CN111582006A - Video analysis method and device - Google Patents

Abstract

The embodiment of the invention provides a video analysis method and a video analysis device, wherein the method comprises the following steps: detecting a monitoring target in the collected video stream; intercepting a video image containing a monitoring target; identifying the intercepted video image to obtain the classification information of each monitoring target; therefore, in the scheme, the whole video stream is not classified and identified, but the video image containing the monitoring target is intercepted, and only the intercepted video image is classified and identified, so that the calculation amount is reduced.

Description

A video analysis method and device

technical field

The present invention relates to the technical field of monitoring, and in particular, to a video analysis method and device.

Background technique

In the related scheme, monitoring equipment is usually set in the area that needs to be monitored, the monitoring equipment collects the video stream, and analyzes the video stream to determine whether there are people or vehicles illegally intruding into the area. In this solution, the overall analysis of the video stream requires a large amount of computation.

SUMMARY OF THE INVENTION

The purpose of the embodiments of the present invention is to provide a video analysis method and apparatus to reduce the amount of calculation.

To achieve the above purpose, an embodiment of the present invention provides a video analysis method, including:

Detect surveillance targets in the captured video stream;

intercepting a video image containing the monitoring target;

By identifying the intercepted video images, the classification information of each monitoring target is obtained.

Optionally, the detecting a monitoring target in the collected video stream includes: detecting a moving target in the collected video stream;

The intercepting the video image containing the monitoring target, including:

One or more frames of video images containing the moving object are captured.

Optionally, the classification information of each monitoring target is obtained by identifying the intercepted video images, including:

Input the intercepted video image into the first neural network model obtained by pre-training, use the first neural network model to classify the moving objects in the video image, and obtain each output of the first neural network model. classification information of a moving target.

Optionally, the detecting the monitoring target in the collected video stream includes: performing face recognition in the collected video stream to obtain a recognition result;

The intercepting the video image containing the monitoring target, including:

According to the recognition result, the face area is intercepted in the image containing the face;

Described by identifying the intercepted video images, the classification information of each monitoring target is obtained, including:

By matching the intercepted face region with the face data stored in the face database, the classification information of the face region is obtained.

Optionally, by matching the intercepted face region with the face data stored in the face database, the classification information of the face region is obtained, including:

Inputting the intercepted face region into the second neural network model obtained by pre-training, and using the second neural network model to convert the face region into modeling data;

By matching the modeling data with the face data stored in the face database, the classification information of the face region is obtained, and the classification information includes: existence or non-existence in the face database and the construction The face data for which the model data is successfully matched.

Optionally, after the classification information of each monitoring target is obtained by identifying the intercepted video images, the method further includes:

judging whether the classified information meets the preset alarm condition;

If it matches, output an alarm message.

Optionally, after obtaining the classification information of each moving target output by the first neural network model, the method further includes:

Determine whether the classification information meets the preset alarm conditions; if so, output the alarm information;

The preset alarm conditions include:

The classification information of the moving object is a person; or, the classification information of the moving object is a vehicle.

Optionally, after obtaining the classification information of the face region, the method further includes:

Determine whether the classification information meets the preset alarm conditions; if so, output the alarm information;

The preset alarm conditions include:

The classification information of the face region is: the existence or non-existence of face data that is successfully matched with the modeling data in the face database.

To achieve the above purpose, an embodiment of the present invention also provides a video analysis device, including:

The detection module is used to detect the monitoring target in the collected video stream;

An interception module for intercepting a video image containing the monitoring target;

The classification module is used to obtain classification information of each monitoring target by identifying the intercepted video images.

Optionally, the detection module is specifically used for: detecting moving objects in the collected video stream;

The intercepting module is specifically configured to: intercept one or more frames of video images including the moving object.

Optionally, the classification module is specifically used for:

Input the intercepted video image into the first neural network model obtained by pre-training, use the first neural network model to classify the moving objects in the video image, and obtain each output of the first neural network model. classification information of a moving target.

Optionally, the detection module is specifically used for: performing face recognition in the collected video stream to obtain a recognition result;

The intercepting module is specifically used for: intercepting the face region in the image containing the human face according to the recognition result;

The classification module is specifically configured to obtain classification information of the face region by matching the intercepted face region with the face data stored in the face database.

