A system and method for unattended monitoring

Abstract

The application provides a system and method for unmanned monitoring, and belongs to the technical field of monitoring.The application comprises a collection module, a classification module, a plurality of prediction modules, a comprehensive module and a comparison module.In the application, environment images and monitoring real-time images of a region to be monitored are collected from multiple angles at different time periods, the plurality of prediction modules are trained respectively by using the environment images at different time periods from multiple angles, and a standard result is obtained by processing through the comprehensive module.The monitoring real-time images are input into the prediction modules to obtain a prediction result by processing through the comprehensive module.The prediction result and the standard result are compared to determine whether an abnormality occurs in the region to be monitored.The application is trained by a specific region video, is suitable for multiple scenes such as contact prediction in a dangerous place, protection of an unmanned guard place, specific scene prediction, face recognition clock-in, face recognition lock control and the like, and effectively solves the problems of low accuracy and the like in current manual video recognition.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of monitoring, and particularly relates to a system and method for unmanned monitoring. BACKGROUND

[0002] The existing public field is usually equipped with a camera to collect real-time images of the monitored area, and a security personnel manually monitors the real-time images of the monitored area to determine whether an anomaly occurs in the monitored area. This traditional video monitoring method requires a dedicated personnel to identify the video content, which consumes a large amount of manpower. Meanwhile, a single personnel may make identification errors when identifying videos for a long time due to fatigue and other reasons. SUMMARY

[0003] In order to overcome the deficiencies of the prior art, the present application provides a system and method for unmanned monitoring.

[0004] In order to achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0005] A system for unmanned monitoring comprises:

[0006] A collection module is configured to collect environmental images and monitoring real-time images of a monitored area from multiple angles at different time periods.

[0007] A classification module is in communication connection with the collection module and is configured to divide the collected environmental images and monitoring real-time images into multiple groups of processing images according to angles.

[0008] A plurality of prediction modules are in communication connection with the first classification module, respectively. The plurality of prediction modules are configured to respectively predict and judge the monitoring real-time images in the groups of processing images to obtain a plurality of first prediction results, and are configured to respectively predict and judge the environmental images in the groups of processing images to obtain a plurality of first standard results.

[0009] A comprehensive module is in communication connection with the plurality of prediction modules and is configured to combine the plurality of first prediction results to obtain a plurality of combined prediction results, and is configured to combine the plurality of first standard results to obtain a plurality of final standard results.

[0010] A comparison module is in communication connection with the comprehensive module and is configured to compare the combined prediction results with the final standard results to determine whether an anomaly occurs in the monitored area.

[0011] Preferably, each prediction module is a convolutional neural network.

[0012] Preferably, the convolutional neural network comprises:

[0013] A convolutional layer is in communication connection with the classification module.

[0014] a pooling layer, which is in signal communication with the convolution layer;

[0015] a full connection layer, which is in signal communication with the pooling layer.

[0016] Preferably, the application further comprises an alarm module, wherein the alarm module comprises:

[0017] a voice alarm module, which is in signal communication with the comparison module;

[0018] a light alarm module, which is in signal communication with the comparison module.

[0019] A monitoring method for unmanned monitoring, comprising the following steps:

[0020] collecting environmental images and monitoring real-time images of a region to be monitored from multiple angles at different time periods;

[0021] dividing the collected environmental images and monitoring real-time images into multiple groups of processing images according to angles;

[0022] training multiple prediction modules respectively using multiple environmental images in the multiple groups of processing images to obtain multiple groups of first feature maps, and extracting multiple groups of first standard results from the multiple groups of first feature maps;

[0023] combining each group of first standard results to obtain a final standard result;

[0024] inputting monitoring real-time images in the multiple groups of processing images into the prediction modules according to groups to obtain multiple groups of second feature maps, and extracting multiple first prediction results from the multiple groups of first feature maps;

[0025] combining the multiple first prediction results to obtain a combined prediction result;

[0026] performing a comparison operation on the combined prediction result and the final standard result, and determining whether an abnormality occurs in the region to be monitored according to a comparison operation result.

[0027] Preferably, the application further comprises the following steps:

[0028] until a value of the final standard result obtained by the current operation tends to be stable; otherwise, training multiple prediction modules respectively using multiple environmental images in the multiple groups of processing images to obtain multiple groups of first feature maps, and extracting multiple groups of first standard results from the multiple groups of first feature maps, and combining each group of first standard results to obtain a final standard result.

[0029] Preferably, the step of determining whether the value of the final standard result tends to be stable comprises:

[0030] calculating a first standard result s j obtained by the jth training according to the following formula: j-1Previous difference d j ,

[0031] d j = s j -s j-1

[0032] If the difference between the first standard result s j obtained by the jth training and the first standard result s j-1 obtained by the (j-1)th training is less than a threshold value, it is determined that the numerical value of the final standard result tends to be stable, otherwise it is determined that the numerical value of the final standard result is unstable; the threshold value is calculated by the following formula:

[0033]

[0034] In the formula, M is the number of training times of the plurality of prediction modules.

