A remnant detection method based on yolo target detection

Abstract

The invention discloses a remnant detection method based on YOLO target detection, and relates to the fields of intelligent monitoring, computer vision and deep learning. The present invention obtains the target category and its corresponding specific coordinates in each frame of image data through real-time detection of the YOLO target. Non-object targets such as pedestrians and animals are accurately filtered through the target category, which greatly reduces the interference to subsequent judgments of leftovers. At the same time, the background time uses YOLO to detect the background target, and a very clean residue detection background is obtained. Then classify the detected targets through the target category and the overlapping degree of the two coordinates, track and time the suspicious targets, and judge the background moving objects to obtain accurate leftovers. Applying YOLO target detection to remnant detection largely guarantees the accuracy and real-time performance of remnant detection. And it can also be well adapted to various public places and the interference caused by some complex environmental changes.

Description

technical field [0001] The invention relates to the fields of intelligent video monitoring, computer vision, machine learning and the like, and in particular to a method for detecting remnants based on YOLO target detection. Background technique [0002] With the popularization and widespread use of network surveillance cameras, remnant detection technology has become an important branch of intelligent video surveillance in the field of security protection. It is an interdisciplinary technology spanning image processing, pattern recognition, machine learning and other disciplines. The detection of remnants is widely used in the field of safety protection and is closely related to our life. For example: In banks, military bases, airports, subways, railway stations, shopping malls and other places, monitor and report to the police in time for leftover items that appear. [0003] Although there have been many studies on carryover detection, there are still some unresolved prob...

AI Technical Summary

Problems solved by technology

[0003] Although there have been many studies on carryover detection, there are still some unresolved problems

At present, most of the relic detection is based on the improvement of the double background model and the mixed Gaussian model. In complex scenes, it is easy to be disturbed by the movement of non-object targets (pedestrians, animals, etc.), and the influence of other disturbances cannot be completely eliminated.

There are some problems such as the background model is not clean enough, the misjudgment of the leftover objects, the high false detection rate of the leftover objects, etc. At the same time, the calculation complexity that needs to be run is large, and it is difficult to meet the real-time processing requirements of the intelligent video surveillance system

Images