CN117975405A

CN117975405A - Inspection track monitoring method and device based on deep learning

Info

Publication number: CN117975405A
Application number: CN202311659587.2A
Authority: CN
Inventors: 王宁; 徐旺; 陈蕾; 张宪法; 张耀方; 卢丹; 卜晓燕; 陆凤岭
Original assignee: Shangfei Intelligent Technology Co ltd
Current assignee: Shangfei Intelligent Technology Co ltd
Priority date: 2023-12-05
Filing date: 2023-12-05
Publication date: 2024-05-03

Abstract

The invention provides a method and a device for monitoring a patrol track based on deep learning, belonging to the technical field of computer vision, wherein the method comprises the following steps: acquiring at least one monitoring image corresponding to each inspection point in a target area; detecting each monitoring image based on a target detection network, and acquiring at least one identification image comprising personnel information in all the monitoring images; the target detection network is obtained by training a patrol scene sample set based on a deep learning algorithm; determining a target image comprising at least one target person based on the identification algorithm and each identification image; determining a patrol track corresponding to the target personnel based on each target image; the inspection track is used for representing the sequence of the target personnel appearing at each inspection point; the inspection track is compared with the standard inspection track, and when the inspection track is inconsistent with the standard inspection track, the first warning information is sent out.

Description

Inspection track monitoring method and device based on deep learning

Technical Field

The invention relates to the technical field of computer vision, in particular to a method and a device for monitoring a patrol track based on deep learning.

Background

Patrol inspection is an important task, for example, in boiler houses, high-voltage substations and other scenes, if the patrol inspection is not performed at fixed points on time, serious injury can be caused to personnel and equipment. The fixed-point patrol inspection in a specific scene is one of indispensable security measures, and the security of personnel and equipment can be effectively protected.

In the prior art, whether the patrol personnel appear at the corresponding patrol points is monitored through the control cameras, but the real patrol track cannot be obtained, and whether the patrol personnel patrol according to the specified patrol track cannot be confirmed.

How to reduce the occurrence rate of safety accidents caused by inspection missing and out-of-place is a technical problem which needs to be solved by the technicians in the field.

Disclosure of Invention

Aiming at the problems existing in the prior art, the invention provides a method and a device for monitoring a patrol track based on deep learning.

The invention provides a patrol track monitoring method based on deep learning, which comprises the following steps:

acquiring at least one monitoring image corresponding to each inspection point in a target area; the target area comprises at least one inspection point;

detecting each monitoring image based on a target detection network, and acquiring at least one identification image comprising personnel information in all the monitoring images; the target detection network is obtained by training a patrol scene sample set based on a deep learning algorithm;

Determining a target image comprising at least one target person based on the identification algorithm and each of the identification images;

Determining a patrol track corresponding to the target person based on each target image; the inspection track is used for representing the sequence of the target personnel appearing at each inspection point;

and comparing the inspection track with a standard inspection track, and sending out first alarm information when the inspection track is inconsistent with the standard inspection track.

According to the inspection track monitoring method based on deep learning provided by the invention, the determination of the target image comprising at least one target person based on the identification algorithm and each identification image comprises the following steps:

Identifying each identification image based on a pedestrian re-identification algorithm to obtain first characteristic information corresponding to each identification image;

And determining at least one target image containing target personnel based on the similarity of the first characteristic information and the second characteristic information.

According to the inspection track monitoring method based on deep learning provided by the invention, the inspection track corresponding to the target personnel is determined based on each target image, and the inspection track monitoring method comprises the following steps:

Acquiring a time frame of each target image in each target personnel set; sequencing the target images based on the sequence of the time frames to obtain a first sequence;

and determining the inspection track corresponding to the target personnel based on the inspection points corresponding to the target images in the first sequence.

According to the inspection track monitoring method based on deep learning, provided by the invention, the method further comprises the following steps:

Detecting each monitoring image based on the target detection network, and executing timing operation when the personnel information is not included in each monitoring image to obtain timing duration;

and when the timing time is longer than a preset threshold value, sending out second alarm information.

Acquiring picture information of the target personnel at each inspection point in each target image; performing feature analysis on each piece of picture information to obtain picture features corresponding to each piece of picture information; the visual features include at least one of a pose, a position, and a contour of the target person;

based on the picture characteristics, judging whether the inspection action of the target personnel at each inspection point is standard, and if not, sending out third alarm information.

