Personnel statistical method and system for exhibition activities

Abstract

The invention is applicable to the technical field of exhibition management, and particularly relates to a personnel statistical method and system for exhibition activities, and the method comprises the steps: collecting a field monitoring image through an image collection device; splicing the field monitoring images at the same moment together to obtain monitoring spliced images, and arranging the monitoring spliced images according to a time sequence; carrying out portrait recognition, extracting portrait data contained in the portrait data, comparing all the portrait data, and counting the total number of the portrait data; and obtaining an exhibition region map, counting the total area of the exhibition region, calculating the proportion of the monitoring region, and calculating the number of personnel based on the proportion of the monitoring region. According to the invention, the image acquisition device is used for carrying out field image acquisition, the portrait data contained in the image is actively identified through the portrait identification technology, and the portrait data is counted so as to calculate and obtain the number of exhibition participants in the whole exhibition activity range, so that the automatic counting of the number of exhibition participants is realized, and the privacy of the exhibition participants is protected.

Description

Technical Field

[0001] The invention belongs to the technical field of exhibition management, and in particular relates to a personnel counting method and system for exhibition activities. Background Art

[0002] Exhibition activity crowd counting plays a vital role in evaluating and optimizing exhibition results. It not only helps organizers understand the number of visitors on site, but also provides data support for future event planning by analyzing the peaks and troughs of traffic in different time periods. In addition, accurate crowd counting helps measure the effectiveness of marketing, determine which publicity channels are most attractive to the target group, and adjust marketing strategies accordingly. At the same time, it is also one of the key indicators for evaluating exhibitors' return on investment (ROI), enabling exhibitors to make more effective business decisions based on the actual flow of visitors. Finally, from a safety perspective, reasonable control of the number of people in the venue ensures the safe and orderly conduct of the event and avoids safety hazards caused by overcrowding. In summary, exhibition activity crowd counting is an important bridge connecting organizers, exhibitors and visitors, and is indispensable for improving exhibition quality and efficiency.

[0003] In current exhibition activities, the number of exhibitors is mainly counted through the access control system, and the number of exhibitors is determined based on the difference in the number of people entering and leaving the access control system. For exhibition activities without an access control system, the number of people is difficult to count directly. Summary of the invention

[0004] The purpose of the present invention is to provide a method for counting people in exhibition activities, aiming to solve the problem that the number of people in exhibition activities without a door control system is difficult to directly count.

[0005] The present invention is implemented as follows: a method for counting people in exhibition activities, the method comprising: Multiple image acquisition devices are set up within the scope of the exhibition activities, and on-site monitoring images are acquired through the image acquisition devices; Acquire all on-site surveillance images, stitch together the on-site surveillance images at the same time to obtain a surveillance stitching image, and arrange the surveillance stitching images in chronological order; Use the preset face recognition model to perform portrait recognition on the surveillance spliced ​​images, extract the portrait data contained therein, compare all the portrait data, and count the total number of portrait data; The exhibition area map is obtained, the total area in the exhibition area map is counted, and the monitoring area ratio of the monitoring area corresponding to the image acquisition device is calculated, and the number of people is calculated based on the monitoring area ratio.

[0006] Preferably, the step of acquiring all the on-site surveillance images, splicing the on-site surveillance images at the same time together to obtain a surveillance spliced ​​image, and arranging the surveillance spliced ​​images in chronological order specifically includes: Acquire data from the image acquisition device at preset time intervals, and store the on-site monitoring images acquired at the same time together; Preprocessing the on-site monitoring image, performing image correction, and grayscale processing the corrected on-site monitoring image to obtain a grayscale image; According to the preset stitching order, all grayscale images collected at the same time are stitched together to obtain a monitoring stitching image, and the images are sorted based on the time sequence.

[0007] Preferably, the steps of performing portrait recognition on the surveillance spliced ​​image by using a preset face recognition model, extracting the portrait data contained therein, comparing all the portrait data, and counting the total number of the portrait data specifically include: Import the surveillance stitching images into the face recognition model in chronological order, and use the face recognition model to identify the position of the portrait in the image; According to the position of the portrait, extract the facial features data, extract the facial features position data and facial features pixel data, and obtain the portrait data; All the portrait data obtained are compared, duplicate portrait data are eliminated, and the total number of remaining portrait data is counted.

