the first image region; in the image storage process after the end of the image preview, determining a second image from the plurality of images according to the editing operation acting on the display screen of the electronic equipment; wherein the image acquisition module of the second image is different from the image acquisition module of the first image region; cutting off the part of the second image that has no overlap with the content of the editing range to obtain a second image region; and fusing the first image region and the second image region to obtain a cropped target image.