According to the inspection track monitoring method based on deep learning provided by the invention, after at least one monitoring image in the target area is obtained, the method further comprises the following steps:

preprocessing each monitoring image; the preprocessing includes at least one of image denoising, image enhancement, and color correction.

The invention also provides a device for monitoring the inspection track based on deep learning, which comprises the following steps:

the acquisition module is used for acquiring at least one monitoring image corresponding to the inspection points in the target area respectively; the target area comprises at least one inspection point;

The detection module is used for detecting each monitoring image based on a target detection network and acquiring at least one identification image comprising personnel information in all the monitoring images; the target detection network is obtained by training a patrol scene sample set based on a deep learning algorithm;

A first determining module, configured to determine, based on an identification algorithm and each of the identification images, a target image including at least one target person;

The second determining module is used for determining a patrol track corresponding to the target personnel based on each target image; the inspection track is used for representing the sequence of the target personnel appearing at each inspection point;

And the comparison module is used for comparing the inspection track with the standard inspection track, and sending out first alarm information when the inspection track is inconsistent with the standard inspection track.

The invention also provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the deep learning-based inspection track monitoring method when executing the program.

The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a deep learning based inspection trace monitoring method as described in any of the above.

The invention also provides a computer program product comprising a computer program which, when executed by a processor, implements the deep learning-based inspection track monitoring method as described in any one of the above.

The method comprises the steps of obtaining a plurality of monitoring images corresponding to all inspection points of a target area respectively, detecting the monitoring images through a target detection network obtained by training an inspection scene sample set in advance based on a deep learning algorithm, obtaining a plurality of identification images containing personnel information in the monitoring images, further, identifying each identification image through an identification algorithm, determining at least one target image containing target personnel, determining the inspection track of the target personnel based on the inspection points corresponding to each target image, comparing the inspection track with a standard inspection track, and sending alarm information under the condition that the inspection track and the standard inspection track are inconsistent. According to the inspection track monitoring method based on deep learning, each inspection point is monitored in real time, the inspection track is obtained by identifying and processing the monitoring image through the target detection network and the identification algorithm, and if the inspection track is wrong in the target area, an alarm is given, so that the occurrence rate of safety accidents caused by missing or out-of-place inspection is reduced, and the safety of field personnel and equipment is greatly ensured.

Drawings

In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic flow chart of a method for monitoring a patrol track based on deep learning;

FIG. 2 is a second flow chart of the inspection track monitoring method based on deep learning provided by the invention;

FIG. 3 is a third flow chart of the inspection track monitoring method based on deep learning according to the present invention;

FIG. 4 is a schematic flow chart of a method for monitoring a patrol track based on deep learning;

fig. 5 is a schematic structural diagram of the inspection track monitoring device based on deep learning;

Fig. 6 is a schematic structural diagram of an electronic device provided by the present invention.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The following describes a method and a device for monitoring a patrol track based on deep learning with reference to fig. 1 to 5.

Fig. 1 is one of flow diagrams of a deep learning-based inspection track monitoring method provided by the present invention, and as shown in fig. 1, the deep learning-based inspection track monitoring method provided by the embodiment of the present invention includes:

step 110, obtaining at least one monitoring image corresponding to each inspection point in a target area; the target area comprises at least one inspection point;

Specifically, the application scenario of the deep learning-based inspection track monitoring method of the embodiment of the invention is generally a complex and high-risk scenario, such as a boiler room, a high-voltage transformer substation and the like, in which the requirements for regular inspection are high, a fixed inspection route is generally preset, a plurality of inspection points are arranged in the scenario, and inspection results are qualified only by sequentially performing inspection according to standard inspection tracks for the plurality of inspection points in the scenario. Namely, aiming at the inspection track, corresponding inspection point fixed-point inspection and sequential inspection are required.

Specifically, the target area includes a plurality of inspection points, and in the embodiment of the present invention, the target area refers to an area including a complete inspection track. In the target area, including a plurality of inspection points that set up, set up at least one camera to every inspection point, through a plurality of cameras that every inspection point corresponds, the monitoring image of different angles to this inspection point that can gather. In the step, each camera can acquire images of corresponding inspection points in real time to obtain a plurality of monitoring images.