[0008] Preferably, the steps of obtaining an exhibition area map, counting the total area in the exhibition area map, calculating the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculating the number of personnel based on the monitoring area ratio specifically include: Obtain an exhibition area map, and count the total area of ​​the exhibition area map according to a preset scale; Identify the reference objects in the on-site monitoring image, determine the scale of the image based on the number of pixels occupied by the reference objects, and calculate the area of ​​each monitoring area; The ratio of the regional area to the total area is used as the proportion of the monitored area, and the total number of people in the exhibition activity is inferred based on the total number of portrait data in the monitored area.

[0009] Preferably, when the image acquisition devices are set, the monitoring areas between the image acquisition devices do not overlap.

[0010] Another object of the present invention is to provide a personnel counting system for exhibition activities, the system comprising: An image acquisition module is used to set up multiple image acquisition devices within the scope of the exhibition activities, and acquire on-site monitoring images through the image acquisition devices; An image stitching module is used to obtain all on-site monitoring images, stitch together the on-site monitoring images at the same time to obtain a monitoring stitching image, and arrange the monitoring stitching images in chronological order; A portrait data extraction module is used to perform portrait recognition on the surveillance spliced ​​image through a preset face recognition model, extract the portrait data contained therein, compare all portrait data, and count the total amount of portrait data; The people counting module is used to obtain the exhibition area map, count the total area in the exhibition area map, calculate the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculate the number of people based on the monitoring area ratio.

[0011] Preferably, the image stitching module includes: A data acquisition unit, used to acquire data from the image acquisition device at preset time intervals, and store the on-site monitoring images acquired at the same time together; An image and preprocessing unit is used to preprocess the on-site monitoring image, perform image correction, and perform grayscale processing on the corrected on-site monitoring image to obtain a grayscale image; The image sorting unit is used to splice all grayscale images collected at the same time according to a preset splicing order to obtain a monitoring spliced ​​image, and sort it based on the time sequence.

[0012] Preferably, the portrait data extraction module includes: A portrait recognition unit, used to import the surveillance spliced ​​images into the face recognition model in chronological order, and recognize the position of the portrait in the image through the face recognition model; A facial features data extraction unit, used to extract facial features data according to the position of the portrait, extract facial features position data and facial features pixel data, and obtain portrait data; The data sorting unit is used to compare all the obtained portrait data, eliminate duplicate portrait data, and count the total number of remaining portrait data.

[0013] Preferably, the headcount counting module includes: A total area statistics unit is used to obtain an exhibition area map and to count the total area included in the exhibition area map according to a preset scale; The area statistics unit is used to identify the reference objects in the on-site monitoring image, determine the scale of the image according to the number of pixels occupied by the reference objects, and calculate the area of ​​each monitoring area; The exhibitor number calculation unit is used to use the ratio of the regional area to the total area as the monitoring area ratio, and reversely calculate the total number of exhibition activities based on the total number of portrait data in the monitoring area.

[0014] Preferably, when the image acquisition devices are set, the monitoring areas between the image acquisition devices do not overlap.