Optionally, the classification module is specifically used for:

Inputting the intercepted face region into the second neural network model obtained by pre-training, and using the second neural network model to convert the face region into modeling data;

By matching the modeling data with the face data stored in the face database, the classification information of the face region is obtained, and the classification information includes: existence or non-existence in the face database and the construction The face data for which the model data is successfully matched.

Optionally, the device further includes:

a first judging module for judging whether the classified information meets the preset alarm condition; if so, triggering the first alarm module;

The first alarm module is used for outputting alarm information.

Optionally, the device further includes:

The second judgment module is used for judging whether the classification information complies with the preset alarm conditions; the preset alarm conditions include: the classification information of the moving object is a person; or, the classification information of the moving object is a vehicle; if In line with, trigger the second alarm module;

The second alarm module is used for outputting alarm information.

Optionally, the device further includes:

The third judging module is used to judge whether the classification information meets the preset alarm conditions; the preset alarm conditions include: the classification information of the face area is: existence or non-existence in the face database and the The face data whose modeling data matches successfully; if it matches, the third alarm module is triggered;

The third alarm module is used for outputting alarm information.

In the embodiment of the present invention, the monitoring target is detected in the collected video stream; the video image containing the monitoring target is intercepted; the classification information of each monitoring target is obtained by identifying the intercepted video image; it can be seen that in this scheme, and Instead of classifying and identifying the entire video stream, the video images containing the monitoring target are intercepted, and only the intercepted video images are classified and identified, thus reducing the amount of calculation.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

1 is a first schematic flowchart of a video analysis method provided by an embodiment of the present invention;

2 is a second schematic flowchart of a video analysis method provided by an embodiment of the present invention;

3 is a schematic diagram of interaction between a monitoring point and an NVR according to an embodiment of the present invention;

4 is a third schematic flowchart of a video analysis method provided by an embodiment of the present invention;

FIG. 5 is a schematic diagram of another interaction between a monitoring point and an NVR according to an embodiment of the present invention;

6 is a schematic structural diagram of a video analysis apparatus according to an embodiment of the present invention;

7 is a schematic structural diagram of an electronic device according to an embodiment of the present invention;

FIG. 8 is a schematic structural diagram of a video analysis system according to an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In order to solve the above technical problems, embodiments of the present invention provide a video analysis method and device, which can be applied to cameras, such as IPC (IP Camera, network camera), or can be applied to NVR (Network Video Recorder, network hard disk) video recorder), or can be applied to other electronic devices, or can be applied to a video analysis system, which is not specifically limited. The video analysis method provided by the embodiment of the present invention is first introduced in detail below.

1 is a first schematic flowchart of a video analysis method provided by an embodiment of the present invention, including:

S101: Detect a monitoring target in the collected video stream.

For example, in one embodiment, the monitoring target may be a moving target; in this case, S101 may include: detecting the moving target in the captured video stream. For example, an algorithm such as a frame difference method, a background subtraction algorithm, or an optical flow method can be used to detect moving objects in the video stream.

In another implementation manner, the monitoring target may be a face; in this case, S101 may include: performing face recognition in the collected video stream to obtain a recognition result. For example, face recognition algorithms can be used to identify faces in video streams.

S102: Capture a video image including the monitoring target.

In the above-mentioned embodiment, the monitoring target is a moving target. In this case, S102 may include: capturing one or more frames of video images including the moving target. For example, the multi-frame video image may be a small video, such as a small video several seconds before and after the key frame.

In the above another embodiment, the monitoring target is a human face. In this case, S102 may include: according to the above recognition result, intercepting a human face region from an image containing a human face. Alternatively, one or more frames of video images including the face region may also be intercepted from the video stream according to the recognition result.

S103: Obtain classification information of each monitoring target by identifying the intercepted video images.

In the above-mentioned one embodiment, the monitoring target is a moving target. In this case, S103 may include: inputting the intercepted video image into the first neural network model obtained by pre-training, and using the first neural network model to The moving objects in the video image are classified to obtain classification information of each moving object output by the first neural network model.

