Crowd density monitoring systems and methods

Abstract

Crowd density monitoring systems and methods arc disclosed, comprising a computer vision density monitoring system comprising a camera and a vision processing module adapted to count persons in a given area of in real time during an event. Information from the computer vision processing module and secondary crowd density and crowd counting sensors are displayed on a density monitoring station adapted to issue alerts in the event density or counts exceed a predetermined threshold.

Description

TITLE: Crowd Density Monitoring Systems and MethodsCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of, and priority to, U.S. Provisional Application Serial No. 63 / 517,833, filed August 4th, 2023, which is hereby incorporated by reference in its entirety for all that it contains (including all references therein) for all purposes as if restated and set forth fully herein to the maximum extent allowable by law.BACKGROUND

[0002] The present invention relates to systems and methods of monitoring crowd density at events such as musical concerts, theatrical shows, rallies, pilgrimages, and any other location where large numbers of people are free to move in a given space. When people gather in such spaces, there may be a tendency for people to push toward a specific location. For example, at a music festival, when a performer comes onto the stage, the members of the audience may begin to push forward to get closer to the stage, or at a political rally, people may push forward to get closer to a speaker. At the beginning of the performance, people may also push to get through an entranceway in hopes of obtaining a better location for viewing the event, and at the end people may push to get to an exit. Wherever a large number of people move toward a given location, people who are closer to that location can get crowded as the density of persons in that space increases. There have been multiple tragic occurrences in such situations where audience members have been suffocated, trampled, or otherwise harmed as a result of such movements.

[0003] In an effort to prevent such tragedies, law enforcement and security personnel at such venues try to look for areas of high or growing crowd density in hopes of intervening before the density becomes so high as to become dangerous. Relying on individuals on the ground or live camera feeds, however, in hopes that people will notice and predict areas of rapidly growing density in real time, has not been a reliable solution, particularly where there is a large crowd in place and locations of high density are changing over time.

[0004] The present invention provides systems and methods for using a variety of sensors to monitor crowd density in real time and identify times of higher risk and notify safety personnel so as to enable intervention prior to injury occurring. Such sensors may also track the flow of persons at an event, thereby providing early indications of potential density risks and providing analytics data about how crowds behave at the venue. In one embodiment, a computer vision density monitoring system uses a camera adapted to operate in the lighting conditions commonly present in concert venues linked to a computer vision processing module adapted to count persons appearing in a pre-determined area of interest in the camera image. The computer vision processing module is connected to a density monitoring station that allows an operator to monitor the fluctuations in crowd density in the area of interest throughout the event. In alternative embodiments, additional sensor types are added and combined with the computer vision data. The resulting systems and methods thereby allow centralized security personnel to monitor crowd density changes in real time and direct roving security or safety personnel to those areas where density appears likely to exceed safety thresholds, in hopes of intervening before density becomes dangerous. The use of multiple sensor types allows for improved count qualities and tracking of flow of persons within the venue. Moreover, when multiple sensors are used within the area of interest, the computer vision density monitoring system can be used to calibrate the secondarydensity monitoring systems, and the data from the disparate systems can be used to increase confidence.

[0005] With the ability to leverage density and flow data, venue personnel may make decisions and address safety concerns in a more timely fashion, especially if serious injury or loss of life or limb may be imminent. This can be done via any number of established emergency standard operating procedures or other techniques known to persons having ordinary skill in the arts of crowd science and / or entertainment venue operations. Examples include modifying the intensity, duration, and / or location of one or more performances to direct and control crowd behavior; cutting sound entirely; taking control of lighting; automating public address messaging or by geo-locating attendee phones and sending communication messages to specific recipients at selected locations; and / or releasing enclosure locks to relieve pressure; as appropriate.BRIEF SUMMARY OF THE INVENTION

[0006] One embodiment of the present invention comprises at least one camera mounted above and adapted to view an area of interest where crowd density is expected to be high. A computer vision processing module receives image data from the camera and adaptively calculates the density of persons within the predetermined area of interest and transmits camera images and density information to a density monitoring station adapted to issue alerts when the density values approach or exceed pre-determined threshold values. At least one person-counter is optionally provided at an entrance to a venue or location to provide overall attendance counts in addition to density information and alerts. The data from the person-counter and the computer vision processing module are displayed on the density monitoring system display to enable an operator to monitor density in real time. The system may optionally be enhanced by incorporatingadditional computer vision density monitoring systems or secondary density monitoring systems such as LIDAR density monitoring systems, radio frequency attenuation density monitoring systems, radio frequency density monitoring systems, and additional people counters. Advertising ID data may also be incorporated as an additional source of data about attendees. The accuracy and setup of the secondary density monitoring systems may optionally be improved by providing configuration data from the computer vision density monitoring system.

[0007] A method of configuring a computer vision density monitoring system is also provided. Markers or mats are used to determine the boundaries of the area of interest within the camera view. Pixel distances are then correlated with physical area measurements to determine the physical area of the area of interest for density calculations.

[0008] Other features of the present invention will be apparent to persons having ordinary skill in the art in light of this disclosure and description.BRIEF DESCRIPTION OF THE DRAWINGS

[0009] For purpose of explanation, aspects of several embodiments are depicted in the following illustrative figures, wherein:

[0010] Fig. 1 is a schematic view of an exemplary venue in which crowd counting and crowd density monitoring sensors are deployed;

[0011] Fig. 2 is a block diagram view of an embodiment of a multi-sensor, crowd density monitoring system according to the disclosure;

[0012] Fig. 3 is a flow chart representation of an embodiment of a method for configuring a computer vision processing module and camera as part of a crowd density monitoring system according to the present disclosure;

[0013] Fig. 4 is a flow chart representation of an alternate embodiment of a method for configuring a computer vision processing module and camera as part of a crowd density monitoring system according to the present disclosure; and

[0014] Fig. 5 is a flow chart representation of a further embodiment of a method for configuring a computer vision processing module and camera as part of a crowd density monitoring system according to the present disclosure.DETAILED DESCRIPTION

[0015] In the disclosure herein, details are set forth for the purpose of description. However, a person having ordinary skill in the art will realize the systems and methods disclosed herein may be practiced without all the specifics described herein. The embodiments and explanations are intended, therefore, to be illustrative only, and not limiting. Similarly, where examples are used, they are not intended to be limiting unless the context clearly indicates otherwise. Accordingly, “for example” or “e.g.” should be read as “for example, and without limitation,” and “including” should be read as “including, but without limitation,” unless the context clearly indicates limitation is intended.

[0016] As used herein, “adapted” means: adjusted, arranged, automatized, calibrated, configured, connected, constructed, coordinated, designed, directed, disposed, established, fashioned, formatted, formed, instructed, modified, mechanized, modulated, organized, positioned, programmed, regulated, set up, shaped, sized, structured, and tailored, as appropriate to render an item suitable for use with another item, step, or apparatus. The terms “adapted,” “adaption,” “communicable,” “communicably,” “connected,” “connection,” “configured,” “operable,”“operably,” and “operate” are intended as indications of structure, and not function.