[0015] The present invention provides a method for counting people for exhibition activities. A plurality of image acquisition devices are arranged within the scope of the exhibition activities. The image acquisition devices are used to acquire images of the scene. The portrait data contained in the images are actively recognized by portrait recognition technology. The portrait data are counted to calculate the number of exhibitors within the entire scope of the exhibition activities. The method realizes automatic counting of the number of exhibitors and protects the privacy of the exhibitors. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of a method for counting people at an exhibition event provided by an embodiment of the present invention; Figure 2 A flowchart of the steps of obtaining all on-site monitoring images, splicing the on-site monitoring images at the same time together to obtain a monitoring spliced ​​image, and arranging the monitoring spliced ​​images in chronological order provided in an embodiment of the present invention; Figure 3 A flowchart of the steps of performing portrait recognition on a surveillance spliced ​​image using a preset face recognition model, extracting the portrait data contained therein, comparing all the portrait data, and counting the total number of portrait data provided by an embodiment of the present invention; Figure 4 A flowchart of the steps of obtaining an exhibition area map, counting the total area in the exhibition area map, calculating the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculating the number of people based on the monitoring area ratio provided in an embodiment of the present invention; Figure 5 An architectural diagram of a personnel counting system for exhibition activities provided by an embodiment of the present invention; Figure 6 An architecture diagram of an image stitching module provided by an embodiment of the present invention; Figure 7 An architecture diagram of a portrait data extraction module provided by an embodiment of the present invention; Figure 8 An architecture diagram of a people counting module provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] like Figure 1FIG. 1 is a flowchart of a method for counting people in an exhibition activity provided by an embodiment of the present invention, the method comprising: S100, multiple image acquisition devices are set up within the scope of the exhibition activity, and on-site monitoring images are acquired by the image acquisition devices.

[0019] In this step, multiple image acquisition devices are set within the exhibition activity range. The image acquisition devices can be installed in multiple positions, but the monitoring areas of the image acquisition devices do not overlap. The angle between the acquisition angle of the image acquisition device and the horizontal plane is between -30° and 0°. The number of image acquisition devices is at least 3. The image acquisition devices should be installed in the edge area of ​​the exhibition activity range so that images can be acquired from the edge position. The image acquisition devices acquire images at preset time intervals, such as acquiring a set of on-site monitoring images every one minute. Smaller intervals can also be used. The time interval for image acquisition is set according to the flow of people on site. The larger the flow of people, the smaller the time interval, and the smaller the flow of people, the larger the time interval.

[0020] S200, acquiring all on-site surveillance images, splicing the on-site surveillance images at the same time together to obtain a surveillance splicing image, and arranging the surveillance splicing images in chronological order.

[0021] In this step, all the on-site surveillance images are acquired. The on-site surveillance images are acquired at the end of the exhibition activity or at intervals. If the on-site surveillance images are acquired at the end of the exhibition activity, a large number of on-site surveillance images need to be stored. Therefore, the images can be acquired at intervals. After the acquired on-site surveillance images are processed, the above-mentioned on-site surveillance images are deleted to reduce the amount of data storage and reduce costs. The on-site surveillance images are extracted in chronological order, and the on-site surveillance images generated at the same time are divided into a group. The on-site surveillance images belonging to the same group are spliced ​​into a whole to facilitate batch processing. After the splicing is completed, a series of surveillance splicing images will be obtained, which are arranged in chronological order. In the subsequent processing process, the surveillance splicing images are processed in chronological order.

[0022] S300, performing portrait recognition on the surveillance spliced ​​image using a preset face recognition model, extracting the portrait data contained therein, comparing all the portrait data, and counting the total amount of the portrait data.

[0023] In this step, portrait recognition is performed on the surveillance spliced ​​image through a preset face recognition model. In the present invention, the face recognition model is used to identify the position of the portrait in the picture, and does not directly identify the identity of the portrait. When a portrait is recognized in the picture, the corresponding portrait data is recorded and stored as portrait data. Since different people have different portrait data, when performing statistics, the portrait data can be compared to determine whether two groups of portrait data belong to the same person, thereby performing deduplication processing. After the deduplication processing is completed, the number of remaining portrait data is the total number of people appearing in the surveillance picture obtained by statistics.

[0024] S400, obtaining an exhibition area map, counting the total area in the exhibition area map, calculating the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculating the number of people based on the monitoring area ratio.

[0025] In this step, an exhibition area map is obtained. The exhibition area map records the on-site layout of the current exhibition activity. The total area of ​​the current exhibition activity can be known based on the exhibition area map. Different image acquisition devices can monitor different areas. The areas of the monitoring areas of all image acquisition devices are counted to determine the ratio of the monitored area to the total area. The total number of people in the entire exhibition area can be inferred based on the number of portrait data within the current monitoring range and the ratio of the monitored area to the total area.