For example, the moving target may be a person, a vehicle, or the like. The first neural network model is a model for classifying moving objects. The process of training to obtain the first neural network model may include: acquiring a sample image to be trained, and the sample image may include moving objects such as people or vehicles; adding labels to various moving objects in the sample images, the label is The type of moving target, such as vehicles, people, etc.; input the sample image into the neural network of the preset structure, and use the label as the supervision to iteratively adjust the neural network. When the conditions for the end of the iteration are met, the training is obtained. The completed first neural network model.

The video image intercepted in S102 is input into the first neural network model, and the first neural network model can output the classification information of each moving target in the video image, and the classification information is also the type of moving target, such as vehicles, personnel, etc. Wait.

For example, some scenarios have a high level of security, and perimeter protection needs to be carried out for these scenarios, that is, to determine whether people or vehicles enter the scenario. Applying this embodiment, on the one hand, the classification information of each moving target is obtained. If the classification information is a person or a vehicle, the relevant personnel can be reminded to carry out follow-up processing in time, so as to achieve effective perimeter prevention; For moving target detection, the moving target detection algorithm can be understood as a coarse detection algorithm with a small amount of calculation, and then intercepts a small part of the video images in the video stream, and only performs fine identification on this small part of the video images, that is, using

The first neural network model identifies the classification information of the moving target. Compared with the overall analysis of the video stream, this solution reduces the amount of computation.

In the above-mentioned another embodiment, the monitoring target is a human face. In this case, S103 may include: by matching the intercepted human face region with the human face data stored in the human face database, obtain the face region of the human face. Classified information.

For example, in some scenarios, only authorized personnel are allowed to enter, and a stranger (unauthorized personnel) identification scheme needs to be implemented for these scenarios. In this case, this embodiment can be used. For example, the face data of the authorized person may be stored in the face database, and the face area intercepted in S102 is matched with the face database, that is, it is determined whether the person in the video stream is an authorized person. The classification information of the face area may be: the presence or absence of face data matching the face area in the face database; or, the classification information of the face area may also be: authorized personnel or unauthorized personnel (strangers). ).

As another example, in some scenarios, it is necessary to identify a designated person, such as an attendance scenario, or a VIP (very important person, important person) identification scenario, and this embodiment can also be used in these scenarios. For example, the face data of the designated person may be stored in the face database, and the face area intercepted in S102 is matched with the face database, that is, it is determined whether the person in the video stream is the designated person. The classification information of the face area may be: existence or non-existence of face data successfully matched with the face area in the face database; or, the classification information of the face area may also be: a designated person or a non-designated person.

In one case, S103 may include: inputting the intercepted face region into a second neural network model obtained by pre-training, and using the second neural network model to convert the face region into modeling data; Matching the modeling data with the face data stored in the face database to obtain classification information of the face area, the classification information includes: the presence or absence in the face database and the modeling The data matches the face data that is successful.

The second neural network model may be a face modeling model, and the second neural network model may convert the face image into modeling data, that is, a kind of structural data. In this case, what is also stored in the face database is the modeling data (structure data) transformed by the second neural network model. Match the converted modeling data of the face region intercepted in S102 with the modeling data in the face database. If the matching is successful, it means that the person corresponding to the face region is an authorized person or a designated person. If the matching is unsuccessful , it means that the person corresponding to the face area is an unauthorized person (stranger) or a non-designated person.

Applying this embodiment, on the one hand, the classification information of the face area can be obtained, according to the classification information, it can be judged whether the person is an authorized person or a designated person, and according to the judgment result, the relevant personnel can be reminded to carry out follow-up processing in time, which can achieve effective strangers. On the other hand, first perform face recognition on the video stream, the face recognition algorithm can be understood as a rough detection algorithm with a small amount of calculation, and then intercept a small part of the video image in the video stream ( (or image area), only the intercepted part is finely identified, that is, face matching is performed. Compared with the overall analysis of the video stream, this solution reduces the amount of calculation.

As an implementation manner, after S103, the method may further include: judging whether the classification information meets the preset alarm condition; if so, outputting alarm information.

In the above embodiment, the monitoring target is a moving target. In this case, the preset alarm condition may include: the classification information of the moving target is a person; or, the classification information of the moving target is a vehicle.