[0017] “LIDAR” means light detection and ranging and includes devices capable of performing light detection and ranging.

[0018] “RF” means radiofrequency or radio frequency and, depending on context, may include protocols and standards adapted to facilitate communication among devices using radio frequencies, or devices adapted to send or receive data using radio frequency signals.

[0019] “Venue” means an area where persons group or gather to participate in an event or observe a performance. Examples of venues include general admission concert floors and standing areas (whether inside or outside), stadiums, theaters, parks, markets, fairgrounds, sporting event locations, public locations where persons gather for an event including rallies or religious pilgrimages or demonstrations or store openings, and concert halls. Venues may have some areas where persons are seated, and others where persons may stand and / or move about more freely. Venues may also include multiple stages or performance areas, as well as concession areas, open- air markets, sites where rally or protest speeches are occurring, limited entry VIP areas, store entrance gathering areas, etc.

[0020] The meaning of other terms may be defined herein explicitly or will be apparent to persons having ordinary skill in the arts of crowd density management, venue management, software development, network engineering, computer vision, or electrical engineering.

[0021] Referring to FIG. 1, a schematic view of a venue 10 is shown with an embodiment of a system according to the present disclosure installed. Venue 10 is an exemplary venue and, in this case, a venue suitable for a music concert in which spectators stand and move about in spectator area 20 which is “general admission” without assigned seating. As will become apparent, the present disclosure is not limited to any specific venue or venue configuration. The configuration of venue 10 is merely used as an example.

[0022] Venue 10 comprises a venue front area 22 and a venue rear area 23. Stage 40 is separated from spectator area 20 by barrier 44 (typically a fence or similar barrier). The area between barrier 44 and stage 40 becomes a walkway 46 where security personnel 52 may be positioned, in addition to roaming in other areas of the venue 10. It is notable that, while venue 10 as illustrated comprises a single stage and set of entrances, venues may be much larger, comprising multiple stages, concession areas, venue entrance checkpoints, choke points, and the like. Herein, the term venue is not intended to be limited only to the area of a particular stage 40 but is instead intended to also encompass multi-stage venues and venues comprising areas in addition to spectator areas 20, it being understood that determining overall attendance at a venue and flow of persons among different areas of a venue can be desirable. For the purposes of this exemplary embodiment, spectator area 20 is assumed to comprise at least some general admission area where spectators are permitted to move freely and are not limited to assigned seats or individual standing areas. It is in such general admission areas, and in other areas where spectators do not have assigned locations, that crowd density risks tend to arise more often. In addition to general admission viewing areas, other areas where density risks may arise include avenues of ingress and egress, ticket or concession purchase lines, recreation and rest facilities, stairwells, hallways, and areas around event features such as live performers or sculptures or displays that are supplemental to the main attraction.

[0023] As illustrated, spectators gain access to spectator area 20 through entrance doorways 32 and larger open entrance 30. Entrance doorways 32 are smaller and adapted to allow one to fifteen people to pass through in one direction at a time. Entrance doorway 32 in VIP area 28 provides and controls access to that area. It will be understood that the particular configuration of each venue may vary based on characteristics of the venue and particular events taking place there. Thesystems and methods described herein are not intended to be limited to any particular venue configuration. It will also be understood that, while stage 40 in a performance venue will typically be a raised stage at venue front area 22, stages on the same level or below spectator area 20, or stages more centrally located in spectator area 20 are also possible. As described, stage 40 is just one example of a feature that may draw people to collect in high densities, and the present disclosure is not intended to be limited to areas of interest near concert stages.

[0024] As was noted above, open entrance 30 is a larger opening that allows many spectators to pass through at a time and may allow for a bi-directional flow in which some persons are entering while others are exiting. In a pavilion (not illustrated) where spectators approach from multiple sides, large entrance(s) 30 would be, for example, the open areas between the pillars supporting the pavilion roof through which spectators may pass.

[0025] Because of their smaller areas, doorway entrances 32 can be equipped with optical, weightbased, turnstile, rotating door, Near-Field Communication (NFC) or Personal Area Network (PAN) or Wireless Fidelity (WiFi) receivers, or other types of person counters. A person counter system may comprise at least one person-counter sensor 602 adapted to count persons entering or exiting a particular doorway entrance 32. In this context, humans at the doorway or cameras being viewed by humans may also be used as person-counting sensors provided the counts are maintained in real time and available for continuous transmission to security personnel. Additional options for person-counter sensors 602 include human or machine operated tally counters and / or people counters; LIDAR, radio frequency detectors, and / or 3D imaging sensors.

[0026] Person-counter sensors 602 are adapted to count the number of individuals passing through doorway entrances 32, into entry a spectator area 20, typically (but not necessarily) assuming that all traffic is predominantly going in a single direction at a given time. They do not, however,provide any data regarding where and how spectators move within spectator area 20, with one exception — speed of movement may indicate density in the area just beyond doorway entrances 32, with a fast speed of movement indicating a low density in such areas, and a slow speed of movement indicating a higher density in such areas. Instead, person-counter sensors 602 are more often used to monitor overall attendance counts or overall counts for a given area (e.g. VIP area 28). Density risks are assumed to increase as spectator area 20 approaches anticipated capacity or predetermined expected occupancy levels. Thus, overall attendance data from person-counter sensors 602 may be used as early warning indicators or may be used to adjust threshold values over time for density alerts discussed below.

[0027] Person-counter sensors 602 are not always suited to larger entrances such as open entrance 30 that are more open, may have much higher ceiling heights, and through which many spectators can pass at one time, in either direction. It will be understood, then, that in certain embodiments, person-counter sensors 602 may provide data at most regarding the number of persons that pass- through a given doorway entrance 32 and cannot provide overall attendance data or density data throughout spectator area 20, except through extrapolation. It should be understood, however, that knowing that a given number of persons entered through doorway entrances 32 can provide an early warning of density issues, even if it cannot provide a completely accurate overall attendance count. Such warnings may arise by monitoring the speed of flow through doorway entrances 32 or operating based on assumptions that after a given number of persons have passed through doorway entrances 32, spectator area 20 is likely to be reaching a given capacity level.