[0026] like Figure 2 As a preferred embodiment of the present invention, the steps of acquiring all the on-site surveillance images, splicing the on-site surveillance images at the same time together to obtain a surveillance spliced ​​image, and arranging the surveillance spliced ​​images in chronological order specifically include: S201, acquiring data from an image acquisition device at preset time intervals, and storing together on-site monitoring images acquired at the same time.

[0027] In this step, data is acquired from the image acquisition device at a preset time interval, which is set to ten minutes. The field surveillance images stored in the image acquisition device are acquired once every ten minutes. Then, the field surveillance images generated at the same time are stored together. For example, if the three field surveillance images A, B and C are all generated at time K, then the above three groups of field surveillance images are stored together.

[0028] S202, pre-processing the on-site monitoring image, performing image correction, and gray-scale processing on the corrected on-site monitoring image to obtain a gray-scale image.

[0029] In this step, the on-site monitoring image is preprocessed. Since the present invention only needs to identify whether the picture contains a human portrait and the position of the human portrait, a grayscale image can be used for identification. Then, in the preprocessing process, the on-site monitoring image is grayscale processed to reduce the amount of data contained in the on-site monitoring image. Since the edges of the on-site monitoring image captured by the image acquisition device will have a certain degree of distortion, the distortion is corrected through image processing to obtain a distortion-free grayscale image.

[0030] S203, stitching all grayscale images collected at the same time in a preset stitching order to obtain a monitoring stitching image, and sorting the images based on the time sequence.

[0031] In this step, a group of grayscale images are retrieved each time according to the preset stitching order, and this group of grayscale images are stitched in a preset order. For example, if the image acquisition devices are numbered 1, 2, and 3 respectively, the grayscale images are stitched in the order of the numbers. After the stitching is completed, a group of monitoring stitched images are obtained. The monitoring stitched images are named according to the acquisition time of the on-site monitoring images and are sorted based on the chronological order.

[0032] like Figure 3 As shown, as a preferred embodiment of the present invention, the steps of performing portrait recognition on the surveillance spliced ​​image by using a preset face recognition model, extracting the portrait data contained therein, comparing all the portrait data, and counting the total number of portrait data specifically include: S301, importing the surveillance stitching image into the face recognition model in chronological order, and identifying the position of the portrait in the image through the face recognition model.

[0033] In this step, the surveillance stitching image is imported into the face recognition model in chronological order. Specifically, the face recognition model can adopt the SSD target detection algorithm, take the surveillance stitching image as the input image, adjust the input image to a preset size, and annotate it to ensure the consistency of the input data. The features of the input image are extracted by a pre-trained convolutional neural network to generate a series of feature maps of different scales. At the position of each feature map, multiple default boxes are defined. The default box is a rectangular box with different proportions and aspect ratios. The default box covers different areas of the entire image. For each default box, the offset relative to the default box (that is, the difference between the real bounding box and the default box) and the probability of containing a specific person in the box are predicted, thereby determining the position of the portrait.

[0034] S302, extracting facial features data according to the position of the portrait, extracting facial features position data and facial features pixel data, and obtaining portrait data.

[0035] In this step, facial features data is extracted based on the position of the portrait. An image of each portrait position is captured from the screen. A perspective transformation is performed to obtain a frontal portrait image, and the facial features corresponding to the portrait are identified, including eyes, nose, mouth and ears. The center positions of the eyes, nose, mouth and ears are identified, and the coordinates of each center position are determined by constructing a coordinate system. The distance between any two points can then be calculated, and the coordinates of the above center positions are stored as portrait data.

[0036] S303: compare all the obtained portrait data, remove duplicate portrait data, and count the total number of remaining portrait data.

[0037] In this step, all the portrait data obtained are compared. If the two groups of portrait data are the same, it means that the above portrait data are from the same person, so one of the two groups of portrait data is eliminated, and the portrait data retained through deduplication processing is the number of people who appear in the monitoring range during the entire exhibition activity.

