CN117392743A - Human body running identification method, device, electronic equipment and storage medium - Google Patents

Abstract

The application is applicable to the technical field of image processing, and provides a human running identification method, a device, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring target images of at least two frames including a target pedestrian; based on each target image, determining a human body center point and a human head detection frame corresponding to each target image; determining the moving speed of the head pixels of the target pedestrians based on each target image and the human body center point and the head detection frame corresponding to each target image; if the moving speed of the human head pixels is greater than or equal to the moving speed threshold of the human head pixels, determining that the target pedestrian is in a running state, compared with the problem that the classification accuracy is low or the recognition time is long in the human body running recognition method in the prior art, the human body running recognition accuracy is improved, and the human body running recognition efficiency in multiple scenes is also improved.

Description

Human body running recognition method, device, electronic equipment and storage medium

Technical field

The present application belongs to the field of image processing technology, and in particular relates to a method, device, electronic equipment and storage medium for identifying human running.

Background technique

In many scenarios such as urban governance, image processing is needed to identify whether pedestrians are running.

The current recognition methods of human running have low classification accuracy or long recognition time and slow response speed, resulting in low recognition efficiency of human running.

There is a problem in the existing technology that the recognition method of human running is inefficient in multiple scenes.

Contents of the invention

Embodiments of the present application provide a method, device, electronic device, and storage medium for identifying human body running, which can solve the problem in the existing technology that the identification method of human body running is inefficient in multiple scenes.

In the first aspect, embodiments of the present application provide a method for identifying human running, including:

Obtaining at least two frames of target images including the target pedestrian;

Based on each of the target images, determine the human body center point and the human head detection frame corresponding to each of the target images;

Based on each of the target images and the human body center point and the head detection frame corresponding to each of the target images, determine the head pixel movement speed of the target pedestrian;

If the moving speed of the head pixel is greater than or equal to the moving speed threshold of the head pixel, it is determined that the target pedestrian is in a running state.

In one of the embodiments, each of the target images includes a first frame target image and a second frame target image;

Based on each of the target images, determine the human body center point and head detection frame corresponding to each of the target images, including:

Based on the first frame target image and the second frame target image, determine the first human body detection frame of the first frame target image and the second human body detection frame of the second frame target image;

Based on the first human body detection frame and the second human body detection frame, determine the first human body center point corresponding to the first human body detection frame and the second human body center point corresponding to the second human body detection frame;

Based on the first human body detection frame and the second human body detection frame, determine a first human head detection frame corresponding to the first human body center point and a second human head detection frame corresponding to the second human body center point. , the first head detection frame is within the first human body detection frame, and the second head detection frame is within the second human body detection frame.

In one of the embodiments, each of the target images includes a first frame target image and a second frame target image;

Based on each of the target images, determining the human body center point and the head detection frame corresponding to each of the target images also includes:

Based on the first frame of the target image, synchronously determine the first human body detection frame and the first head detection frame of the first frame of the target image;

Based on the second frame target image, synchronously determine a second human body detection frame and a second head detection frame of the second frame target image;

Based on the first human body detection frame and the second human body detection frame, a first human body center point corresponding to the first human body detection frame and a second human body center point corresponding to the second human body detection frame are respectively determined.

In one embodiment, each of the target images includes a first frame of target image and a second frame of target image. The first frame of target image corresponds to a first human body center point and a first head detection frame, and the second frame of target image corresponds to a first human body center point and a first head detection frame. The frame target image corresponds to the second human body center point and the second head detection frame;

Based on each of the target images and the human body center point and the head detection frame corresponding to each of the target images, determining the head pixel movement speed of the target pedestrian includes:

Based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the first frame target image and the second frame target image. The interval time between, determine the head pixel moving speed of the target pedestrian.

In one embodiment, the method is based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the first frame target image. The interval time between the second frame of the target image and the second frame of the target image determines the head pixel movement speed of the target pedestrian, including:

Determine the human body pixel movement distance of the target pedestrian based on the first human body center coordinate of the first human body center point and the second human body center coordinate of the second human body center point;

Determine the head of the target pedestrian based on the first corner point coordinates and the second corner point coordinates of the first head detection frame, or based on the third corner point coordinates and the fourth corner point coordinates of the second head detection frame Pixel width, wherein the first corner point corresponding to the first corner point coordinates and the second corner point corresponding to the second corner point coordinates are diagonal points of the first head detection frame, and the second corner point The third corner point corresponding to the corner point coordinates and the fourth corner point corresponding to the fourth corner point coordinates are diagonal points of the second head detection frame;

Based on the human body pixel movement distance, the head pixel width and the interval time, the head pixel movement speed of the target pedestrian is determined.

In one embodiment, the head pixel moving speed includes a first head pixel moving speed along a first direction and a second head pixel moving speed along a second direction, wherein the first direction is perpendicular to the second direction;

The method is based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame, the first frame target image and the second frame target. The interval time between images determines the head pixel movement speed of the target pedestrian, and also includes:

Based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the first frame target image and the second frame target image. The interval time between, determine the pixel movement speed of the first head;

Based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the first frame target image and the second frame target image. determine the pixel movement speed of the second head;

Based on the first head pixel moving speed and the second head pixel moving speed, the head pixel moving speed of the target pedestrian is determined.