[0028] In certain preferred embodiments, density in a given area may be estimated based on wireless signals emitted and measured, based on the assumption that the more persons in the path of the wireless signal, the more attenuated the signal will be when it is received. Herein, suchsystems are referred to as radio frequency attenuation density monitoring systems and comprise a plurality of radio frequency attenuation transceivers 502 combined with a radio frequency attenuation processing module (not illustrated) adapted to correlate signal enfeeblement with density along a given signal path, and to combine path data to estimate overall density within the region covered by radio frequency attenuation transceivers 502. For example, in FIG. 1, radio frequency attenuation transceivers 502 may assist in determining crowd density and movement for spectators within the forward and middle regions of spectator area 20. While radio frequency attenuation transceivers can be deployed on as few as two sides, deployment on at least three sides is preferred. One example of radio frequency attenuation transceiver 502 are those made by Belgium company CrowdScan (see crowdscan.be), and in such example each sensor comprises an emitter and one or more receivers. Radio frequency attenuation transceivers 502 transmit a wireless signal (not illustrated) from one to the other and measure the signal strength on arrival at the receiving transceiver. The radio frequency density processing module (not illustrated) can then estimate person density on the paths between radio frequency attenuation transceivers 502 based on the fact that, as the number of humans in the signal propagation area increases, the strength of the signal will be increasingly enfeebled. By measuring the degree of enfeeblement, an estimate of the number of persons in the signal path may be calculated.

[0029] Notably, however, estimating crowd density from signal attenuation requires calibration as signal attenuation can be affected by venue characteristics, local conditions, and event characteristics, some of which may change over time. For this reason, it is typical to perform a manual calibration step prior to the commencement of the event with known numbers of persons standing in designated areas. Such calibration, however, can be performed more quickly and can be updated over time using data from the computer vision density monitoring system describedbelow. It will be understood, therefore that both the initial setup of radio frequency attenuation transceivers 502 and real-time operation of radio frequency attenuation transceivers 502 can be improved by the use a computer vision density monitoring system such as is described herein as a source of calibration data.

[0030] By having multiple radio frequency attenuation transceivers 502 arranged in a mesh pattern, lines of density can be compared to create a 2-dimensional map of the estimated density across spectator area 20, and that map can be overlaid on a schematic or photographic representation of venue 10. Ideally, radio frequency attenuation transceivers 502 will be arranged entirely around spectator area 20, or on at least two, and preferably at least three opposing sides. Data from radio frequency attenuation transceivers 502 may then be combined and displayed in the form of a heat-map superimposed on the schematic or photographic representation indicating crowd density across spectator area 20. Because of the number of units required to form a mesh, it may be preferred that the units be battery-powered and communicate with radio frequency attenuation processing module (not illustrated) wirelessly with the processor (not illustrated) being adapted to combine the data and produce a density heat map or density alerts based on predetermined density thresholds.

[0031] As is noted above, various factors impact enfeeblement and signal strength, including the distance the signal travels, weather conditions, presence and character of persons and other objects (whether trees, cars, etc.), among others. Given this variability, calibration is needed — ideally in an empty environment initially, such as at a venue prior to the event in the same weather conditions as are expected at the time of the event. Additional persons can be added, thereby training the radio frequency attenuation density monitoring system to discern which signal loss characteristics are associated with the addition of persons, and which are associated with other factors.Calibration can then be improved over time by comparing calculated densities with densities calculated by other means. Such calibration and training, however, is difficult and often in need of improvement, including because changes in signal enfeeblement characteristics are not always easy to predict, and the need for multiple stand-in people for use in calibration.

[0032] A further system for monitoring attendance and crowd density is through the use of a radio frequency density monitoring system. Radio frequency multi-band transceivers 652 may be used as additional sensors and may essentially operate as further exemplary counting means in one or more redundancy modules. Radio frequency multi-band transceivers 652 may conveniently be software-controlled radios adapted to scan and receive signals on various frequencies. Examples of suitable radios include the Chupacabra offering from Weathered Security, the ELSAG EOC Plus offering from Leonardo Defense, or the WarDragon offering from DragonOS. Such scanning is based on the understanding that various technologies are commonly used in personal electronic devices such as cell phones, tablets, headsets, smart watches, RFID tags, and other devices commonly carried by spectators to allow wireless communications over distances ranging from a couple centimeters to several meters. Examples include RFID, Near Field Communication (NFC), Bluetooth, and wireless transmitters / receivers for devices such as key fobs, wireless computer peripherals, item locator tags, and the like. Many devices such as cell phones, tablets, and laptops also include cellular transmitters and receivers (e.g. GSM, CDMA, or UMTS capabilities) or wireless networking capabilities (i.e. WiFi networking technology). While the different technologies are designed to operate over varying distances, each is a wireless technology that allows communications within the range of the transmitter / receiver in the device.

[0033] Radio frequency multi-band transceivers 652 may alternately scan the frequencies associated with standards expected to be supported by common consumer devices. Where thestandards (e.g. Bluetooth, 802.11 , etc.) involve a handshake protocol, radio frequency multi-band transceiver 652 may not only determine the existence of a personal electronic device, it may also determine a unique identifier (e.g. a MAC address or Bluetooth ID, or RFID serial number) for the device as a result of the handshake protocol. Given that such identifiers may reasonably be expected to be unique across devices, and that it may reasonably be expected that such devices are owned by particular individuals, Radio frequency multi-band transceivers 652 may assemble an inventory of devices present within range and that inventory can then be used to estimate crowd density or identify individuals. Using such sensors, density can be determined in a variety of ways including estimating the number of persons within the range of a given radio frequency multi-band transceiver 652 based on an assumed average count of devices per person, triangulating the location of a device using directional antennas on a plurality of radio frequency multi-band transceivers 652, and tracking the flow of persons throughout venue 10 as identified devices move in and out of range of radio frequency multi-band transceivers 652. In a preferred embodiment, radio frequency multi-band transceivers 652 are adapted to communicate with a radio frequency density processing module (not illustrated), adapted to receive the identifiers received from each radio frequency multi-band transceiver 652 over time, together with signal strength and, where available direction, to build an inventory of device identifiers and calculate crowd density and crowd flow throughout venue 10. The radio frequency density processing module (not illustrated) may then superimpose that data on a schematic or photographic representation of venue 10 or spectator area 20, or may simply issue alerts based upon pre-determined thresholds (e.g. if a given number of devices is determined to be present within a given radio frequency multi-band transceiver 652, it is assumed that crowd density within the range of that radio frequency multiband transceiver 652 is reaching a dangerous level).

[0034] It is further noted that radio frequency multi-band transceivers 652 can serve the further purpose of monitoring device theft. At events such as festivals, it is not uncommon for teams of pickpockets to work together to steal electronic devices. One thief collects the device and then hands it off to a cohort who places it in a faraday bag, thereby preventing it from transmitting or receiving signals. Where a pattern of devices going offline in a given area is identified, security personnel may be directed to the location to apprehend the thieves.