[0038] like Figure 4 As shown, as a preferred embodiment of the present invention, the steps of obtaining an exhibition area map, counting the total area in the exhibition area map, calculating the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculating the number of people based on the monitoring area ratio specifically include: S401, obtaining an exhibition area map, and counting the total area included in the exhibition area map according to a preset scale.

[0039] In this step, an exhibition area map is obtained. The exhibition area map is used to record the layout of the exhibition activities. When calculating the total area, only the area within the exhibition area that exhibitors are allowed to enter is counted. Locations that exhibitors cannot enter are subtracted from the total area.

[0040] S402, identifying the reference object in the on-site monitoring image, determining the scale of the image according to the number of pixels occupied by the reference object, and calculating the area of ​​each monitoring area.

[0041] In this step, the reference objects in the exhibition area map are identified. In order to facilitate the identification of the scope of the current monitoring area, reference objects can be set in the exhibition area map, such as spherical reference objects of known size. According to the number of pixels occupied by the reference objects in the picture, the size corresponding to each pixel is determined, thereby identifying the area opposite the monitoring area. The area of ​​the monitoring area can also be obtained through manual input to determine the area of ​​each monitoring area.

[0042] S403: Taking the ratio of the area of ​​the region to the total area as the ratio of the monitored area, the total number of people in the exhibition activity is inferred from the total number of portrait data in the monitored area.

[0043] In this step, the ratio of the area S1 to the total area S0 is used as the monitoring area ratio, then the number of exhibitors in all monitoring areas can be obtained, that is, Q0, and the total number of exhibitors in the entire exhibition activity is Q0 / (S1 / S0). Through the personnel counting method provided by the present invention, the number of exhibitors can be automatically estimated based on the set monitoring equipment.

[0044] like Figure 5 As shown, a personnel counting system for exhibition activities provided by an embodiment of the present invention includes: The image acquisition module 100 is used to set up multiple image acquisition devices within the scope of the exhibition activity, and acquire on-site monitoring images through the image acquisition devices.

[0045] In the present system, the image acquisition module 100 sets a plurality of image acquisition devices within the scope of the exhibition activity. The image acquisition devices can be installed at a plurality of positions, but the monitoring areas of the image acquisition devices do not overlap. The angle between the acquisition angle of the image acquisition device and the horizontal plane is between -30° and 0°. The number of image acquisition devices is at least 3. The image acquisition devices should be installed at the edge area of ​​the exhibition activity scope so that images can be acquired from the edge position. The image acquisition devices acquire images at preset time intervals, such as acquiring a set of on-site monitoring images every one minute. Smaller intervals can also be used. The time interval for image acquisition is set according to the flow of people on site. The larger the flow of people, the smaller the time interval, and the smaller the flow of people, the larger the time interval.

[0046] The image stitching module 200 is used to obtain all the on-site monitoring images, stitch together the on-site monitoring images at the same time to obtain a monitoring stitching image, and arrange the monitoring stitching images in chronological order.

[0047] In this system, the image stitching module 200 acquires all the on-site surveillance images, and acquires the on-site surveillance images once at the end of the exhibition activity or at a certain interval. If the on-site surveillance images are acquired at the end of the exhibition activity, a large number of on-site surveillance images need to be stored. Therefore, the images can be collected once at a certain interval, and after the collected on-site surveillance images are processed, the above-mentioned on-site surveillance images are deleted to reduce the data storage volume and reduce the cost. The on-site surveillance images are extracted in chronological order, and the on-site surveillance images generated at the same time are divided into a group. The on-site surveillance images belonging to the same group are stitched into a whole to facilitate batch processing. After the stitching is completed, a series of surveillance stitching images will be obtained, which are arranged in chronological order. In the subsequent processing process, the surveillance stitching images are processed in chronological order.

[0048] The portrait data extraction module 300 is used to perform portrait recognition on the surveillance spliced ​​image through a preset face recognition model, extract the portrait data contained therein, compare all the portrait data, and count the total amount of portrait data.