In one embodiment, determining the head pixel movement speed threshold includes:

Obtain a first sample data set, a second sample data set and a preset head pixel moving speed. The first sample data set includes the first pixel moving speed of multiple sample pedestrians in a running state. The second sample data The set includes a plurality of second pixel movement speeds of the sample pedestrians in the walking state;

If the preset head pixel movement speed is less than or equal to the first pixel movement speed of the preset proportion of the first sample data set, and the preset head pixel movement speed is greater than or equal to the second sample data set The preset proportional number of second pixel movement speeds determines the preset head pixel movement speed as the head pixel movement speed threshold.

In the second aspect, embodiments of the present application provide a device for identifying human body running, including:

An acquisition module, configured to acquire at least two frames of target images including the target pedestrian;

A first determination module, configured to determine the human body center point and head detection frame corresponding to each target image based on each of the target images;

A second determination module, configured to determine the head pixel movement speed of the target pedestrian based on each of the target images and the human body center point and the head detection frame corresponding to each of the target images;

The third determination module is used to determine that the target pedestrian is in a running state if the moving speed of the head pixel is greater than or equal to the moving speed threshold of the head pixel.

In a third aspect, embodiments of the present application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any of the contents in the first aspect are implemented. The method described in one item.

In the fourth aspect, embodiments of the present application provide a computer-readable storage medium. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, any method in the first aspect is implemented.

Compared with the prior art, the beneficial effects of the embodiments of the present application are:

The first aspect of the embodiment of the present application provides a method for identifying human running by acquiring at least two frames of target images including target pedestrians; based on each target image, determining the human body center point and head detection frame corresponding to each target image; based on each target image The target image and the human body center point and head detection frame corresponding to each target image determine the head pixel moving speed of the target pedestrian; if the head pixel moving speed is greater than or equal to the head pixel moving speed threshold, it is determined that the target pedestrian is in a running state. Since it can pass The human body center point and head detection frame corresponding to the target pedestrian's target image are used to determine the head pixel moving speed of the target pedestrian, and compared with the head pixel moving speed threshold, it can be determined whether the target pedestrian is in a running state. Compared with the existing technology This method not only improves the accuracy of recognition of human running, but also improves the recognition efficiency of human running in multiple scenes.

It can be understood that the beneficial effects of the above-mentioned second aspect, third aspect and fourth aspect can be referred to the relevant description in the above-mentioned first aspect, and will not be described again here.

Description of the drawings

In order to more clearly explain the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description The drawings illustrate some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without exerting creative efforts.

Figure 1 is a schematic flowchart of a human running recognition method provided by an embodiment of the present application;

Figure 2 is a schematic flow chart of determining the human body center point and head detection frame corresponding to each target image based on each target image provided by the embodiment of the present application;

Figure 3 is a schematic diagram of the first frame target image and the second frame target image synthesized in the image coordinate system according to an embodiment of the present application;

Figure 4 is a schematic flowchart of determining the human body center point and the head detection frame corresponding to each target image based on each target image according to another embodiment of the present application;

Figure 5 shows an embodiment of the present application based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the relationship between the first frame target image and the second frame target image. Interval time, process diagram for determining the pixel movement speed of the target pedestrian’s head;

Figure 6 shows another embodiment of the present application based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the relationship between the first frame target image and the second frame target image. The schematic flow chart of determining the pixel moving speed of the target pedestrian's head at the interval;

Figure 7 shows another embodiment of the present application based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the relationship between the first frame target image and the second frame target image. The schematic flow chart of determining the pixel moving speed of the target pedestrian's head at the interval;

Figure 8 is a schematic flowchart of determining that the target pedestrian is in a running state if the head pixel movement speed is greater than or equal to the head pixel movement speed threshold provided by an embodiment of the present application;

Figure 9 is a schematic structural diagram of a running recognition device provided by an embodiment of the present application.

Detailed ways

In the following description, for the purpose of explanation rather than limitation, specific details such as specific system structures and technologies are provided to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be practiced in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the application with unnecessary detail.

It will be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integers, steps, operations, elements and/or components but does not exclude one or more other The presence or addition of features, integers, steps, operations, elements, components and/or collections thereof.

It will also be understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

As used in this specification and the appended claims, the term "if" may be interpreted as "when" or "once" or "in response to determining" or "in response to detecting" depending on the context. ". Similarly, the phrase "if determined" or "if [the described condition or event] is detected" may be interpreted, depending on the context, to mean "once determined" or "in response to a determination" or "once the [described condition or event] is detected ]" or "in response to detection of [the described condition or event]".