As described above, if perimeter prevention needs to be performed, that is, to determine whether a person or vehicle enters the scene, this embodiment can be used to determine whether the classification information of the moving target is a person or a vehicle, and if the determination result is yes, output alarm information .

In the above-mentioned another embodiment, the monitoring target is a human face. In this case, the preset alarm conditions may include: the classification information of the face area is: the existence or non-existence in the face database and the modeling The data matches the face data that is successful.

As mentioned above, if it is necessary to implement a stranger (unauthorized person) identification scheme, this embodiment can be used to judge whether there is face data in the face database that matches the modeling data corresponding to the face region successfully, and if so, it means The person corresponding to the face area is an authorized person. If it does not exist, it means that the person corresponding to the face area is a stranger (unauthorized person), and an alarm message is output.

If it is necessary to identify a designated person, this embodiment can be used to determine whether there is face data in the face database that matches the modeling data corresponding to the face area successfully. If there is, it means that the person corresponding to the face area is the designated person. personnel, output alarm information.

In one embodiment, S101 and S102 may be performed by the IPC, and then the IPC sends the captured video image to the NVR, and the NVR performs the subsequent steps.

Applying the embodiment shown in FIG. 1 of the present invention, the monitoring target is detected in the collected video stream; the video image containing the monitoring target is intercepted; the classification information of each monitoring target is obtained by identifying the intercepted video image; In the scheme, instead of classifying and identifying the entire video stream, the video images containing the monitoring target are intercepted, and only the intercepted video images are classified and identified, thus reducing the amount of calculation.

2 is a second schematic flowchart of a video analysis method provided by an embodiment of the present invention, including:

S201: Detect moving objects in the collected video stream.

S202: Capture one or more frames of video images including the moving object.

S203: Input the intercepted video image into the first neural network model obtained by pre-training, use the first neural network model to classify the moving objects in the video image, and obtain each motion output by the first neural network model Classification information for the target.

S204: Determine whether the classification information complies with a preset alarm condition; the preset alarm condition includes: the classification information of the moving object is a person; or, the classification information of the moving object is a vehicle. If so, go to S205.

S205: output alarm information.

For example, in a scene where perimeter protection needs to be performed, the embodiment of FIG. 2 of the present invention can be applied to determine whether a person or vehicle enters the scene, and alarm if the determination result is yes.

Applying the embodiment of FIG. 2 of the present invention, in the first aspect, the first neural network model is used to identify video images, and the classification information of each moving object is obtained. If the classification information is a person or a vehicle, the relevant personnel can be reminded in time for follow-up processing. Effective perimeter prevention; on the other hand, the moving target detection algorithm is firstly performed on the video stream. The moving target detection algorithm can be understood as a rough detection algorithm with a small amount of calculation, and then intercepts a small part of the video images in the video stream, only for This small part of the video image is finely identified, that is, the classification information of the moving target is identified by using the first neural network model. Compared with the overall analysis of the video stream, this solution reduces the amount of calculation.

In some related schemes, an infrared detector is used to emit an infrared laser, which forms a monitoring area. When someone breaks into the monitoring area, the waveform of the infrared laser changes. Therefore, it can be judged whether someone breaks into the monitoring area based on the waveform of the infrared laser. . However, in this solution, the monitoring area formed by the infrared laser emitted by one infrared detector is limited. If the area to be monitored is large, multiple infrared detectors need to be set up, which is costly.

However, with this embodiment, monitoring is performed according to the image collected by the image collecting device, and there is no need to set up multiple infrared detectors, which reduces the monitoring cost.

In the following, an implementation method applied to a surrounding defense scenario is introduced in conjunction with FIG. 3 :

The monitoring point (which can be an IPC) collects a video stream, performs moving object detection on the video stream, intercepts one or more frames of video images containing moving objects according to the detection results, and sends the intercepted video images to the NVR.

The NVR receives the video image sent by the monitoring point, inputs the video image into the first neural network model obtained by pre-training, uses the first neural network model to classify the moving objects in the video image, and obtains the output of the first neural network model The classification information of each moving target. The classification information may be persons, vehicles, objects, etc., which is not specifically limited.