Step 120, detecting each monitoring image based on a target detection network, and acquiring at least one identification image including personnel information in all the monitoring images; the target detection network is obtained by training a patrol scene sample set based on a deep learning algorithm;

Specifically, in this step, each of the monitoring images is detected based on the target detection network, and the purpose of this step is to screen out an identification image including personnel information from among a plurality of monitoring images. Because the cameras with different angles corresponding to each monitoring point can acquire monitoring images in the cameras in real time, in the embodiment of the invention, when no personnel passes, the monitoring images acquired by the cameras are meaningless for the subsequent identification of the inspection track, so that images comprising personnel information are screened out from a plurality of monitoring images through the step, and the images comprising the personnel information are defined as identification images.

Furthermore, the target detection network in the embodiment of the invention is trained in advance, and is obtained by training the inspection scene sample set based on a deep learning algorithm. In the inspection scene sample set, a plurality of inspection scene images, for example, inspection scene images under different angles and different light conditions including personnel information are included, and training is performed on the inspection scene sample set based on a deep learning algorithm.

Specifically, in the embodiment of the present invention, the object detection network may be a YOLO-V8 detection network, the output of the YOLO-V8 object detection network is based on the whole picture, in the embodiment of the present invention, the input is a monitoring image, the output is an identification image including personnel information, further, the YOLO-V8 object detection network outputs the object information detected from the monitoring image at one time, in the embodiment of the present invention, the personnel information included in the object information monitoring image, the position of the personnel, and the like.

Step 130, determining a target image comprising at least one target person based on the identification algorithm and each of the identification images;

Specifically, based on the above step 120, identification images including person information are selected from among the plurality of monitoring images, and in this step, each identification image is identified based on an identification algorithm, from which a target image including a target person is selected. It should be noted that, the target person in the embodiment of the present invention does not refer to a certain person, but includes the target image of the same person in the plurality of identification images.

In a specific implementation, in the multiple identification images, the person 1 and the person 2 can appear, and in the multiple identification images, all the identification images of the person 1 appear are target images corresponding to the person 1, and all the identification images of the person 2 appear are target images corresponding to the person 2.

In the embodiment of the invention, a plurality of target images comprising target personnel can be obtained based on the recognition algorithm. The recognition algorithm used in the embodiment of the invention can be a pedestrian re-recognition algorithm, and a target image comprising the same person is selected from a plurality of identification images comprising personnel information and shot by different cameras through the pedestrian re-recognition algorithm, namely, the target person appears in a target area, and monitoring images of all inspection points are selected.

Step 140, determining a patrol track corresponding to the target person based on each target image; the inspection track is used for representing the sequence of the target personnel appearing at each inspection point;

Specifically, in this step, based on the set of target personnel acquired in the above step 140, the inspection track of the target personnel is determined, and the inspection track of the target personnel is determined through all the monitoring images of the target personnel in the target area, where the inspection track may represent the sequence of the target personnel passing through each inspection point.

And 150, comparing the inspection track with a standard inspection track, and sending out first alarm information when the inspection track is inconsistent with the standard inspection track.

Specifically, in this step, based on the inspection track obtained in the above step 140, the inspection track is compared with a pre-designated standard inspection track, if the inspection track of the target person is inconsistent with the standard inspection track, that is, the sequence of the target person appearing at each inspection point is wrong, an alarm message is sent to indicate that the current inspection track is wrong, and inspection needs to be performed again.

According to the inspection track monitoring method based on the deep learning, through obtaining a plurality of monitoring images corresponding to all inspection points of a target area respectively, then detecting the monitoring images through a target detection network obtained by training an inspection scene sample set based on a deep learning algorithm in advance, obtaining a plurality of identification images including personnel information in the monitoring images, further, identifying each identification image through a pedestrian re-identification algorithm, determining at least one target image including target personnel, then determining the inspection track of the target personnel based on the inspection points corresponding to each target image, comparing the inspection track with a standard inspection track, and sending alarm information under the condition that the inspection track and the standard inspection track are inconsistent. According to the inspection track monitoring method based on deep learning, each inspection point is monitored in real time through the camera, the monitoring image is identified and processed through the target detection network and the identification algorithm, the inspection track corresponding to the target personnel is obtained, if the inspection track is wrong in the target area, an alarm is given, the safety accident occurrence rate caused by the lack and the lack of in-place inspection is reduced, and the safety of the field personnel and the equipment is greatly ensured.