[0035] Preferably, the final standard result is obtained by combining each group of first standard results by the following formula,

[0036]

[0037] In the formula, x 标准 is the final standard result, x1, x2, …, x N are a plurality of first standard results, N is the number of groups of processed images, a1, a2, …, a N are weight values corresponding to each of the first standard results.

[0038] The system and method for unmanned monitoring provided by the present application have the following beneficial effects: the present application comprises a collection module, a classification module, a plurality of prediction modules, a comprehensive module and a comparison module. In the present application, environmental images and monitoring real-time images of a region to be monitored are collected from multiple angles at different time periods, a plurality of prediction modules are trained respectively by using the environmental images from multiple angles at different time periods, and a standard result is obtained by processing through the comprehensive module. The monitoring real-time images are input into the prediction modules to obtain prediction results by processing through the comprehensive module. The prediction results and the standard result are compared to determine whether an abnormality occurs in the region to be monitored. The present application is trained for specific region videos, and is suitable for multiple scenes such as contact prediction in dangerous places, protection of unmanned guard places, specific scene prediction, face recognition clock-in, face recognition lock control, etc., and effectively solves the problem of low accuracy of current manual video recognition. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application and the design scheme thereof, the drawings required by the present embodiments will be briefly introduced as follows. The drawings in the following description are only part of the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0040] Figure 1 Structure diagram of the system for unattended monitoring of embodiment 1 of the present application;

[0041] Figure 2 Structure diagram of the monitoring method for unattended monitoring of embodiment 1 of the present application. DETAILED DESCRIPTION

[0042] In order to better understand the technical solutions of the present application and to enable one skilled in the art to carry out the present application, the present application will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present application, and cannot be used to limit the protection scope of the present application.

[0043] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "axial", "radial", "circumferential" and the like indicate the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the technical solutions of the present application and simplify the description, and are not intended to indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0044] In addition, the terms "first", "second", etc. are only used for description purposes and cannot be understood as indicating or implying relative importance. In the description of the present application, it should be noted that unless otherwise explicitly specified or limited, the terms "connected", "connected" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances. In the description of the present application, unless otherwise specified, the meaning of "multiple" is two or more, which will not be described in detail here.

[0045] Embodiment 1

[0046] Reference Figure 1The system for unmanned monitoring comprises a collection module, a classification module, a plurality of prediction modules, a comprehensive module and a comparison module. The collection module is configured to collect environment images and monitoring real-time images of a region to be monitored from multiple angles at different time periods. The classification module is in communication connection with the collection module and is configured to divide the collected environment images and monitoring real-time images into a plurality of groups of processing images according to angles. The plurality of prediction modules are in communication connection with the first classification module respectively, and the plurality of prediction modules are configured to respectively perform prediction judgment on the monitoring real-time images in the groups of processing images to obtain a plurality of first prediction results, and to respectively perform prediction judgment on the environment images in the groups of processing images to obtain a plurality of first standard results. The comprehensive module is in communication connection with the plurality of prediction modules and is configured to combine the plurality of first prediction results to obtain a plurality of combined prediction results, and to combine the plurality of first standard results to obtain a plurality of final standard results. The comparison module is in communication connection with the comprehensive module and is configured to compare the combined prediction results with the final standard results to determine whether an abnormality occurs in the region to be monitored.

[0047] In the embodiment, each prediction module is a convolutional neural network. The convolutional neural network comprises a convolutional layer, a pooling layer and a fully connected layer. The convolutional layer is in communication connection with the classification module; the pooling layer is in communication connection with the convolutional layer; and the fully connected layer is in communication connection with the pooling layer.

[0048] In order to facilitate the transmission of monitoring results and timely take measures, the system for unmanned monitoring further comprises an alarm module, which comprises a voice alarm module and a light alarm module. The voice alarm module is in communication connection with the comparison module. The light alarm module is in communication connection with the comparison module.

[0049] Reference is made to Figure 2 A monitoring method for unmanned monitoring comprises the following steps: collecting environment images and monitoring real-time images of a region to be monitored from multiple angles at different time periods; dividing the collected environment images and monitoring real-time images into a plurality of groups of processing images according to angles; until the value of a final standard result obtained by current operation tends to be stable; otherwise, training a plurality of prediction modules respectively by using a plurality of environment images in the plurality of groups of processing images to obtain a plurality of first feature maps, extracting a plurality of first standard results from the plurality of first feature maps respectively, and combining each first standard result to obtain a final standard result; inputting monitoring real-time images in the plurality of groups of processing images into the prediction modules according to groups to obtain a plurality of second feature maps, extracting a plurality of first prediction results from the plurality of first feature maps respectively, combining the plurality of first prediction results to obtain a combined prediction result; and comparing the combined prediction result with the final standard result, and determining whether an abnormality occurs in the region to be monitored according to a comparison result.