In addition, in the description of this application and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

Reference in this specification to "one embodiment" or "some embodiments" or the like means that a particular feature, structure or characteristic described in connection with the embodiment is included in one or more embodiments of the application. Therefore, the phrases "in one embodiment", "in some embodiments", "in other embodiments", "in other embodiments", etc. appearing in different places in this specification are not necessarily References are made to the same embodiment, but rather to "one or more but not all embodiments" unless specifically stated otherwise. The terms “including,” “includes,” “having,” and variations thereof all mean “including but not limited to,” unless otherwise specifically emphasized. "Multiple" means "two or more".

In many scenarios such as urban governance, image processing is needed to identify whether pedestrians are running.

The current recognition methods of human running have low classification accuracy or long recognition time and slow response speed, resulting in low recognition efficiency of human running.

Therefore, there is a problem in the existing technology that the human running recognition method has low recognition efficiency in multiple scenes.

In response to the above problems, the embodiment of the present application provides a method for identifying human running by acquiring at least two frames of target images including target pedestrians; based on each target image, determining the human body center point and head detection frame corresponding to each target image; based on each target image. The target image and the human body center point and head detection frame corresponding to each target image determine the head pixel moving speed of the target pedestrian; if the head pixel moving speed is greater than or equal to the head pixel moving speed threshold, it is determined that the target pedestrian is in a running state. Since it can pass The human body center point and head detection frame corresponding to the target pedestrian's target image are used to determine the head pixel moving speed of the target pedestrian, and compared with the head pixel moving speed threshold, it can be determined whether the target pedestrian is in a running state. Compared with the existing technology This method not only improves the accuracy of recognition of human running, but also improves the recognition efficiency of human running in multiple scenes.

The following is an exemplary description of the running recognition method provided by this application in conjunction with specific embodiments.

In the first aspect, as shown in Figure 1, this embodiment provides a method for identifying human running, including:

S100, obtain at least two frames of target images including the target pedestrian.

In one embodiment, the target image including at least two frames of the target pedestrian is acquired through a device with image acquisition or video acquisition, for example, by capturing at least 2 frames of the image including the target pedestrian through a mobile terminal such as an electronic eye, or from a surveillance camera. Obtain at least 2 frames of images of the target pedestrian from the video stream, where the target pedestrian is one or more pedestrians to be identified as running human body, so that the running state of the target pedestrian can be identified through at least 2 frames of images, and the target image is specific The number of frames is determined according to the scene requirements. The more frames of the target image, the more accurate the recognition of the running state of the human body, but the recognition time of the running state will also increase and the recognition efficiency will decrease.

S200: Based on each target image, determine the human body center point and head detection frame corresponding to each target image.

In one embodiment, based on each target image, the human body center point and the human head detection frame corresponding to each target image are determined, so that the interval time between images can be obtained through each target image, and the pedestrian pixel movement distance can be obtained through each human body center point. It is convenient to obtain the width of the head through the head detection frame to further obtain the pixel movement speed of the head.

In one embodiment, each target image includes a first frame of target image and a second frame of target image, wherein both the first frame of target image and the second frame of target image include a target pedestrian, which facilitates obtaining the human body detection frame and head of the target pedestrian. Detection box.

In one embodiment, as shown in Figure 2, based on each target image, the human body center point and the head detection frame corresponding to each target image are determined, including:

S211. Based on the first frame of the target image and the second frame of the target image, determine the first human body detection frame of the first frame of the target image and the second human body detection frame of the second frame of the target image.

In one embodiment, as shown in Figure 3, based on the first frame of the target image (the left image in the figure) and the second frame of the target image (the right image in the figure), the target pedestrian is tracked through the trained human body detection model , and determine the first human body detection frame of the target pedestrian in the first frame of the target image and the second human body detection frame in the second frame of the target image, which is beneficial to determining the pixel movement distance of the target pedestrian through each human body detection frame.

In one embodiment, the human detection model includes a Yolo target detection model or a CenterNet network model, and the first frame target image and the second frame target image are detected through the built Yolo (You Only Look Once) target detection model or CenterNet network model. Whether the target pedestrian is included, and the first human body detection frame of the target pedestrian in the first frame of the target image and the second human body detection frame of the second frame of the target image are obtained. Among them, the CenterNet network model includes a backbone network model and a target detection model. The target images of multiple frames are images after normalizing the pixel values of each frame image.

S212: Based on the first human body detection frame and the second human body detection frame, determine the first human body center point corresponding to the first human body detection frame and the second human body center point corresponding to the second human body detection frame.

In one embodiment, as shown in Figure 3, based on the first human body detection frame and the second human body detection frame, the geometric center of the first human body detection frame is used as the corresponding first human body center point, and the second human body detection frame is used as the corresponding first human body center point. The geometric center serves as the corresponding second human body center point, which facilitates quick confirmation of the coordinates of the first human body center point and the second human body center point, and improves the efficiency of human running recognition.

S213. Based on the first human body detection frame and the second human body detection frame, determine the first human head detection frame corresponding to the first human body center point and the second human head detection frame corresponding to the second human body center point.