It is assumed that the preset alarm condition is: the classification information of the moving object is a person; or, the classification information of the moving object is a vehicle. If the classification information of the moving target output by the first neural network model is a vehicle or a person, alarm information is output.

In this embodiment, the alarm information is output only when the classification information meets the preset alarm conditions, which can reduce false alarms caused by nuisances, pet interference, and lighting changes, and improve the alarm accuracy.

4 is a third schematic flowchart of a video analysis method provided by an embodiment of the present invention, including:

S401: Perform face recognition in the collected video stream to obtain a recognition result.

S402: According to the recognition result, cut out the face area in the image including the face.

S403: Input the intercepted face region into a second neural network model obtained by pre-training, and use the second neural network model to convert the face region into modeling data.

S404: Obtain classification information of the face region by matching the modeling data with the face data stored in the face database; the classification information includes: existence or non-existence in the face database and the construction The face data for which the model data is successfully matched.

For example, a face image of an authorized person can be collected in advance, the face image can be transformed into modeling data by using the second neural network model, and the transformed modeling data can be stored in a face database as face data.

S405: Determine whether the classification information complies with preset alarm conditions; the preset alarm conditions include: the classification information of the face area is: the existence or absence in the face database and the modeling data are successfully matched face data. If so, go to S406.

S406: output alarm information.

For example, in a scenario where identification of strangers (unauthorized persons) is required, the embodiment of FIG. 4 of the present invention can be applied to store the face data of authorized persons in the face database, and convert the intercepted face area into The modeling data is matched against the face database, that is, to determine whether the person in the video stream is an authorized person. If it is determined that the person in the video stream is a stranger (unauthorized person), an alarm will be issued.

To give another example, if the designated person needs to be identified, the embodiment of Fig. 4 of the present invention can be applied, and the face data of the designated person is stored in the face database, and the modeling data converted from the intercepted face area is compared with the face database. Matching is performed, that is, to determine whether the person in the video stream is the designated person. If it is determined that the person in the video stream is the designated person, an alarm will be issued.

Applying the embodiment shown in FIG. 4 of the present invention, on the one hand, the classification information of the face area can be obtained, according to the classification information, it can be judged whether the person is an authorized person or a designated person, and according to the judgment result, the relevant personnel can be reminded to carry out follow-up processing in time, so that It can realize effective stranger alarm or designated person identification; on the other hand, first perform face recognition on the video stream, and the face recognition algorithm can be understood as a rough detection algorithm with a small amount of calculation, and then intercept the video stream. For a small part of the video image (or image area), only the intercepted part is finely identified, that is, face matching is performed. Compared with the overall analysis of the video stream, this solution reduces the amount of calculation.

Below in conjunction with FIG. 5, an implementation manner applied to a stranger alarming scenario is introduced:

The monitoring point (which can be an IPC) collects a video stream, performs face recognition on the video stream, and intercepts one or more frames of a face image containing a human face according to the recognition result, or intercepts the face area in the image; The face image or face area is sent to the NVR. For the convenience of description, the captured face images or face regions are collectively referred to as face images.

The NVR receives the face image sent by the monitoring point, inputs the face image into the second neural network model obtained by pre-training, and uses the second neural network model to convert the face image into modeling data; The model data is matched with the face data stored in the face database; if the match is successful, it means that the person corresponding to the face area is an authorized person; if the match is unsuccessful, it means that the person corresponding to the face area is a stranger. Output alarm information.

Corresponding to the foregoing method embodiments, an embodiment of the present invention further provides a video analysis apparatus, as shown in FIG. 6 , including:

The detection module 601 is used to detect the monitoring target in the collected video stream;

An interception module 602, configured to intercept a video image containing the monitoring target;

The classification module 603 is configured to obtain classification information of each monitoring target by identifying the intercepted video images.

As an implementation manner, the detection module 601 is specifically configured to: detect moving objects in the collected video stream;

The interception module 602 is specifically configured to: intercept one or more frames of video images including the moving object.

As an implementation manner, the classification module 603 is specifically used for:

Input the intercepted video image into the first neural network model obtained by pre-training, use the first neural network model to classify the moving objects in the video image, and obtain each output of the first neural network model. classification information of a moving target.