Optionally, according to the method for monitoring a routing track based on deep learning provided by the embodiment of the present invention, the step 130 is specifically implemented as follows, fig. 2 is a second schematic flow chart of the method for monitoring a routing track based on deep learning provided by the present invention, and as shown in fig. 2, the determining, based on an identification algorithm and each of the identification images, a target image including at least one target person includes:

step 210, identifying each identification image based on a pedestrian re-identification algorithm to obtain first characteristic information corresponding to each identification image;

Specifically, the pedestrian re-recognition algorithm is a technique for judging whether a specific pedestrian exists in an image by using a computer vision technique; in other words, pedestrian re-recognition refers to the recognition of a target pedestrian in an existing video sequence of possible sources and non-overlapping camera views. In the step, each identification image is identified based on a pedestrian re-identification algorithm, and the pedestrian re-identification algorithm can extract the characteristics of the personnel information based on the personnel information in each identification image to obtain first characteristic information corresponding to each identification image.

Step 220, determining at least one target image containing the target person based on the similarity of the first feature information and the second feature information.

Specifically, the similarity between the first feature information corresponding to each identification image and the second feature information corresponding to the target person is calculated, when the similarity is larger than a similarity threshold, the person information included in the current identification image is considered to be consistent with the target person information, and the identification image is determined to be the target image, namely the image including the target person.

According to the inspection track monitoring method based on deep learning, first characteristic information corresponding to each person in each identification image is obtained through a pedestrian re-identification algorithm, then the first characteristic information is compared with second characteristic information corresponding to the target person, the similarity between the first characteristic information and the second characteristic information is calculated, and when the similarity is larger than a preset threshold value, the person in the identification image is the target person, and the identification image is the target image. According to the inspection track monitoring method based on deep learning, through a pedestrian re-recognition algorithm, target images comprising the same person can be obtained from a plurality of identification images shot by different cameras, and a foundation is laid for subsequent determination of the inspection track.

Optionally, according to the method for monitoring a routing inspection track based on deep learning provided by the embodiment of the present invention, the step 140 specifically includes the following steps, fig. 3 is a third schematic flow chart of the method for monitoring a routing inspection track based on deep learning provided by the present invention, and as shown in fig. 3, determining, based on each target image, a routing inspection track corresponding to the target person specifically includes:

Step 310, obtaining a time frame of each target image;

Specifically, in this step, a time frame of each target image is acquired, the inspection points are monitored in real time by a plurality of cameras corresponding to each inspection point, a plurality of target images are obtained by performing subsequent screening based on the monitored images, each target image has a corresponding shooting time frame and is included in the target image information, and the time frame of the target image can be directly acquired.

Step 320, sorting the target images based on the sequence of the time frames to obtain a first sequence;

Specifically, in this step, based on the time frame of each target image, the plurality of target images in the target person set are ordered according to the sequence of the time frames, so as to obtain the first sequence.

Step 330, determining a patrol track corresponding to the target person based on the patrol points corresponding to each target image in the first sequence.

Specifically, in this step, based on the inspection points corresponding to each target image in the first sequence, the inspection track of the target personnel may be obtained, and the first sequence ordered according to the sequence of the time frames of the target images, where the inspection points corresponding to the first sequence are also in sequence, and the sequence of the target personnel passing through each inspection point may be represented, so as to determine the inspection track corresponding to the target personnel.

According to the inspection track monitoring method based on deep learning, the time frames of each target image in the target personnel set are acquired, the target images are sequenced based on the sequence of the time frames corresponding to the target images to obtain the first sequence, then the inspection track corresponding to the target personnel is determined based on the inspection points corresponding to each target image in the first sequence, the inspection track is determined through sequencing of the time frames, and the accuracy of inspection track confirmation is improved.

Optionally, according to the method for monitoring a routing inspection track based on deep learning provided by the embodiment of the present invention, the method further includes the following steps, fig. 4 is a schematic flow chart diagram of the method for monitoring a routing inspection track based on deep learning provided by the present invention, and as shown in fig. 4, the method for monitoring a routing inspection track based on deep learning according to the embodiment of the present invention further includes:

step 410, detecting each monitoring image based on the target detection network, and executing timing operation to obtain timing duration when each monitoring image does not include the personnel information;

Specifically, this step is another branch of step 120 described above. Detecting a plurality of monitoring images based on the target detection network, and executing the subsequent step of the step 120 when the monitoring images are detected to include personnel information. In another case, that is, when the monitoring image does not include personnel information, it is indicated that the current inspection point does not pass by an inspection personnel, in this case, a timer is started, a timing operation is performed, and when the timing operation is performed, the plurality of cameras corresponding to the inspection point still can shoot the monitoring image corresponding to the inspection point in real time, if the personnel information does not exist in the monitoring image, the timing is not interrupted, and the timing duration is obtained.