In one embodiment, as shown in Figure 3, based on the first human body detection frame and the second human body detection frame, the first human head detection frame corresponding to the first human body center point and the first human head detection frame and the second human body detection frame are determined through the trained human head detection model. The second head detection frame corresponding to the center point of the second human body; wherein, the first head detection frame is within the first human body detection frame, and the second head detection frame is within the second human body detection frame. After obtaining the human body detection frame, Determining the target pedestrian's head detection frame from the human body detection frame reduces the image range for identifying the target pedestrian's head, which is beneficial to improving the computing speed of human running recognition and improving the efficiency of human running recognition. At the same time, the first head detection frame Corresponding to the first human body center point, the second human head detection frame corresponds to the second human body center point. Corresponding the human head detection frame to the corresponding human body center point avoids misidentification of the human head detection frame and improves the accuracy of human running recognition. .

In one embodiment, training the head detection model includes: obtaining a first training set and a first verification set. The first training set includes annotated objects similar to human heads and annotated target images containing target pedestrians. The first verification set The set includes unlabeled target images containing target pedestrians; the first training set is used to train the head detection model until the accuracy of the head detection model predicting the head detection frame of the first verification set is greater than or equal to the preset percentage, and the preset The fluctuation range of the prediction accuracy is less than or equal to the preset percentage range. Among them, the value range of the preset percentage is greater than or equal to 99%, and the preset percentage range is 0.1%. The head detection model includes at least one of the Yolo target detection model, CenterNet network model or Faster R-CNN. Since the training set includes The object is similar to the human head for training, which improves the accuracy of identifying the human head detection frame in the target image.

In another embodiment, as shown in Figure 4, based on each target image, determining the human body center point and head detection frame corresponding to each target image also includes:

S221. Based on the first frame of the target image, synchronously determine the first human body detection frame and the first head detection frame of the first frame of the target image.

S222: Based on the target image of the second frame, synchronously determine the second human body detection frame and the second head detection frame of the target image of the second frame.

S223: Based on the first human body detection frame and the second human body detection frame, respectively determine the first human body center point corresponding to the first human body detection frame and the second human body center point corresponding to the second human body detection frame.

In another embodiment, since the first human body detection frame and the first head detection frame of the first frame target image are determined synchronously through the human body detection model and the head detection model directly in the first frame target image, in the second frame target image In the image, the human detection model and the human head detection model are used to simultaneously determine the second human body detection frame and the second human head detection frame of the second frame of the target image, without having to first determine the first human body detection frame in the first frame of the target image, and then determine the second human body detection frame in the first frame of the target image. The first head detection frame within a human body detection frame does not need to first determine the second human body detection frame in the second frame of the target image, and then determine the second head detection frame within the second human body detection frame, thus improving the recognition of head detection. The speed of the frame improves the efficiency of human running recognition.

S300: Determine the head pixel movement speed of the target pedestrian based on each target image and the human body center point and head detection frame corresponding to each target image.

In one embodiment, based on each target image and the human body center point and head detection frame corresponding to each target image, the head pixel moving speed of the target pedestrian is determined without mapping to the physical speed of the target pedestrian in the actual scene, which improves the determination The speed of the target pedestrian's head pixel movement speed improves the efficiency of human running recognition.

In one embodiment, each target image includes a first frame target image and a second frame target image. The first frame target image corresponds to the first human body center point and the first head detection frame, and the second frame target image corresponds to the second human body. Center point and second head detection frame.

In one embodiment, determining the head pixel movement speed of the target pedestrian based on each target image and the human body center point and head detection frame corresponding to each target image includes:

Determine the head pixel movement of the target pedestrian based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the interval time between the first frame of the target image and the second frame of the target image. Since the speed does not need to be mapped to the physical speed of the target pedestrian in the actual scene, the speed of determining the head pixel movement speed of the target pedestrian is improved, and the efficiency of human running recognition is improved.

In one embodiment, as shown in Figure 5, based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the first frame target image and the second frame target image. The interval time between the target pedestrian's head and pixel movement speed is determined, including:

S311. Determine the human body pixel movement distance of the target pedestrian based on the first human body center coordinate of the first human body center point and the second human body center coordinate of the second human body center point.

In one embodiment, since the human body detection model can form a motion trajectory of the target pedestrian at different times in the image coordinate system, as shown in Figure 3, the first frame of the target image and the second frame of the target image including the target pedestrian are synthesized. In the same image coordinate system, that is, it is equivalent to a composite target image obtained by overlapping the first frame target image and the second frame target image in the same image coordinate system. The image coordinate system of the target pedestrian's movement trajectory is based on the composite image. The upper left corner is the origin, the vertical downward direction is the X axis, and the horizontal right direction is the Y axis. The first human body center coordinate of the target pedestrian in the first frame of the target image on the left is The second human body center coordinate of the target pedestrian in the second frame of the target image on the right is/> Based on the first human body center coordinates of the first human body center point A0 and the second human body center coordinates of the second human body center point A1, the human body pixel movement distance L of the target pedestrian is determined through the Euclidean distance calculation formula, that is, the human body pixel movement distance L is The pixel distance between the first human body center point A0 and the second human body center point A1.