[0049] In this system, the portrait data extraction module 300 performs portrait recognition on the surveillance spliced ​​image through a preset face recognition model. In the present invention, the face recognition model is used to identify the position of the portrait in the picture, and does not directly identify the identity of the portrait. When a portrait is recognized in the picture, the corresponding portrait data is recorded and stored as portrait data. Since different people have different portrait data, when performing statistics, the portrait data can be compared to determine whether two groups of portrait data belong to the same person, thereby performing deduplication processing. After the deduplication processing is completed, the number of remaining portrait data is the total number of people appearing in the surveillance picture obtained by statistics.

[0050] The people counting module 400 is used to obtain an exhibition area map, count the total area in the exhibition area map, calculate the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculate the number of people based on the monitoring area ratio.

[0051] In this system, the crowd counting module 400 obtains the exhibition area map, which records the on-site layout of the current exhibition activity. The total area of ​​the current exhibition activity can be known based on the exhibition area map. Different image acquisition devices can monitor different areas. The areas of the monitoring areas of all image acquisition devices are counted to determine the ratio of the monitored area to the total area. The total number of people in the entire exhibition area can be inferred based on the number of portrait data within the current monitoring range and the ratio of the monitored area to the total area.

[0052] like Figure 6As shown, as a preferred embodiment of the present invention, the image stitching module 200 includes: The data acquisition unit 201 is used to acquire data from the image acquisition device at preset time intervals, and store the on-site monitoring images acquired at the same time together.

[0053] In this module, the data acquisition unit 201 acquires data from the image acquisition device at a preset time interval. The preset time interval is set to ten minutes. The field monitoring images stored in the image acquisition device are acquired once every ten minutes. Then the field monitoring images generated at the same time are stored together. For example, if the three field monitoring images A, B and C are all generated at time K, then the above three groups of field monitoring images are stored together.

[0054] The image and preprocessing unit 202 is used to preprocess the on-site monitoring image, perform image correction, and perform grayscale processing on the corrected on-site monitoring image to obtain a grayscale image.

[0055] In this module, the image and preprocessing unit 202 preprocesses the on-site monitoring image. Since the present invention only needs to identify whether the picture contains a portrait and the position of the portrait, a grayscale image can be used for identification. Then, during the preprocessing process, the on-site monitoring image is grayscale processed to reduce the amount of data contained in the on-site monitoring image. Since the edges of the on-site monitoring image captured by the image acquisition device will have a certain degree of distortion, the distortion is corrected through image processing to obtain a distortion-free grayscale image.

[0056] The image sorting unit 203 is used to splice all grayscale images collected at the same time according to a preset splicing order to obtain a monitoring spliced ​​image, and sort the spliced ​​images based on the time sequence.

[0057] In this module, the image sorting unit 203 retrieves a group of grayscale images each time according to a preset stitching order, and stitches the group of grayscale images in a preset order. For example, if the image acquisition devices are numbered 1, 2, and 3, the grayscale images are stitched in the order of the numbers. After the stitching is completed, a group of monitoring stitched images is obtained. The monitoring stitched images are named according to the acquisition time of the on-site monitoring images and are sorted based on the time sequence.

[0058] like Figure 7 As shown, as a preferred embodiment of the present invention, the portrait data extraction module 300 includes: The portrait recognition unit 301 is used to import the surveillance spliced ​​images into the face recognition model in chronological order, and recognize the position of the portrait in the image through the face recognition model.

[0059] In this module, the portrait recognition unit 301 imports the surveillance stitching image into the face recognition model in chronological order. Specifically, the face recognition model can adopt the SSD target detection algorithm, take the surveillance stitching image as the input image, adjust the input image to a preset size, and annotate it to ensure the consistency of the input data. The features of the input image are extracted through a pre-trained convolutional neural network to generate a series of feature maps of different scales. At the position of each feature map, multiple default boxes are defined. The default box is a rectangular box with different proportions and aspect ratios. The default box covers different areas of the entire image. For each default box, the offset relative to the default box (that is, the difference between the real bounding box and the default box) and the probability that a specific person is contained in the box are predicted, thereby determining the position of the portrait.

[0060] The facial feature data extraction unit 302 is used to extract the facial feature data according to the position of the portrait, extract the facial feature position data and the facial feature pixel data, and obtain the portrait data.