And 420, when the timing time is longer than a preset threshold value, sending out alarm information.

Specifically, in this step, when the timing duration is greater than the preset threshold, it is indicated that no personnel information appears in the monitoring images shot by the cameras corresponding to the monitoring point within a period of time, and this condition indicates that the patrol personnel does not pass the patrol point, needs to send alarm information, and characterizes the current patrol problem. In the specific implementation summary, the preset threshold may be set to 3 hours, and may specifically be adjusted according to the inspection scene, which is not specifically limited herein.

According to the deep learning-based inspection track monitoring method, when each monitoring image is detected based on the target detection network, if no personnel information exists in the monitoring image, in this case, timing operation is needed, if no personnel information exists in the monitoring image, timing is not interrupted, corresponding timing duration is obtained, if the timing duration exceeds a preset threshold value, it is indicated that inspection points corresponding to the monitoring image do not have the appearance of inspection personnel within a preset time, it is indicated that the current inspection work is wrong, alarm information is needed to be sent, and the accuracy of inspection work monitoring is improved.

Optionally, according to the method for monitoring a patrol track based on deep learning provided by the embodiment of the present invention, the method further includes:

acquiring picture information of the target personnel at each inspection point in each target image;

Specifically, in this step, based on the above-described target detection network, it is possible to acquire the screen information corresponding to the target person in addition to the person information that appears in the monitoring image.

Performing feature analysis on each piece of picture information to obtain picture features corresponding to each piece of picture information; the visual features include at least one of a pose, a position, and a contour of the target person;

Specifically, in this step, the screen information corresponding to the target person is analyzed to obtain the screen feature corresponding to the screen information, and the specific analysis method is not limited herein. The visual characteristics include the pose, position and contour of the target person.

It should be noted that, in the embodiment of the present invention, the first, second and third alarm information are all alarm information, but the alarm modes of the display interface at the front end are different, which is not limited herein.

Specifically, in this step, by comparing the picture feature with the preset standard feature, it can be determined whether the inspection action and state of the target person are normal, for example, some inspection points require inspection of the inspection person at a fixed position, if it is determined from the gesture that some inspection points need squatting inspection, then the gesture feature of the target person should be squatting gesture, if the picture feature is not matched with the standard feature, it is indicated that the inspection state of the target person at the inspection point is not qualified and does not meet the inspection standard, and in this case, alarm information needs to be sent.

According to the deep learning-based inspection track monitoring method, the image information of the target person at the inspection point is obtained, then the image information is subjected to characteristic analysis, so that the image characteristics of the target person can be obtained, whether the inspection action of the target person at the inspection point meets the specification or not can be judged, and if the inspection action of the target person at the inspection point does not meet the specification, alarm information is required to be sent out, so that the normalized management of the inspection action of the target person is realized.

Optionally, according to an embodiment of the present invention, after the at least one monitoring image in the target area is obtained, the method further includes:

Specifically, after a plurality of monitoring images in a target area are acquired, in order to facilitate subsequent processing of the monitoring images by a subsequent target detection network and a pedestrian re-recognition algorithm, the monitoring images need to be preprocessed, and as the monitoring images in the embodiment of the invention are shot by a camera in the target area, certain problems exist in definition and color, the preprocessing in the embodiment of the invention comprises at least one of image denoising, image enhancement and color correction, and preprocessing is performed based on specific conditions of the monitoring images.

The deep learning-based inspection track monitoring device provided by the invention is described below, and the deep learning-based inspection track monitoring device described below and the deep learning-based inspection track monitoring method described above can be correspondingly referred to each other.