In one embodiment, the Euclidean distance calculation formula for obtaining the human body pixel movement distance L1 is as follows:

S312. Determine the head pixel width of the target pedestrian based on the first corner point coordinates and the second corner point coordinates of the first head detection frame, or based on the third corner point coordinates and the fourth corner point coordinates of the second head detection frame.

In one embodiment, as shown in Figure 3, the first corner point corresponding to the first corner point coordinates and the second corner point corresponding to the second corner point coordinates are diagonal points of the first head detection frame, that is, the first The corner point H01 and the diagonal point H03 of the head detection frame; the third corner point corresponding to the coordinates of the second corner point and the fourth corner point corresponding to the coordinates of the fourth corner point are the diagonal points of the second head detection frame, that is, the second The corner point H11 and the diagonal point H13 of the head detection frame are conducive to determining the head pixel width of the target pedestrian in the horizontal or vertical direction.

In one embodiment, as shown in Figure 3, based on the first corner point coordinates of the corner point H01 of the first head detection frame and the coordinates of the second corner point of the opposite corner/> Or based on the third corner point coordinates of the second head detection frame/> and the fourth corner coordinate of the opposite corner/> Determine the head pixel width of the target pedestrian, that is, the head pixel width/> Or head pixel width

S313: Determine the head pixel moving speed of the target pedestrian based on the human body pixel moving distance, head pixel width and interval time.

In one embodiment, based on the human body pixel movement distance L, the head pixel width and the interval time t, the head pixel movement speed v of the target pedestrian is determined through the head pixel movement speed calculation formula. Since there is no need to map the head pixel movement speed to the target pedestrian The physical speed in the actual scene, and only determining the head pixel width for one head detection frame, further improves the speed of determining the head pixel moving speed of the target pedestrian, and further improves the efficiency of human running recognition.

In one embodiment, the calculation formula of the head pixel movement speed is:

v＝L/W/t

Among them, v is the moving speed of the human head pixels, L is the moving distance of the human head pixels, W is the pixel width of the human head, and t is the interval time, that is, the interval time is between the moment corresponding to the first frame of the target image and the moment corresponding to the second frame of the target image. period.

In one embodiment, the value range of the interval time is less than or equal to 1 second. In this embodiment, the specific value of the interval time is set according to the needs of human running recognition in different scenarios. For example, the interval time can also be 0.5 seconds.

In another embodiment, the head pixel moving speed includes a first head pixel moving speed along a first direction and a second head pixel moving speed along a second direction, where the first direction is perpendicular to the second direction, for example, The first direction is the horizontal direction (ie, the Y-axis direction of the image coordinate system), and the second direction is the vertical direction (ie, the X-axis direction of the image coordinate system).

In another embodiment, as shown in Figure 6, based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame, the first frame target image and the second frame target image The interval time between determines the head pixel movement speed of the target pedestrian, and also includes:

S321: Determine the first person's avatar based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the interval between the first frame target image and the second frame target image. element movement speed.

In another embodiment, based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the interval time between the first frame target image and the second frame target image, To determine the moving speed of the first head pixel, since the moving speed of the first head pixel in the horizontal direction is calculated, the accuracy of the moving speed of the head pixel is further improved, and the accuracy of human running recognition is further improved.

In another embodiment, as shown in Figure 7, based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame, the first frame target image and the second frame target image The interval time between determines the moving speed of the first head pixel, including:

S3211. Based on the first human body center coordinate of the first human body center point and the second human body center coordinate of the second human body center point, determine the first human body pixel movement distance of the target pedestrian along the first direction.

In another embodiment, the first body center coordinate of the target pedestrian based on the first frame of the target image is and the second human body center coordinates of the target pedestrian in the second frame of the target image are/> Determine the movement distance of the first human pixel of the target pedestrian along the first direction/>

S3212: Determine the target pedestrian based on the first corner point coordinates and the second corner point coordinates of the diagonal point of the first head detection frame, and the third corner point coordinates and the fourth corner point coordinates of the diagonal point of the second head detection frame. The width of the first head pixel along the first direction,

In another embodiment, the position of the first corner point corresponding to the first corner point coordinates in the first head detection frame is the same as the position of the third corner point corresponding to the third corner point coordinates in the second head detection frame. For example, the first The position of the corner point H01 of the head detection frame is the same as the position of the corner point H11 of the second head detection frame.

In another embodiment, based on the first corner point coordinates of the corner point H01 of the first head detection frame and the second corner point coordinates of the diagonal point H03/> And the third corner point coordinates of the corner point H11 of the second head detection frame/> and the coordinates of the fourth corner point of diagonal point H13 The first head pixel width of the target pedestrian along the first direction is determined by the first head pixel width calculation formula, because by taking the average head pixel width of the first head detection frame and the second head detection frame in the first direction is the first head pixel width W1, which improves the accuracy of the first head pixel width and also improves the accuracy of human running recognition.

In another embodiment, the calculation formula of the first head pixel width is:

Among them, W1 is the pixel width of the first head.

S3213. Based on the first human body pixel movement distance and the first head pixel width, determine the first head pixel movement speed of the target pedestrian along the first direction.