As an embodiment, the detection module 601 is specifically used to: perform face recognition in the collected video stream to obtain a recognition result;

The interception module 602 is specifically configured to: intercept the face region from the image containing the human face according to the recognition result;

The classification module 603 is specifically configured to obtain classification information of the face region by matching the intercepted face region with the face data stored in the face database.

As an implementation manner, the classification module 603 is specifically used for:

Inputting the intercepted face region into the second neural network model obtained by pre-training, and using the second neural network model to convert the face region into modeling data;

By matching the modeling data with the face data stored in the face database, the classification information of the face region is obtained, and the classification information includes: existence or non-existence in the face database and the construction The face data for which the model data is successfully matched.

As an embodiment, the device further includes: a first judgment module and a first alarm module (not shown in the figure), wherein,

a first judging module for judging whether the classified information meets the preset alarm condition; if so, triggering the first alarm module;

The first alarm module is used for outputting alarm information.

As an embodiment, the device further includes: a second judgment module and a second alarm module (not shown in the figure), wherein,

The second judgment module is used for judging whether the classification information complies with the preset alarm conditions; the preset alarm conditions include: the classification information of the moving object is a person; or, the classification information of the moving object is a vehicle; if In line with, trigger the second alarm module;

The second alarm module is used for outputting alarm information.

As an embodiment, the device further includes: a third judgment module and a third alarm module (not shown in the figure), wherein,

The third judging module is used to judge whether the classification information meets the preset alarm conditions; the preset alarm conditions include: the classification information of the face area is: existence or non-existence in the face database and the The face data whose modeling data matches successfully; if it matches, the third alarm module is triggered;

The third alarm module is used for outputting alarm information.

In the embodiment of the present invention, the monitoring target is detected in the collected video stream; the video image containing the monitoring target is intercepted; the classification information of each monitoring target is obtained by identifying the intercepted video image; it can be seen that in this scheme, and Instead of classifying and identifying the entire video stream, the video images containing the monitoring target are intercepted, and only the intercepted video images are classified and identified, thus reducing the amount of calculation.

An embodiment of the present invention further provides an electronic device, as shown in FIG. 7 , including a processor 701 and a memory 702,

a memory 702 for storing computer programs;

The processor 701 is configured to implement any of the above video analysis methods when executing the program stored in the memory 702 .

The memory mentioned in the above electronic device may include random access memory (Random Access Memory, RAM), and may also include non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory. As an embodiment, the memory may also be at least one storage device located remote from the aforementioned processor.

The above-mentioned processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; may also be a digital signal processor (Digital Signal Processing, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

An embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any one of the foregoing video analysis methods is implemented.

An embodiment of the present invention further provides a video analysis system, as shown in FIG. 8 , including: a monitoring point and a processing device, wherein:

a monitoring point for detecting a monitoring target in the collected video stream; intercepting a video image containing the monitoring target; sending the intercepted video image to the processing device;

The processing device is used for receiving the video image, and by identifying the received video image, the classification information of each monitoring target is obtained.

For example, the monitoring point may be an IPC, and the processing device may be an NVR, which is not specifically limited.

It should be noted that, in this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any relationship between these entities or operations. any such actual relationship or sequence exists. Moreover, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion such that a process, method, article or device comprising a list of elements includes not only those elements, but also includes not explicitly listed or other elements inherent to such a process, method, article or apparatus. Without further limitation, an element qualified by the phrase "comprising a..." does not preclude the presence of additional identical elements in a process, method, article or apparatus that includes the element.

Each embodiment in this specification is described in a related manner, and the same and similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the apparatus embodiments, device embodiments, computer-readable storage medium embodiments, and system embodiments, since they are basically similar to the method embodiments, the descriptions are relatively simple, and for relevant details, please refer to the partial descriptions of the method embodiments That's it.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims (16)