Fig. 5 is a schematic structural diagram of a deep learning-based inspection track monitoring device provided by the invention, and as shown in fig. 5, the deep learning-based inspection track monitoring device according to an embodiment of the invention includes:

The acquiring module 510 is configured to acquire at least one monitoring image corresponding to each of the inspection points in the target area; the target area comprises at least one inspection point;

The detection module 520 is configured to detect each of the monitoring images based on a target detection network, and obtain at least one identification image including personnel information in all the monitoring images; the target detection network is obtained by training a patrol scene sample set based on a deep learning algorithm;

A first determining module 530, configured to determine, based on the identification algorithm and each of the identification images, a target image including at least one target person;

a second determining module 540 for

And the comparison module 550 is configured to compare the inspection track with a standard inspection track, and send out first alarm information when the inspection track is inconsistent with the standard inspection track.

According to the inspection track monitoring device based on deep learning, the effective monitoring of the inspection track of the target area is improved through the mutual matching of the modules. The method comprises the steps of obtaining a plurality of monitoring images corresponding to all inspection points of a target area respectively, detecting the monitoring images through a target detection network obtained by training an inspection scene sample set based on a deep learning algorithm in advance, obtaining a plurality of identification images containing personnel information in the monitoring images, further, identifying each identification image through a pedestrian re-identification algorithm, determining a target personnel set corresponding to the target image of the same target personnel, determining an inspection track of the target personnel based on the inspection points corresponding to each target image, comparing the inspection track with a standard inspection track, and sending alarm information under the condition that the inspection track and the standard inspection track are inconsistent. According to the inspection track monitoring method for deep learning, each inspection point is monitored in real time through the camera, the monitoring image is identified and processed through the target detection network and the identification algorithm, the inspection track corresponding to the target personnel is obtained, if the inspection track is wrong in the target area, an alarm is given, the safety accident occurrence rate caused by the lack and the lack of in-place inspection is reduced, and the safety of the field personnel and the equipment is greatly ensured.

Optionally, the first determining module is specifically configured to:

identifying each identification image based on a pedestrian re-identification algorithm to obtain first characteristic information corresponding to each identification image in each identification image;

determining the target personnel set based on the similarity of the first characteristic information and the second characteristic information; the second characteristic information is characteristic information corresponding to the target person.

Optionally, the second determining module is specifically configured to:

acquiring a time frame of each target image in the target personnel set;

sequencing the target images based on the sequence of the time frames to obtain a first sequence;

Fig. 6 is a schematic structural diagram of an electronic device provided by the present invention. Fig. 6 illustrates a physical schematic diagram of an electronic device, as shown in fig. 6, which may include: processor 610, communication interface (Communications Interface) 620, memory 630, and communication bus 640, wherein processor 610, communication interface 620, memory 630 communicate with each other via communication bus 640. The processor 610 may invoke logic instructions in the memory 630 to perform the deep learning based patrol trajectory monitoring method described above, the method comprising:

Further, the logic instructions in the memory 630 may be implemented in the form of software functional units and stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

In another aspect, the present invention also provides a computer program product, where the computer program product includes a computer program, where the computer program can be stored on a non-transitory computer readable storage medium, and when the computer program is executed by a processor, the computer can execute the method for monitoring a patrol track based on deep learning provided by the above methods, and the method includes:

In yet another aspect, the present invention further provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method for deep learning based inspection trace monitoring provided by the above methods, the method comprising:

The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. The inspection track monitoring method based on deep learning is characterized by comprising the following steps of:

2. The deep learning based patrol trajectory monitoring method of claim 1, wherein the recognition algorithm comprises a pedestrian re-recognition algorithm, wherein the determining, based on the recognition algorithm and each of the identification images, a target image comprising at least one target person comprises:

3. The method for monitoring a patrol trajectory based on deep learning according to claim 2, wherein determining a patrol trajectory corresponding to the target person based on each of the target images comprises:

acquiring a time frame of each target image;

4. The deep learning based patrol trajectory monitoring method of claim 1, further comprising:

5. The deep learning based patrol trajectory monitoring method of claim 1, further comprising:

performing feature analysis on each piece of picture information to obtain picture features corresponding to each piece of picture information; the visual features include at least one of a pose, a position, and a profile of the target person;

6. The method for monitoring a patrol trajectory based on deep learning according to claim 1, wherein after the at least one monitoring image corresponding to at least one patrol point of the target area is obtained, the method further comprises:

7. Inspection track monitoring device based on deep learning, characterized by comprising:

8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the deep learning based patrol trajectory monitoring method of any one of claims 1 to 6 when the program is executed by the processor.

9. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the deep learning based patrol trajectory monitoring method according to any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements a deep learning based patrol trajectory monitoring method according to any one of claims 1 to 6.