In another embodiment, based on the first body pixel movement distance and the first head pixel width, the first head pixel movement speed of the target pedestrian along the first direction is determined through a first head pixel movement speed calculation formula.

In another embodiment, the calculation formula of the first head pixel moving speed is:

v1＝L1/W1/t

Among them, v1 is the moving speed of the first human head pixel, L1 is the moving distance of the first human head pixel, W1 is the pixel width of the first human head, t is the interval time, that is, the interval time is the moment corresponding to the first frame of the target image and the second The time period between the moments corresponding to the frame target image.

S322: Determine the second head pixel based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the interval between the first frame target image and the second frame target image. Moving speed.

In another embodiment, based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the interval time between the first frame target image and the second frame target image, Determine the pixel movement speed of the second head, including:

S3211. Based on the first human body center coordinate of the first human body center point and the second human body center coordinate of the second human body center point, determine the second human body pixel movement distance of the target pedestrian along the second direction.

In another embodiment, the first body center coordinate of the target pedestrian based on the first frame of the target image is and the second human body center coordinates of the target pedestrian in the second frame of the target image are/> Determine the movement distance of the second human body pixel of the target pedestrian in the second direction/>

S3212: Determine the target pedestrian based on the first corner point coordinates and the second corner point coordinates of the diagonal point of the first head detection frame, and the third corner point coordinates and the fourth corner point coordinates of the diagonal point of the second head detection frame. The pixel width of the second head along the second direction.

In another embodiment, the position of the first corner point corresponding to the first corner point coordinates in the first head detection frame is the same as the position of the third corner point corresponding to the third corner point coordinates in the second head detection frame. For example, the first The position of the corner point H01 of the head detection frame is the same as the position of the corner point H11 of the second head detection frame.

In another embodiment, based on the first corner point coordinates of the corner point H01 of the first head detection frame and the second corner point coordinates of the diagonal point H03/> And the third corner point coordinates of the corner point H11 of the second head detection frame/> and the coordinates of the fourth corner point of diagonal point H13 The second head pixel width of the target pedestrian along the second direction is determined by the second head pixel width calculation formula. Since the average head pixel width of the first head detection frame and the second head detection frame in the second direction is the The pixel width of the second head W2 improves the accuracy of the pixel width of the second head and also improves the accuracy of human running recognition.

In another embodiment, the calculation formula of the second head pixel width is:

Among them, W2 is the pixel width of the second head.

S3213. Based on the second human body pixel movement distance and the second head pixel width, determine the second head pixel movement speed of the target pedestrian along the second direction.

In another embodiment, based on the second body pixel movement distance and the second head pixel width, the second head pixel movement speed of the target pedestrian along the second direction is determined through a second head pixel movement speed calculation formula.

In another embodiment, the calculation formula of the second head pixel movement speed is:

v2＝L2/W2/t

Among them, v2 is the pixel moving speed of the second human head, L2 is the moving distance of the second human head pixel, W2 is the pixel width of the second human head, t is the interval time, that is, the interval time is the moment corresponding to the first frame of the target image and the second frame of the target. The time period between the corresponding moments of the image.

S323: Determine the head pixel moving speed of the target pedestrian based on the first head pixel moving speed and the second head pixel moving speed.

In another embodiment, based on the first head pixel moving speed v1 and the second head pixel moving speed v2, the head pixel moving speed of the target pedestrian is determined by synthesizing the head pixel moving speed calculation formula. The moving speed of the first human body pixel in one direction is determined, and then the moving speed of the second human body pixel in the second direction of the target pedestrian is determined, and then the first human body pixel moving speed and the second human body pixel moving speed are synthesized into the head pixel moving speed, and further It improves the accuracy of the pixel movement speed of the human head and further improves the accuracy of human running recognition.

In another embodiment, the calculation formula of the synthetic human head pixel movement speed is:

Among them, v is the pixel movement speed of the head, v1 is the pixel movement speed of the first head, and v2 is the pixel movement speed of the second head.

S400: If the head pixel moving speed is greater than or equal to the head pixel moving speed threshold, determine that the target pedestrian is in a running state.

In one embodiment, if the head pixel movement speed is greater than or equal to the head pixel movement speed threshold, the target pedestrian is determined to be in a running state. Since there is no need to map the head pixel movement speed to the target pedestrian's physical movement speed in the scene, human running recognition is reduced. The computing time improves the efficiency of human running recognition.

In one embodiment, as shown in Figure 8, determining the head pixel movement speed threshold includes:

S410: Obtain the first sample data set, the second sample data set and the preset human head pixel moving speed. The first sample data set includes the first pixel moving speed of multiple sample pedestrians in the running state. The second sample data set includes The second pixel movement speed of multiple sample pedestrians in the walking state.

S420, if the preset head pixel movement speed is less than or equal to the first pixel movement speed of the preset proportion number of the first sample data set, and the preset head pixel movement speed is greater than or equal to the preset proportion number of the second sample data set The second pixel moving speed is determined to be the preset head pixel moving speed as the head pixel moving speed threshold.