[0050] In the embodiment, the step of determining that the value of the final standard result tends to be stable comprises: calculating a first standard result s jand the first standard result s obtained from the j-1th training j-1 The previous difference d j ,

[0051] d j =s j -s j-1

[0052] If the first standard result s is obtained in the jth training j and the first standard result s obtained from the j-1th training j-1 If the difference between the two values ​​is less than the threshold, the final standard result value is judged to be stable, otherwise the final standard result value is judged to be unstable. The threshold value is calculated using the following formula:

[0053]

[0054] Where M is the number of training times for multiple prediction modules.

[0055] In this embodiment, the final standard result is obtained by combining each set of first standard results using the following formula:

[0056]

[0057] Where x 标准 is the final standard result, x1,x2,…,x N are multiple first standard results, N is the number of processed image groups, a, a2,…, a N are the weights corresponding to each first standard result.

[0058] The above embodiments are only preferred specific implementation methods of the present invention, and the protection scope of the present invention is not limited thereto. Any simple changes or equivalent replacements of the technical solutions that can be obviously obtained by any technician familiar with the field within the technical scope disclosed in the present invention fall within the protection scope of the present invention.

Claims

1. A system for unmanned monitoring, characterized in that: include: The acquisition module is used to collect environmental images and real-time monitoring images of the monitored area from multiple angles at different time periods; A classification module, in communication with the acquisition module, for dividing the collected environmental images and monitoring real-time images into multiple groups of processed images according to angles; Multiple prediction modules are respectively connected to the first classification module for communication, and the multiple prediction modules respectively predict and judge the real-time monitoring images in the group of processed images to obtain multiple initial prediction results. The multiple prediction modules are respectively trained using multiple environmental images in the multiple groups of processed images to obtain multiple groups of first feature maps, and multiple groups of first standard results are respectively extracted from the multiple groups of first feature maps. a synthesis module, communicatively connected to the plurality of prediction modules, for combining the plurality of initial prediction results to obtain a combined prediction result, and for combining the plurality of first standard results to obtain a final standard result; The comparison module is in communication with the comprehensive module and is used to compare the combined prediction result with the final standard result to determine whether an abnormality occurs in the monitored area.

2. The system for unmanned monitoring according to claim 1, characterized in that: Each of the prediction modules is a convolutional neural network.

3. The system for unmanned monitoring according to claim 2, characterized in that: The convolutional neural network includes: A convolutional layer, communicating with the classification module; A pooling layer, communicating with the convolutional layer; A fully connected layer is connected to the pooling layer.

4. The system for unmanned monitoring according to claim 1, characterized in that: Also included is an alarm module, the alarm module including: a voice alarm module, communicatively connected to the comparison module; The light alarm module is communicatively connected with the comparison module.

5. A monitoring method for unmanned monitoring, characterized in that: The following steps are involved: Collect environmental images and real-time monitoring images of the monitored area from multiple angles at different time periods; Divide the collected environmental images and real-time monitoring images into multiple groups of processed images according to angles; Using multiple environmental images in the multiple sets of processed images to train multiple prediction modules respectively, to obtain multiple sets of first feature maps, and extracting multiple sets of first standard results from the multiple sets of first feature maps respectively; Combining each group of first standard results to obtain the final standard result; The monitoring real-time images in the multiple groups of processed images are input into the prediction module according to the groups to obtain multiple groups of second feature maps, and multiple first prediction results are extracted from the multiple groups of first feature maps respectively; Combining multiple initial prediction results to obtain a combined prediction result; The combined prediction results are compared with the final standard results, and whether an abnormality occurs in the monitored area is determined based on the comparison results.

6. The system for unmanned monitoring according to claim 5, characterized in that: The following steps are also included: Until the value of the final standard result obtained by the current operation tends to be stable; otherwise, multiple prediction modules are trained separately using multiple environmental images in multiple groups of processed images to obtain multiple groups of first feature maps, and multiple groups of first standard results are extracted from the multiple groups of first feature maps, and each group of first standard results is combined to obtain the final standard result.

7. The system for unmanned monitoring according to claim 6, characterized in that: The steps to determine whether the value of the final standard result is stable include: The first standard result is obtained by calculating the j-th training according to the following formula The first standard result obtained from the j-1th training The difference before , If the first standard result is obtained in the jth training The first standard result obtained from the j-1th training If the difference between them is less than the threshold, the final standard result is judged to be stable, otherwise the final standard result is judged to be unstable. The threshold is calculated using the following formula: Where, is the number of training times for multiple prediction modules.

8. The system for unmanned monitoring according to claim 5, characterized in that: The final standard result is obtained by combining each group of first standard results using the following formula: Where, The final standard result is are multiple first standard results, is the number of image groups to be processed, are the weights corresponding to each of the first standard results.

Images