1. a video analysis method, is characterized in that, comprises: Detect surveillance targets in the captured video stream; intercepting a video image containing the monitoring target; By identifying the intercepted video images, the classification information of each monitoring target is obtained. 2. The method according to claim 1, wherein the detecting a monitoring target in the collected video stream comprises: detecting a moving target in the collected video stream; The intercepting the video image containing the monitoring target, including: One or more frames of video images containing the moving object are captured. 3. method according to claim 2, is characterized in that, described by the video image being intercepted to be identified, obtains the classification information of each monitoring target, comprises: Input the intercepted video image into the first neural network model obtained by pre-training, use the first neural network model to classify the moving objects in the video image, and obtain each output of the first neural network model. classification information of a moving target. 4. method according to claim 1, is characterized in that, described detecting monitoring target in the video stream of collection, comprises: carry out face recognition in the video stream of collection, obtain recognition result; The intercepting the video image containing the monitoring target, including: According to the recognition result, the face area is intercepted in the image containing the face; Described by identifying the intercepted video images, the classification information of each monitoring target is obtained, including: By matching the intercepted face region with the face data stored in the face database, the classification information of the face region is obtained. 5. method according to claim 4, is characterized in that, described by the face data stored in the face region that is intercepted and face database are matched, obtain the classification information of described face region, comprising: Inputting the intercepted face region into the second neural network model obtained by pre-training, and using the second neural network model to convert the face region into modeling data; By matching the modeling data with the face data stored in the face database, the classification information of the face region is obtained, and the classification information includes: existence or non-existence in the face database and the construction The face data for which the model data is successfully matched. 6. The method according to claim 1, characterized in that, after the classification information of each monitoring target is obtained by identifying the intercepted video image, the method further comprises: judging whether the classified information meets the preset alarm condition; If it matches, output an alarm message. 7. The method according to claim 3, wherein after obtaining the classification information of each moving target output by the first neural network model, the method further comprises: Determine whether the classification information meets the preset alarm conditions; if so, output the alarm information; The preset alarm conditions include: The classification information of the moving object is a person; or, the classification information of the moving object is a vehicle. 8. The method according to claim 5, wherein after obtaining the classification information of the face region, further comprising: Determine whether the classification information meets the preset alarm conditions; if so, output the alarm information; The preset alarm conditions include: The classification information of the face region is: the existence or non-existence of face data that is successfully matched with the modeling data in the face database. 9. A video analysis device, comprising: The detection module is used to detect the monitoring target in the collected video stream; An interception module for intercepting a video image containing the monitoring target; The classification module is used to obtain classification information of each monitoring target by identifying the intercepted video images. 10. The device according to claim 9, wherein the detection module is specifically used for: detecting a moving target in the collected video stream; The intercepting module is specifically configured to: intercept one or more frames of video images including the moving object. 11. The device according to claim 10, wherein the classification module is specifically used for: Input the intercepted video image into the first neural network model obtained by pre-training, use the first neural network model to classify the moving objects in the video image, and obtain each output of the first neural network model. classification information of a moving target. 12. The device according to claim 9, wherein the detection module is specifically used for: performing face recognition in the collected video stream to obtain a recognition result; The intercepting module is specifically used for: intercepting the face region in the image containing the human face according to the recognition result; The classification module is specifically configured to obtain classification information of the face region by matching the intercepted face region with the face data stored in the face database. 13. The device according to claim 12, wherein the classification module is specifically used for: Inputting the intercepted face region into the second neural network model obtained by pre-training, and using the second neural network model to convert the face region into modeling data; By matching the modeling data with the face data stored in the face database, the classification information of the face region is obtained, and the classification information includes: existence or non-existence in the face database and the construction The face data for which the model data is successfully matched. 14. The apparatus of claim 9, wherein the apparatus further comprises: a first judging module for judging whether the classified information meets the preset alarm condition; if so, triggering the first alarm module; The first alarm module is used for outputting alarm information. 15. The apparatus of claim 11, wherein the apparatus further comprises: The second judgment module is used for judging whether the classification information complies with the preset alarm conditions; the preset alarm conditions include: the classification information of the moving object is a person; or, the classification information of the moving object is a vehicle; if In line with, trigger the second alarm module; The second alarm module is used for outputting alarm information. 16. The apparatus of claim 13, wherein the apparatus further comprises: The third judging module is used to judge whether the classification information meets the preset alarm conditions; the preset alarm conditions include: the classification information of the face area is: existence or non-existence in the face database and the The face data whose modeling data matches successfully; if it matches, the third alarm module is triggered; The third alarm module is used for outputting alarm information.

Images