In one embodiment, the value range of the preset ratio is greater than or equal to 95%, which is beneficial to obtaining more accurate human head pixel movement speed thresholds in multiple scenes, and improves the accuracy of human running recognition in multiple scenes.

Compared with the prior art, the beneficial effects of the embodiments of the present application are:

The first aspect of the embodiment of the present application provides a method for identifying human running by acquiring at least two frames of target images including target pedestrians; based on each target image, determining the human body center point and head detection frame corresponding to each target image; based on each target image The target image and the human body center point and head detection frame corresponding to each target image determine the head pixel moving speed of the target pedestrian; if the head pixel moving speed is greater than or equal to the head pixel moving speed threshold, it is determined that the target pedestrian is in a running state. Since it can pass The human body center point and head detection frame corresponding to the target pedestrian's target image are used to determine the head pixel moving speed of the target pedestrian, and compared with the head pixel moving speed threshold, it can be determined whether the target pedestrian is in a running state. Compared with the existing technology This method not only improves the accuracy of recognition of human running, but also improves the recognition efficiency of human running in multiple scenes.

The following is an exemplary description of the running recognition device provided by the present application with reference to the accompanying drawings.

Corresponding to the running identification method described in the above embodiment, in the second aspect, as shown in Figure 9, an embodiment of the present application provides a human running identification device 100, which includes:

The acquisition module 110 is configured to acquire at least two frames of target images including the target pedestrian.

The first determination module 120 is configured to determine, based on each of the target images, the human body center point and the human head detection frame corresponding to each of the target images.

The second determination module 130 is configured to determine the head pixel moving speed of the target pedestrian based on each of the target images and the human body center point and the head detection frame corresponding to each of the target images.

The third determination module 140 is configured to determine that the target pedestrian is in a running state if the moving speed of the head pixel is greater than or equal to the moving speed threshold of the head pixel.

It should be noted that the information interaction, execution process, etc. between the above-mentioned modules/units are based on the same concept as the method embodiments of this application, and their specific functions and technical effects can be found in the method embodiments section. No further details will be given.

Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above functional units and modules is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs. Module completion means dividing the internal structure of the device into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be hardware-based. It can also be implemented in the form of software functional units. In addition, the specific names of each functional unit and module are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present application. For the specific working processes of the units and modules in the above system, please refer to the corresponding processes in the foregoing method embodiments, and will not be described again here.

In a third aspect, embodiments of the present application also provide an electronic device, including a memory, a processor 902, and a computer program stored in the memory and executable on the processor. The processor 902 executes the The computer program implements various steps of the run recognition method as described above.

In applications, the electronic device may include, but is not limited to, a processor and a memory. The electronic device may also include more or less components than shown, or a combination of certain components, or different components, such as input and output devices. , network access equipment, etc. Input and output devices can include cameras, audio collection/playback devices, displays, etc. The network access device may include a network module for wireless networking with external devices.

In an application, the processor can be a central processing unit (CPU), which can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit) , ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

In applications, the memory in some embodiments may be an internal storage unit of the electronic device, such as a hard drive or memory of the electronic device. In other embodiments, the memory may also be an external storage device of the electronic device, for example, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SD) card, or a flash memory equipped on the electronic device. Flash Card, etc. Memory may also include both internal storage units of the electronic device and external storage devices. Memory is used to store operating systems, applications, boot loaders, data, and other programs, such as program codes of computer programs. The memory can also be used to temporarily store data that has been output or will be output.

In a fourth aspect, embodiments of the present application also provide a computer-readable storage medium. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps in each of the above method embodiments can be implemented.

This application implements all or part of the processes in the above embodiment method, which can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by the processor, the computer program can Implement the steps of each of the above method embodiments. Wherein, the computer program includes computer program code, which may be in the form of source code, object code, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device capable of carrying computer program code to an electronic device, recording media, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signals, telecommunications signals, and software distribution media. For example, U disk, mobile hard disk, magnetic disk or CD, etc.

In the above embodiments, each embodiment is described with its own emphasis. For parts that are not detailed or documented in a certain embodiment, please refer to the relevant descriptions of other embodiments.

Those of ordinary skill in the art will appreciate that the devices and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may implement the described functionality using different methods for each specific application, but such implementations should not be considered beyond the scope of this application.

In the embodiments provided in this application, it should be understood that the disclosed devices/devices and methods can be implemented in other ways. For example, the apparatus/equipment embodiments described above are only illustrative. For example, the division of modules or units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or units. Components can be combined or can be integrated into another system, or some features can be ignored, or not implemented. On the other hand, the coupling or direct coupling or communication connection between each other shown or discussed may be through some interfaces, indirect coupling or communication connection of devices or units, which may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

The above-described embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still implement the above-mentioned implementations. The technical solutions described in the examples are modified, or some of the technical features are equivalently replaced; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions in the embodiments of this application, and should be included in within the protection scope of this application.

Claims (10)

1. A method for identifying human running, comprising:

acquiring target images of at least two frames including a target pedestrian;

based on each target image, determining a human body center point and a human head detection frame corresponding to each target image;

Determining the moving speed of the head pixels of the target pedestrian based on each target image and the human body center point and the head detection frame corresponding to each target image;

and if the moving speed of the head pixels is greater than or equal to the threshold value of the moving speed of the head pixels, determining the artificial running state of the target pedestrian.