[0035] A challenge with the use of radio frequency multi-band transceivers 652 to estimate crowd density is the need for calibration. Environmental and other factors such as those discussed above can affect signal strength and range. It is also common for persons to carry more than one such device at a time, while others may carry none, meaning that there is not necessarily a one-to-one correlation between devices and persons. Because the number of persons attending an event is typically large, however, it is possible to determine a reasonable average of the number of devices expected per person for density calculations, and the expected range of a given radio frequency multi-band transceiver 652. Such estimates can be improved by combining counts from a computer vision density monitoring system (discussed below) adapted to use the area around radio frequency multi-band transceiver 652 as an area of interest, with device inventory information. It is thus understood that the combination of computer vision density monitoring (as described below) can enhance and improve the results of radio frequency multi-band density monitoring including by easing and improving calibration.

[0036] It will also be understood that although radio frequency multi-band transceivers 652 may conveniently be software-controlled radios capable of operating on a variety of bands and utilizing a variety of protocols implemented by modern electronic devices, nothing herein precludes use of only one frequency range or protocol in situations where the security staff believes one issufficient. One possible example would be an RFID ticket, wristband, or pass, or limiting reception to WiFi signals alone and ignoring Bluetooth based on the assumption that a given person is likely to carry fewer WiFi-capable devices than Bluetooth capable devices.

[0037] Herein, while use of a software-controlled radio may be convenient, nothing precludes the use of other devices for radio frequency multi-band transceiver 652, particularly where the number of frequencies and protocols to be scanned is more limited. Any device capable of receiving signals from devices having such capabilities, using one or more of the known protocols for wireless communications or networking may be used. Examples include, but are not limited to, wireless hotspots, Bluetooth readers, RFID readers, cellular devices, and wireless network monitors that are commercially available. Such devices may be used individually or combined together to create a single multi-function sensor. In certain embodiments, radio frequency multiband transceiver 652 may be as simple as a cell phone or tablet, having an integrated cell feature and integrated radios for multiple protocols together with software capable of inventorying signals within range and, preferably, signal strength. It may also be as simple as a wireless hotspot adapted to provide signal strength and device identifier information that can be used for device location estimates.

[0038] Where data privacy is a concern in connection with identifiers used for radio frequency multi-band transceivers 652, the inventory may be created by applying a one-way hash to the identifier collected. A variety of one-way hashes may be used ranging from MD5 to more secure SHA-512 and others. The result will be that each identifier remains unique and associated with a theoretical device, but without any practical way of associating the hashed identifier with a person or recovering the identifier itself from the inventory. If an event name or date is combined in the hashing process, then there will be no practical way of determining from an inventory if the samedevice appears at multiple events, thereby enhancing privacy. Such enhancements, however, do not prevent the system from providing density and flow information within an event, albeit using only anonymized data.

[0039] The use of radio frequency multi-band transceivers 652 involves communication directly between radio frequency multi-band transceivers 652 and devices carried by event attendees. Notably, devices carried by attendees also communicate with pre-existing cell towers, WiFi hotspots and other devices. Location data collected by or inferred from cell towers, WiFi hotspots, and other devices may be accessible to density monitoring personnel and combined with other density data through the use of advertising IDs such as those utilized in Android and iOS cellular phones. Advertising IDs allow advertisers to uniquely identify specific devices. Advertising IDs may be tracked and combined with geolocation data based on cell tower locations, internal GPS coordinates, hotspot locations, and other sources. Advertising ID data can also be combined with other types of data (e.g. online shopping preferences or app usage) to generate robust profiles of device users and their preferences. Advertising IDs, therefore, provide an ability both to track the location of mobile devices without the use of radio frequency multi-band transceivers 652, and also to build a profile of the activities and interests of a given device user. Advertising ID data may be purchased from third party sources with varying levels of granularity. Some such data may be historical, but other sources allow purchasing of near real time updates of device location based on advertising IDs.

[0040] In the context of monitoring crowd density and flow at a venue, advertising ID data can be used in a variety of ways. For example, near-real-time location information from advertising ID data providers can show how people are flowing throughout a venue. It can also provide an additional source of data regarding the total number of persons at a venue or the number of personsat a particular location in a venue (e.g. in a given parking lot or near a particular entrance or near a particular stage or concession area). Such data provides a high-level view of crowd density and can serve as an additional confirmation with respect to overall attendance numbers and crowd flow. As such, it can be displayed in addition to, or in combination with data from other sensor types discussed herein, both to increase confidence in density and attendance numbers, and to provide an additional warning indicator as density levels begin to increase.

[0041] In certain embodiments, data from advertising IDs may also be combined with data from radio frequency multi-band transceivers 652 or other sensors to track devices as they move throughout venue 10. For example, by correlating the path taken by a given advertising ID and the path taken by a device identifier collected by radio frequency multi-band transceivers 652 to associate an advertising ID (or group of advertising IDs) with a device identifier (or group of device identifiers). Such tracking can also be used for a variety of purposes, including to coordinate with law enforcement to identify individuals present in locations where crimes took place, deliver targeted advertising based on location, and even to build detailed profiles of event attendees based on activities before, during and after events. It will be understood, therefore, that combining use of advertising ID data (i) provides an additional indicator of crowd density and flow itself, (ii) can be combined with other data collected at the event to create more detailed profiles of event attendees, and (iii) can be used to facilitate gaining direct access to location data on a device that can then be combined with data from other sensor types.

[0042] Where a mobile application associated with the event is utilized, permissions on the application may allow for the gathering of more detailed information about the attendee through the device’s network connection. Permission to gather advertising IDs may also be granted, in which case the local GPS location data may be utilized together with advertising ID data to builda profile of the attendee. One example would be an event app for attendee cell phones that provides information about the event. During installation, an event attendee can be asked to consent to have GPS or advertising ID data from the device disclosed to the event organizer while the event is taking place. Where permission is given, data from the GPS on the device and other data approved for collection through application terms and service conditions, can be collected and the GPS data can be used for crowd flow and density analysis. Advertising IDs collected, and other data collected, can then be combined with location and density data from other sensors to create more detailed profiles of attendees. Such profiles can be used for a variety of purposes including, but not limited to, delivering targeted advertising, providing data to law enforcement, and analyzing event design and configuration efficiency.

[0043] It will be understood that other technologies for people-counting and density monitoring may also be added as additional sensor types. One such example would be LIDAR sensors 552, mounted above the crowd, looking down. LIDAR sensors 552 generate a contour map from which individual persons can be counted automatically based on their height above ground. Ouster OS2, OS1, and OS Dome sensors are examples of sensors that could be used in a venue setting. As such, LIDAR sensors 552 may be used as either a person-counting sensor (for example over open entrance 30), or as a density monitor (for example in the area of open entrance 30 or over areas of interest 25). LIDAR sensors 552 may conveniently be connected to a LIDAR density monitoring unit (not illustrated) that processes the LIDAR contour map to identify persons or estimate crowd density or person counts from the LIDAR data.

