High toss act detection method and system based on background modeling and deep learning

Abstract

The invention discloses a high toss act detection method based on background modeling and deep learning. The method comprises the following steps: S1) acquiring front S frames of images of a monitoring camera to establish an initial Gaussian mixture background model B; S2) reading the newest image frame from the monitoring video stream, matching the newest image frame with the background model B, determining a foreground and a background, and updating the background model B; S3) carrying out feature extraction recording on a pixel region of the foreground part through a CNN, and removing a wrong foreground region; S4) repeating the steps S2 and S3, and performing feature matching on moving objects at different moments to form a matching set Z; S5) after the matching in the step S4 is finished, carrying out time sequence analysis on the matching set Z through LSTM, determining a real high toss act event, recording the event, and giving an alarm in time. The invention also provides a system for implementing the high toss act detection method based on background modeling and deep learning. According to the method, high toss act detection can be rapidly carried out on site deployment, and an object with the minimum size of 2 pixel points can be detected.

Description

technical field [0001] The invention relates to the field of deep learning and image processing, in particular to a high-altitude parabolic detection method and system based on background modeling and deep learning. Background technique [0002] With the development of the city, in order to save land, the floors are built higher and higher, high-rise residential buildings emerge in endlessly, and high-altitude parabolic phenomena also occur frequently. This uncivilized behavior is called "pain hanging over the city". It will cause great social harm. Studies have shown that an egg weighing 30 grams dropped from the 4th floor will make a swollen bump on the top of the person's head, dropped from the 18th floor, it can break the human skull, and dropped from the 25th floor, the impact force Enough to kill. In response to the frequent occurrence of injuries caused by high-altitude parabolic objects in recent years, the "Criminal Law Amendment (Eleventh)" that came into effect ...

AI Technical Summary

Problems solved by technology

Considering that the background modeling method is easily affected by the foreground pixels, it is difficult to achieve accurate detection when the background and foreground pixels are relatively close, especially for small targets; while only using the deep learning detection model, the general It is difficult to obtain a training sample due to the high-altitude parabola; and the high-altitude parabola is a continuous event, and the convolutional neural network alone cannot reflect its continuity in time, which will increase the detection system. Occurrence probability of missed detection and false alarm

[0005] From the above analysis, it can be seen that the existing high-altitude parabolic detection algorithm has certain limitations, and cannot meet the detection requirements of multi-type target detection and rapid deployment at the same time.

Images