[0061] In this module, the facial features data extraction unit 302 extracts facial features data according to the position of the portrait, captures the image of each portrait position from the screen, performs perspective transformation to obtain a frontal portrait image, and identifies the facial features corresponding to the portrait, including eyes, nose, mouth and ears, identifies the center positions of the eyes, nose, mouth and ears, and determines the coordinates of each center position by constructing a coordinate system. Then, the distance between any two points can be calculated and stored as the portrait data with the coordinates of the above center positions.

[0062] The data sorting unit 303 is used to compare all the obtained portrait data, eliminate duplicate portrait data, and count the total number of remaining portrait data.

[0063] In this module, the data sorting unit 303 compares all the portrait data obtained. If the two groups of portrait data are the same, it indicates that the above portrait data are from the same person, so one of the two groups of portrait data is eliminated, and the retained portrait data is the number of people appearing in the monitoring range in the entire exhibition activity through deduplication processing.

[0064] like Figure 8 As shown, as a preferred embodiment of the present invention, the people counting module 400 includes: The total area counting unit 401 is used to obtain an exhibition area map and count the total area included in the exhibition area map according to a preset scale.

[0065] In this module, the total area statistics unit 401 obtains the exhibition area map, which is used to record the layout of the exhibition activities. When counting the total area, only the area within the exhibition area that exhibitors are allowed to enter is counted, and the areas that exhibitors cannot enter are subtracted from the total area.

[0066] The area statistics unit 402 is used to identify the reference objects in the on-site monitoring image, determine the scale of the picture according to the number of pixels occupied by the reference objects, and calculate the area of ​​each monitoring area.

[0067] In this module, the regional area statistics unit 402 identifies the reference objects in the exhibition area map. In order to facilitate the identification of the scope of the current monitoring area, reference objects can be set in the exhibition area map, such as setting a spherical reference object of known size. According to the number of pixels occupied by the reference object in the picture, the size corresponding to each pixel is determined, thereby identifying the area opposite the monitoring area. The area of ​​the monitoring area can also be obtained through manual input to determine the regional area of ​​each monitoring area.

[0068] The exhibition attendance calculation unit 403 is used to use the ratio of the area of ​​the region to the total area as the monitoring area ratio, and to infer the total number of people attending the exhibition activity according to the total number of portrait data in the monitoring area.

[0069] In this module, the exhibitor number calculation unit 403 uses the ratio of the regional area S1 to the total area S0 as the monitoring area ratio, so the number of exhibitors in all monitoring areas can be obtained, that is, Q0, and the total number of exhibitors in the entire exhibition activity is Q0 / (S1 / S0). Through the personnel counting method provided by the present invention, the number of exhibitors can be automatically estimated based on the set monitoring equipment.

[0070] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for counting people in exhibition activities, characterized in that: The method comprises: Multiple image acquisition devices are set up within the scope of the exhibition activities, and on-site monitoring images are acquired through the image acquisition devices; Acquire all on-site surveillance images, stitch together the on-site surveillance images at the same time to obtain a surveillance stitching image, and arrange the surveillance stitching images in chronological order; Use the preset face recognition model to perform portrait recognition on the surveillance spliced ​​images, extract the portrait data contained therein, compare all the portrait data, and count the total number of portrait data; The exhibition area map is obtained, the total area in the exhibition area map is counted, and the monitoring area ratio of the monitoring area corresponding to the image acquisition device is calculated, and the number of people is calculated based on the monitoring area ratio.

2. The method for counting people in exhibition activities according to claim 1, characterized in that: The step of acquiring all the on-site monitoring images, splicing the on-site monitoring images at the same time together to obtain a monitoring spliced ​​image, and arranging the monitoring spliced ​​images in chronological order specifically includes: Acquire data from the image acquisition device at preset time intervals, and store the on-site monitoring images acquired at the same time together; Preprocessing the on-site monitoring image, performing image correction, and grayscale processing the corrected on-site monitoring image to obtain a grayscale image; According to the preset stitching order, all grayscale images collected at the same time are stitched together to obtain a monitoring stitching image, and the images are sorted based on the time sequence.