[0044] As is discussed above in connection with other sensor types, redundant coverage of an area of interest improves confidence by allowing users to compare counts from different sensor types.Moreover, counts and density measurements from LIDAR data may be improved by calibratingthem with data from cameras 110. Calibration in this sense refers to embodiments where computer vision processing (described below) is used to evaluate crowd density using cameras 110 during an event. To the extent a given camera 110 overlaps with the region covered by a given LIDAR sensor 552, the camera density data can be used as real-time calibration and learning data for LIDAR density monitoring unit (not illustrated). For example, it will be understood that certain LIDAR sensors may be effective up to a given level of density (for example five persons per square meter) but may have difficulty distinguishing individuals when density increases beyond a certain point. The ability to re-calibrate using vision data as densities increase can allow LIDAR sensor(s) 552 to operate more accurately at higher density levels and can help improve LIDAR data as event conditions evolve.

[0045] It will thus be seen that having multiple sensor types / counting means allows the user to improve the performance of individual sensors by allowing more frequent calibration and may also improve confidence by allowing averaging or weighted averaging of data from different sensors in a given region. Data from cameras 110 is particularly helpful as the images from cameras 110 are directly discernable by users, whereas LIDAR data, radio frequency attenuation data, radio frequency density data, and person-counter data is all processed prior to display to a human. As such, the data from cameras 110 allows users to double-check results and calibrations visually with data that is less processed / filtered both to improve confidence, and to improve correlation of various sensor types.

[0046] In FIG. 1 , camera(s) 110 are mounted such that they can capture images covering primary areas of interest 25 in front of stage 40 in their field of view from a downward angle. It is notable that there could be multiple other areas of interest where density monitoring is relevant including in entrance ways, exits, choke points, concession areas, and other locations where large numbersof people are likely to gather, in which additional cameras 1 10 could be deployed. While the figure shows cameras 110 being mounted above stage 40, other mounting locations are possible, including to the side, or in front of stage 40. Preferably, a camera 110 will be mounted such that the angle between camera 110 and the back edge of primary area of interest 25 is no less than forty-five degrees from horizontal, with the angle between camera 110 and the front edge of primary area of interest 25 of no more than ninety degrees. Such angles allow camera images to capture faces, heads, and shoulders throughout area of interest 25, thereby allowing personcounting based on the identification of one or more of those features of the attendees. Where area of interest 25 is larger than the usable field of view of a camera 110, or where there are secondary areas of interest, multiple cameras 110 may be used. In certain embodiments, a camera 110 may also be mounted to an aerial drone (not illustrated) that hovers and moves over spectator area 20 to allow monitoring of different locations over time. It will be understood, however, that where a camera 110 is viewing spectators at an angle, configuration is required to determine the physical area encompassed by the view of the camera based both on the mounting angle and lens characteristics. Where a camera 110 is on a drone, the angle may be maintained at or near 90 degrees, thereby reducing the need for configuration data to account for mounting angles where height of the drone is known.

[0047] Lighting conditions at musical and theatrical events present unique challenges for camerabased sensor systems. Darkness, smoke machines, on-stage effects, large format monitors, lighting effects, costumes and lighting devices used or worn by participants, and flash photography all create photographic challenges. Camera(s) 110, therefore, are preferably cameras with low-light capability, infrared capability, high quality optics, and high-quality image processing capabilities.One such camera is the Axis Q6225-le PTZ (see https: / / www.axis.com / products / axis-q6225-le).Camera(s) 110 may conveniently have a light sensitivity (B / W) of 0.001 lux, a wide dynamic range, and a minimum sensor size (1 / 2"). During the day, camera(s) 110 may conveniently be left in color mode. At dusk, nighttime, or during particularly cloudy days infrared may be used. Alternatively, an operator may switch a given camera 110 between optical and infrared mode throughout the event, evaluating which mode gives more accurate information at various times based on lighting conditions. While providing a clear, human-readable image is helpful for calibration and human monitoring, increasing contrast and saturation has been shown to make heads / faces easier for a computer vision module to identify. Exact settings, however, will have to be determined based on light conditions in venue 10 throughout the event. Wide dynamic range settings further assist in head / face identification. Maximum image / frame size will also depend on conditions. Typically, however, selecting image / frame sizes where heads are no smaller than 15 by 15 pixels is desirable.

[0048] Referring now to FIG. 2, camera 110 is operably connected to computer vision processing module 120. While camera(s) 110 may be configured to transmit images wirelessly to computer vision processing module 120, where higher resolutions and frame rates are used, camera(s) 110 may be hard-wired to computer vision processing module 120 to reduce risks emanating from wireless interference or slower wireless bandwidth during an event.

[0049] Computer vision processing module 120 is required to process and record data from camera(s) 110 and may need to apply adaptive algorithms to count persons in the image data. In such embodiments, it may be convenient to use a GPU-based computing platform for computer vision processing module 120 to improve processing throughput. The NVIDIA Jetson AGX ORIN (see https: / / www.nvidia.com / en-us / autonomous-machines / embedded-systems / jetson-orin / ) is one such platform. Using a platform designed for Al processing and capable of high-speed operationmay conveniently allow for increased sampling rates, and increased numbers of cameras being supported by one computer vision processing module 120. Such platforms may conveniently support one to three cameras 110 depending on memory and processor speeds, with two cameras 110 being typical.

[0050] Computer vision processing module 120 may conveniently be located underneath stage 40 and may be connected directly to camera(s) 110 by ethernet cables, or through a typical network switch (not illustrated), or wirelessly using a wireless access point (not illustrated) or internal wireless networking adapter (not illustrated). Utilizing a wired connection between camera(s) 110 and computer vision processing module 120, but also providing a wireless access point on the same network, allows for a fast and reliable connection between camera(s) 110 and computer vision processing module 120, and also allows personnel to connect wirelessly to the same network from portable devices such as laptops, tablets, and cell phones for monitoring and diagnostics. As is discussed below, such connections are particularly helpful when setting up and configuring the system as they do not require a wired connection or interaction with a venue network being used for other purposes.

[0051] In other embodiments, camera(s) 110 can be connected to the internet or another wide area network. In such configurations, computer vision processing module 120 could be in a remote office or could be configured as a virtual machine in a cloud computing environment. In such embodiments, lower frame rates are typically used, however, based on available bandwidth.

[0052] As is discussed further below, computer vision processing module 120 is operably connected to density monitoring station 170, which may conveniently be located in a security office 50 away from spectator area 20. Such connection may be achieved either by a direct wire connection, or by placing density monitoring station 170 and computer vision processing module120 on the same network. While WiFi can be used, reliability factors relating to event security and monitoring crowd density will often favor use of a wired connection or a wireless point-to- point connection.

