High-order video event judgment method and device, equipment and medium

Abstract

The invention discloses a high-order video event judgment method and device, equipment and a medium. The method comprises the following steps: acquiring a high-order video of a target parking space, and equally dividing the high-order video into a plurality of video frame groups according to a preset number and an extraction interval; inputting the video frame group into a pre-created end-to-end detection model to obtain a parking space state of the target parking space, wherein the end-to-end detection model is a model obtained by training a deep neural network constructed based on a 3D convolutional neural network and a long short-term memory network by using a high-order video training set containing a parking space state label. The features of the input video frames are automatically extracted through the 3D convolutional neural network in the pre-constructed end-to-end detection model, and the parking space state of the target parking space in the high-order video is obtained in combination with the long short-term memory network, so that the accuracy of vehicle detection and parking space state detection is improved.

Description

technical field [0001] The invention relates to the field of high-level video monitoring, in particular to a high-level video event judgment method, device, equipment and medium. Background technique [0002] At present, the high-level video smart parking system uses cameras installed on the roadside and video analysis algorithms to detect whether the surrounding parking spaces are occupied. After the driver parks the vehicle in the parking space, the video analysis algorithm will automatically identify and record the characteristics of the vehicle and the license plate number. Parking time, after the vehicle leaves, it can still accurately identify the vehicle's departure, and then notify the driver of the charging settlement information according to the parking time of the vehicle, so as to complete the collection of parking fees. In the prior art, the vehicle detection frame is generated according to the characteristics of the vehicle, and then the entry and exit of the v...

AI Technical Summary

Problems solved by technology

In the prior art, the vehicle detection frame is generated according to the characteristics of the vehicle, and then the entry and exit of the vehicle are judged. However, the detection method of displaying the vehicle detection frame on the video frame according to the vehicle characteristics reduces the accuracy of vehicle state detection. Omissions and false positives of vehicle status events may be caused by missed detection of the vehicle detection frame, enlarged detection frame, inaccurate detection frame, etc.

Images