2. The method of claim 1, wherein each of the target images comprises a first frame target image and a second frame target image;

based on each target image, determining a human body center point and a human head detection frame corresponding to each target image, including:

determining a first human body detection frame of the first frame target image and a second human body detection frame of the second frame target image based on the first frame target image and the second frame target image;

determining a first human body center point corresponding to the first human body detection frame and a second human body center point corresponding to the second human body detection frame based on the first human body detection frame and the second human body detection frame;

based on the first human body detection frame and the second human body detection frame, determining a first human head detection frame corresponding to the first human body center point and a second human head detection frame corresponding to the second human body center point, wherein the first human head detection frame is positioned in the first human body detection frame, and the second human head detection frame is positioned in the second human body detection frame.

3. The method of claim 1, wherein each of the target images comprises a first frame target image and a second frame target image;

based on each target image, determining a human body center point and a human head detection frame corresponding to each target image, and further comprising:

based on the first frame target image, synchronously determining a first human body detection frame and a first human head detection frame of the first frame target image;

synchronously determining a second human body detection frame and a second human head detection frame of the second frame target image based on the second frame target image;

and based on the first human body detection frame and the second human body detection frame, respectively determining a first human body center point corresponding to the first human body detection frame and a second human body center point corresponding to the second human body detection frame.

4. The method of claim 1, wherein each of the target images comprises a first frame of target image and a second frame of target image, the first frame of target image corresponding to a first human body center point and a first human head detection frame, the second frame of target image corresponding to a second human body center point and a second human head detection frame;

based on each of the target images and the human body center point and the human head detection frame corresponding to each of the target images, determining a human head pixel movement speed of the target pedestrian includes:

And determining the moving speed of the head pixels of the target pedestrian based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame and the interval time between the first frame target image and the second frame target image.

5. The method of claim 4, wherein the determining the head pixel movement speed of the target pedestrian based on the first human body center point, the second human body center point, the first human head detection frame, the second human head detection frame, and a spacing time between the first frame target image and the second frame target image comprises:

determining a human body pixel moving distance of the target pedestrian based on the first human body center coordinates of the first human body center point and the second human body center coordinates of the second human body center point;

determining the width of a head pixel of the target pedestrian based on a first corner coordinate and a second corner coordinate of the first head detection frame or based on a third corner coordinate and a fourth corner coordinate of the second head detection frame, wherein a first corner corresponding to the first corner coordinate and a second corner corresponding to the second corner coordinate are diagonal points of the first head detection frame, and a third corner corresponding to the second corner coordinate and a fourth corner corresponding to the fourth corner coordinate are diagonal points of the second head detection frame;

And determining the head pixel moving speed of the target pedestrian based on the human body pixel moving distance, the head pixel width and the interval time.

6. The method of claim 4, wherein the head pixel movement speed comprises a first head pixel movement speed in a first direction and a second head pixel movement speed in a second direction, wherein the first direction is perpendicular to the second direction;

the determining the moving speed of the head pixel of the target pedestrian based on the first human body center point, the second human body center point, the first head detection frame, the second head detection frame, and the interval time between the first frame target image and the second frame target image further includes:

determining the first human head pixel moving speed based on the first human body center point, the second human body center point, the first human head detecting frame, the second human head detecting frame and the interval time between the first frame target image and the second frame target image;

determining the second human head pixel moving speed based on the first human body center point, the second human body center point, the first human head detecting frame, the second human head detecting frame and the interval time between the first frame target image and the second frame target image;

And determining the head pixel moving speed of the target pedestrian based on the first head pixel moving speed and the second head pixel moving speed.

7. The method of claim 1, wherein determining the head pixel movement speed threshold comprises:

acquiring a first sample data set, a second sample data set and a preset head pixel moving speed, wherein the first sample data set comprises a plurality of first pixel moving speeds of sample pedestrians in a human running state, and the second sample data set comprises a plurality of second pixel moving speeds of the sample pedestrians in a walking state;

if the preset head pixel moving speed is smaller than or equal to the first pixel moving speed of the preset proportion number of the first sample data set, and the preset head pixel moving speed is larger than or equal to the second pixel moving speed of the preset proportion number of the second sample data set, determining that the preset head pixel moving speed is the head pixel moving speed threshold.

8. An identification device for running by a human body, comprising:

the acquisition module is used for acquiring target images of at least two frames including a target pedestrian;

The first determining module is used for determining a human body center point and a human head detection frame corresponding to each target image based on each target image;

the second determining module is used for determining the moving speed of the head pixels of the target pedestrian based on the target images, the human body center point corresponding to the target images and the head detection frame;

and the third determining module is used for determining the artificial running state of the target pedestrian if the moving speed of the head pixel is greater than or equal to the threshold value of the moving speed of the head pixel.

9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 7 when the computer program is executed.

10. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the method according to any one of claims 1 to 7.