[0053] Due to the many activities taking place during event setup, it can be desirable to have computer vision processing module 120 and camera(s) 110 on their own local area network, and to incorporate a router (not illustrated) to allow that network to connect to an event network or, where no event network is conveniently available, to a cellular modem (not illustrated) or other access point to allow the local area network to connect to the internet. In this way, the combination of camera(s) 110, computer vision processing module 120, a network switch / router combination (not illustrated), a wireless access point, optional battery backup, and cabling sufficient to connect camera(s) 110 to computer vision processing module 120 may be provided as a pre-configured kit for easy deployment at a venue.

[0054] In an exemplary embodiment, computer vision processing module 120 will comprise an image processor with software adapted to count persons in captured images from camera(s) 110. Counting heads, faces, or shoulders is suitable, particularly where camera(s) 110 are positioned above the spectators and looking down (directly or at an angle). Known adaptive algorithms and techniques may be applied to identify heads or faces and improve recognition over time. One example of an artificial intelligence package that has proven suitable for use in face or head counting is the Scale-Adaptive Selection Network (“SASNet”) described in "To Choose or to Fuse? Scale Selection for Crowd Counting" by Qingyu Song, Changan Wang, Yabiao Wang, Ying Tai, Chcngjic Wang, Jilin Li, Jian Wu, and Jiayi Ma, published in 2021 in Proceedings of the AAAI conference on artificial intelligence, vol. 35, no. 3, pp. 2576-2583 (see https: / / ojs.aaai.org / index.php / AAAI / article / view / 16360), which is hereby incorporated byreference. In one embodiment, five crowd counts are calculated per second, with computer vision processing module 120 being adapted to display a one-minute moving average. This is a form of visual density data. An average of multiple averages may be used where the images are intermittently blocked or distorted by, for example, by fog machines or short-duration lighting effects. Averaging allows for reduction of errant counts, and other signal processing methods typically used to account for noise may also be used. In this way computer vision processing module 120 performs the following steps: (i) repeated analysis of image data to count persons in a given region, (ii) averaging or smoothing of count data, (iii) provision of count data in numerical format, and (iv) provision of count data in graphical format using heatmap or color-coded boundaries to indicate situations where density data approaches or exceeds certain pre-determined thresholds. Such data may then be communicated to density monitoring station 170 for display to event security or safety personnel. Users of density monitoring station 170 may adjust camera parameters during the event as needed, for example by switching to infrared mode or adjusting saturation values. Such parameters may selected or adjusted automatically or manually depending on the capabilities of camera(s) 110 and the characteristics of the installation and event.

[0055] Density monitoring station 170 may conveniently be a desktop or laptop computer (or in some cases, a tablet) configured with software to display input from one or more crowd counting or density monitoring systems and connected to such systems through a local area network. Where remote monitoring is needed, density monitoring station 170 may be connected through a wide area network instead, such as the internet, provided that reliability and bandwidth are sufficient to meet safety concerns. As illustrated, density monitoring station 170 is operably connected to two computer vision processing modules 120, one person-counter processing module 604, one LIDAR processing module 556, one radio frequency density processing module 656, and one radiofrequency attenuation processing module 506, in each case with one processing module supporting one or more sensors. It will be understood that as many sensor types as necessary may be included, or that as few as one sensor type may be used, depending on the needs of the event.

[0056] It will be understood that person-counter processing module 604, LIDAR processing module 556, radio frequency density processing module 656 and radio frequency attenuation processing module 506 are each similar to computer vision processing module 120 described above in that each accepts data from one or more sensors of a given type and processes that data to calculate crowd count, crowd density, or device identification values. Where crowd density is calculated, the values presented may conveniently be in the form of current persons per square meter, average persons per square meter over a given time range, and maximum recorded persons per square meter. Crowd density may also be conveniently displayed visually as a heatmap. In the case of computer vision processing module 120 and LIDAR processing module 556, the heatmap may conveniently be superimposed on the image(s) generated by camera(s) 110 or the image generated by LIDAR sensor(s) 552, and may be in the form of a border drawn around an area of interest (e.g. area of interest 25) where the border colors change depending on average density value over a pre-determined time range, or in the form of a traditional heatmap showing color- coded blob shapes over locations in the image based on their calculated density. In the case of any sensor type, the heatmap could conveniently be superimposed on a schematic or photographic representation of venue 10 or spectator area 20. Density monitoring station 170 is conveniently adapted to display the data generated by the processing modules individually, selectively, or side- by-side. Average or combined values may also be displayed in numeric format, preferably color- coded.

[0057] Notably, in the embodiment illustrated, computer vision processing module 120, personcounter processing module 604, radio frequency attenuation processing module 506, radio frequency density processing module 656 and LIDAR processing module 556 are represented as modular computers, each connected to one or more sensors via a network or other connection. Such modular computers may conveniently be local at the venue or could also be deployed remotely or in a cloud computing environment and connected through a wide area network such as the internet. While a locally installed modular computer may be preferable, particularly for processing modules such as computer vision processing module 120, which typically utilizes particularly high processing power and hardware adapted to adaptive algorithms, there is no reason that processing modules may not be built directly into sensors, or incorporated into density monitoring station 170, given sufficient processing power and capacity. Where the processing load of a given processing module is high for a particular’ sensor, it may be preferable to utilize a special purpose device such as the Jetson device discussed above to distribute workload and allow scalability. Where processing load is lower (such as for person-counter sensors 602), it may be more convenient to incorporate processing into density monitoring station 170, or directly into a sensor, or as a cloud resource. The particular configuration used will depend on the requirements of a given venue and event, with the understanding that the embodiments disclosed provide a modular-, scalable solution that can be used in a wide variety of settings, with density monitoring station 170 collecting and displaying processed sensor data in a visual format with alerts at predetermined density levels.

[0058] While density monitoring station 170 may conveniently be located in security office 50 (shown on FIG. 1) with access to power, it may also be in the form of a laptop or tablet that is powered by battery. Such portable configurations of density monitoring station 170 allowflexibility in that density monitoring station 170 may be located anywhere with a sufficient network connection, including backstage or in a ticket office.

[0059] In certain embodiments, it may also be desirable to allow roving security personnel 52 to see all or a portion of what is displayed on density monitoring station 170. This may be accomplished by operably connecting one or more mobile display devices 172 to density monitoring station 170 using a wireless network. Mobile display devices 172 may conveniently be tablet computers or smartphones. Reality augmentation devices such as augmented reality glasses or virtual reality headsets may also be used to provide a heads-up display.

[0060] Mobile display devices 172 are adapted to replicate a portion of the crowd density data displayed on density monitoring station 170 and allow communication via messaging between the operator of density monitoring station 170 and roving security personnel 52. Mobile display devices 172 may, in an alternate embodiment, be adapted to allow security personnel 52 to select different sensor data, or zoom in on images, as opposed to simply replicating all or a portion of density monitoring station 170.