3. The method for counting people in exhibition activities according to claim 1, characterized in that: The steps of performing portrait recognition on the surveillance spliced ​​image by using a preset face recognition model, extracting the portrait data contained therein, comparing all the portrait data, and counting the total amount of the portrait data specifically include: Import the surveillance stitching images into the face recognition model in chronological order, and use the face recognition model to identify the position of the portrait in the image; According to the position of the portrait, extract the facial features data, extract the facial features position data and facial features pixel data, and obtain the portrait data; All the portrait data obtained are compared, duplicate portrait data are eliminated, and the total number of remaining portrait data is counted.

4. The method for counting people in exhibition activities according to claim 1, characterized in that: The steps of obtaining an exhibition area map, counting the total area in the exhibition area map, calculating the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculating the number of personnel based on the monitoring area ratio specifically include: Obtain an exhibition area map, and count the total area of ​​the exhibition area map according to a preset scale; Identify the reference objects in the on-site monitoring image, determine the scale of the image based on the number of pixels occupied by the reference objects, and calculate the area of ​​each monitoring area; The ratio of the regional area to the total area is used as the proportion of the monitored area, and the total number of people in the exhibition activity is inferred based on the total number of portrait data in the monitored area.

5. The method for counting people in exhibition activities according to claim 1, characterized in that: When setting up the image acquisition devices, the monitoring areas between the image acquisition devices do not overlap.

6. A personnel counting system for exhibition activities, characterized in that: The system comprises: An image acquisition module is used to set up multiple image acquisition devices within the scope of the exhibition activities, and acquire on-site monitoring images through the image acquisition devices; An image stitching module is used to obtain all on-site monitoring images, stitch together the on-site monitoring images at the same time to obtain a monitoring stitching image, and arrange the monitoring stitching images in chronological order; A portrait data extraction module is used to perform portrait recognition on the surveillance spliced ​​image through a preset face recognition model, extract the portrait data contained therein, compare all portrait data, and count the total amount of portrait data; The people counting module is used to obtain the exhibition area map, count the total area in the exhibition area map, calculate the monitoring area ratio of the monitoring area corresponding to the image acquisition device, and calculate the number of people based on the monitoring area ratio.

7. The personnel counting system for exhibition activities according to claim 6, characterized in that: The image stitching module comprises: A data acquisition unit, used to acquire data from the image acquisition device at preset time intervals, and store the on-site monitoring images acquired at the same time together; An image and preprocessing unit is used to preprocess the on-site monitoring image, perform image correction, and perform grayscale processing on the corrected on-site monitoring image to obtain a grayscale image; The image sorting unit is used to splice all grayscale images collected at the same time according to a preset splicing order to obtain a monitoring spliced ​​image, and sort it based on the time sequence.

8. The personnel counting system for exhibition activities according to claim 6, characterized in that: The portrait data extraction module comprises: A portrait recognition unit, used to import the surveillance spliced ​​images into the face recognition model in chronological order, and recognize the position of the portrait in the image through the face recognition model; A facial features data extraction unit, used to extract facial features data according to the position of the portrait, extract facial features position data and facial features pixel data, and obtain portrait data; The data sorting unit is used to compare all the obtained portrait data, eliminate duplicate portrait data, and count the total number of remaining portrait data.

9. The personnel counting system for exhibition activities according to claim 6, characterized in that: The people counting module includes: A total area statistics unit is used to obtain an exhibition area map and to count the total area included in the exhibition area map according to a preset scale; The area statistics unit is used to identify the reference objects in the on-site monitoring image, determine the scale of the image according to the number of pixels occupied by the reference objects, and calculate the area of ​​each monitoring area; The exhibitor number calculation unit is used to use the ratio of the regional area to the total area as the monitoring area ratio, and reversely calculate the total number of exhibition activities based on the total number of portrait data in the monitoring area.

10. The personnel counting system for exhibition activities according to claim 6, characterized in that: When setting up the image acquisition devices, the monitoring areas between the image acquisition devices do not overlap.

Images