[0061] Communications among security personnel 52 during loud and crowded events can be difficult. It can also be difficult to use words alone to identify a specific location. Security personnel 52, therefore, may also be outfitted with laser designators (not illustrated) as a means of enhancing communication. Laser designators may conveniently be laser pointers that are invisible to the human eye, but easily detected by camera(s) 110. One possible example is the “Odin” infrared (IR) laser 808nm-980nm-1064nm (see https: / / biglasers.com / product / odin-infrared-ir- laser-808nm-980nm-1064nm / ). When so equipped, security personnel 52 may identify areas of interest in real time by using laser designators to point to particular areas, which would then be displayed on density monitoring station 170. Encircling a given area with a moving beam from alaser designator will draw the attention of the operator of density monitoring station 170 on that particular area. This would allow security personnel 52 to act as additional sensors, calling attention to areas of concern, and enable visual communication with other personnel, in addition to, or instead of, radio-based voice communication. In some environments, white light devices (such as a handheld flashlight) may be used in addition to, or in lieu of, infrared laser designators. [00621 In certain embodiments, it may be desirable to allow density monitoring by persons outside of venue 10. This may be accomplished by ensuring that density monitoring station 170 has access to the internet or another wider area network, and using remote display devices 174 to monitor replicated portions of the display of density monitoring station 170 (as was described in connection with mobile display devices 172) or providing a customized user interface whereby remote display devices 174 may become auxiliary density monitoring stations with the same functionality as density monitoring station 170. It will be understood that, in such configurations, density monitoring station 170 may comprise a web server capable of generating replicated portions of its display, or other custom displays, to remote device. Such configurations would allow offsite persons such as law enforcement, emergency responders, or event managers / producers to monitor the event remotely, in near real time. Remote display devices 174 may conveniently be devices with web browsers with the information from density monitoring station 170 being served as an interactive web page.

[0063] Turning now to FIG. 3, a method of configuring camera(s) 110 and computer vision processing module 120 is disclosed. At a high level, configuring camera(s) 1 10 and computer vision processing module(s) 120 requires (i) identifying the area of interest to be monitored, (ii) mounting camera(s) 110 in a location that allows them to look down on the area of interest, (iii) adjusting the zoom level and camera direction to show the area of interest in the camera image,and (iv) specifying both the location of the area of interest in the camera image, and the size of the boundaries of the area of interest on the ground (or at an assumed head level) so that density may be calculated based on a physical measurement such as persons per square meter. Such configuration may be accomplished in a variety of ways using a portable device (not illustrated) such as a tablet, mobile phone or laptop that is connected wirelessly to the same network to which cameras 110 and computer vision processing module 120 are connected. In a first method, illustrated in FIG. 3, physical markers are placed on the ground to outline the area of interest in step 710. It is notable that the area of interest is unlikely to be a perfect rectangle due to the physical characteristics of venue 10. Instead, it is assumed that it may be a polygon with as many sides as are needed, with each vertex of the polygon marked with a marker (e.g. a disk, tape, bucket, etc.). In step 720, camera 110 is pointed and the zoom level is set so as to encapsulate each of the markers in the image, thereby ensuring the entire area of interest is in the view. As is noted above, the zoom level is preferably set such that a typical person’s head would represent no fewer than 15 by 15 pixels in the camera image. In step 730, distances between the vertices are measured on the ground. Such distances determine the physical area (preferably in square meters) of the area of interest, which is required for a density measurement. While ground measurements and markers can be used, it may be preferable to calculate the boundaries of the area of interest at head level. In step 740, therefore, an elevated marker is placed over each physical marker. The elevated marker may conveniently be a leveled tripod or surveying rod with a height that approximates the average height of a person. Because the elevated marker is over the physical marker, and is level, the polygon represented by the elevated markers is, in effect, the same polygon represented by the physical markers, but raised to the height of an average person’s head.The area of the two polygons is the same, but the height above ground is different. In step 750,the vertexes of the raised polygon are identified in the camera display for computer vision processing module 120 by pointing to them on the portable device. In step 760, the lengths of each side of the polygon (measured in step 730) are entered into the portable device, thereby enabling computer vision processing module 120 to calculate the physical area of the area of interest.

[0064] An alternative method of configuration is illustrated in FIG. 4. In step 810, mats formed of at least four rectangles of a known size (e.g. one-meter squares) are positioned at or near the vertexes of the area of interest and aligned to be parallel to the front of stage 40 (illustrated in FIG. 1). The rectangles should use high contrast colors (e.g. black and white) easily detected by computer vision and easily discerned by human operators. In step 820, the portable device is used to point and zoom camera 110 so as to encapsulate the area of interest in the camera’s field of vision, preferably assuring that zoom level is set so that a typical person’s head will require no less than 15 by 15 pixels in the image. In step 830, computer vision processing module 120 identifies (or a user of a portable device indicates) the location of each mat. In step 840, computer vision processing module 120 identifies the four corners of each rectangle on each mat and calculates the number of pixels horizontally and vertically represented by a rectangle at or near each vertex. As a result, computer vision processing module 120 now has a correlation between pixel count and physical area on the ground at or near each vertex. In step 850, computer vision processing module 120 calculates the number of pixels represented by each side of the polygon and uses the correlation to calculate the physical length of the side. They polygon may be defined by the placement of the mats, or interactively on the mobile device by designating points near the mats. It is noted that, due to lens characteristics, angles and zoom levels, the number of pixels per meter at one vertex, in one direction, may not be identical to the number of pixels per meter at a different vertex in the same direction. Computer vision processing module 120 may, therefore, interpolatethe pixel counts at or each vertex that was calculated based on the different mats in order to obtain a more accurate area calculation. To account for differences in head height, the mats may be supported at an average head height by legs or a tripod instead of being placed on the ground in order to increase accuracy.

[0065] A third configuration method is illustrated in FIG. 5, particularly suitable for venues in which the area of interest is rectangular on the ground. In step 910, a large mat comprising high- contrast rectangles of a known dimension (e.g. 1 -meter squares in a black and white checkerboard pattern) are laid on the ground, covering the area of interest. In step 920, camera 110 is pointed and zoomed to encapsulate the area of interest in the camera image, preferably assuring the zoom level is set such that an average person’s head requires no less than 15 by 15 pixels in the image. In step 930, the physical area of the area of interest is read off by counting the rectangles between the corners of the area of interest and entered into computer vision processing module using the portable device. In step 940 either a raised marker is placed over each vertex, or a person stands over each vertex, and the portable device is used to select the vertex within the camera image, thereby allowing computer vision processing module 120 to know the area of the image that corresponds to the area of interest and the physical area of that area of interest. In certain embodiments, computer vision processing module 120 may be adapted to automatically measure the distance between vertexes using the high-contrast rectangles on the mat, and the known physical size of each rectangle.

[0066] It is notable that the foregoing configuration methods may also be performed using density monitoring system 170 instead of a portable device, or by entering data using a keyboard and mouse connected directly to computer vision processing module 120. Using a portable device connected to the same network as computer vision processing module 120, however, may bepreferable as the portability and connectivity of the device allows information to be collected by a single person working in the event floor.

[0067] The disclosure herein provides description of the systems and methods of the present invention and the various components thereof. It will be understood that such specifics are for illustrative and exemplary purposes only and are not intended to be limiting.

[0068] The foregoing pertains to certain preferred embodiments only. Persons having ordinary skill in the art will readily see variations and improvements that may be included in alternate embodiments in light of this disclosure. The invention described herein is not intended to be limited to the preferred embodiments discussed in the detailed description or shown in the figures.

Claims

CLAIMSWhat is claimed is:Claim 1: A crowd density monitoring system comprising at least one camera mounted above a venue stage and adapted to view an area of interest within a spectator area of the venue from above, a computer vision processing module operably connected to the camera and adapted to count persons in an area of interest within the spectator area and operably connected to a density monitoring station; at least one person-counter at an entrance to the venue and operably connected to the density monitoring station; the density monitoring station comprising a density monitoring station display adapted to display the image captured by the camera, display the boundaries of the area of interest, display the highest density per pre-determined area calculated by the computer vision processing module, display the person-counter counts from the person-counter, and issue alerts in the event the calculated density within the area of interest exceeds at least on pre-determined threshold value whereby an operator of the density monitoring station may observe the area of interest while an event is taking place and receive alerts when density of persons within the area of interest exceeds the pre-determined threshold value.Claim 2: The system of claim 1 further comprising at least one secondary density monitoring system selected from the group consisting of a radio frequency attenuation density monitoring system, a lidar density monitoring system, and a radio frequency density monitoring system, the secondary density monitoring system being operably attached to the density monitoring station, wherein the density monitoring station is further adapted to issue alerts in the event that the secondary density monitoring system determines that the density of persons exceeds a pre-determined threshold value and allow the operator to selectively choose between displays by the computer vision processing module and displays generated by the secondary density monitoring system.Claim 3: The system of claim 2 wherein the at least one secondary density monitoring system is adapted to receive density per pre-determined area from the computer vision processing module as a calibration value and adjusts the determined density values based on the calibration value.Claim 4: the system of claim 3 wherein the calibration value is determined prior to the commencement of an event.Claim 5: the system of claim 3 wherein the calibration of the secondary density monitoring system is adjusted at least once per thirty-minute period during the event.Claim 6: the system of claim 1 further comprising a mobile display device operably connected to the density monitoring station and adapted to replicate at least a portion of the display of the density monitoring station, whereby an operator of the mobile display device in the venue may observe a portion of the area of interest while an event is taking place and receive communications from the operator of the density monitoring station in the form of messages delivered to the portable display device.Claim 7: the system of claim 6 further comprising a laser designator wherein the operator of the portable display device may use the laser designator to identify specific locations in the area ofinterest and the operator of the density monitoring station will see the indicated area on the display of the density monitoring station.Claim 8: the system of claim 1 further comprising a remote display device operably connected to the crowd density monitoring station through a wide area network and adapted to replicate at least a portion of the display of the density monitoring station, whereby an operator of the remote display device outside the venue may observe a portion of the area of interest while an event is taking place.Claim 9: A computer vision crowd density monitoring system comprising at least one camera mounted above a venue stage and adapted to view a portion of a spectator area at an angle no less than forty-five degrees, a computer vision processing module operably connected to the camera and adapted to count persons in an area of interest within the spectator area, and a density monitoring station operably connected to the computer vision processing module wherein the computer vision processing module is adapted to count persons within a pre-defined area of interest determined by pixel locations within the image captured by the camera and transmit the person counts to the density monitoring station in real time.Claim 10: the system of claim 9 wherein the computer vision processing module is integrated with the camera.Claim 11: the system of claim 9 further comprising a network switch, a wireless access point, and a portable configuration display wherein the camera, the computer vision processing moduleand the wireless access point are operably connected to the switch, and the portable configuration display is wirelessly connected to the wireless access point, and wherein the portable configuration display is adapted to display the image generated by the camera and enable an operator to remotely aim and zoom the camera, designate an area of interest in a venue by designating a polygon on the portable configuration display, the polygon being determined by pixel values within the image generated by the camera, and enter the physical dimensions of the area of interest.Claim 12: the system of claim 11 wherein the density monitoring station is operably connected to a secondary network, and further comprising a network router operably connected to the network switch and adapted to provide connectivity to the secondary network wherein the computer vision processing module is operably adapted to communicate with the density monitoring station through the secondary network.Claim 13: the system of claim 11 wherein the router is operably connected to a cellular modem and the cellular modem provides connectivity to the secondary network.Claim 14: A method of configuring a computer vision density monitor system comprising at least one camera mounted above a venue stage and adapted to view a portion of an area of interest within a spectator area, a computer vision processing module operably connected to the camera and adapted to count persons in the area of interest, a density monitoring station wherein the computer vision processing module is adapted to count persons within the area of interest determined by pixel locations within the image captured by the camera and transmit the person-counts to the density monitoring station, and a portable configuration display operably connected to the computer vision module and the camera and adapted to allow an operator of the portable configuration to aim and zoom the camera, designate specific pixels within the camera display area to define the area of interest, and provide configuration parameters to the computer vision processing module including the physical dimensions of the area of interest, the method comprising the steps of pointing and setting the display area of the camera using the remote configuration display, drawing a polygon on the remote configuration display to designate an area of interest, and entering the length of each side of the polygon in the physical venue whereby the physical area represented by the polygon may be calculated.Claim 15: the method of claim 14 further comprising laying a mat comprising rectangles of predetermined size within the area of interest and determining the length of at least one polygon side by counting the rectangles on the mat.Claim 16: the method of claim 14 further comprising laying a mat comprising rectangles of predetermined size within the area of interest, determining the number of horizontal and vertical pixels represented by at least one rectangle on the mat, and calculating the physical area of the area of interest based on the number of pixels represented within the polygon.Claim 17: the method of claim 16 further comprising laying a plurality of mats comprising rectangles of predetermined size at a plurality of locations within the area of interest, determining the number of horizontal and vertical pixels represented by at least one rectangle ineach mat, and interpolating to determine the physical area of the area of interest based on the number of horizontal and vertical pixels associated with each rectangle.Claim 18: the method of claim 16 further comprising the step of positioning the mat at a height that approximates the height of a person.Claim 19: the method of claim 14 further comprising the step of placing markers having a height that approximates the height of a person at each vertex of the area of interest and positioning the vertices of the polygon based on the markers at height.Claim 20: the method of claim 14 further comprising the step of adjusting the location of each corner of the polygon to a height approximating the height of a person.

Images