Patent Application
20200388039
The present invention is in the field of image analysis, specifically, the use of image analysis to detect interactions between occupants in a space.
The ability to detect and monitor occupancy in a space, such as a room or building, enables planning and controlling building systems for better space utilization, to minimize energy use, for security systems and more.
Planning and controlling building systems for better space utilization (for example) may require understanding interaction between occupants in the space. For example, measures taken to prevent the spread of a contagious disease may include physical distancing, namely, maintaining a physical distance between people and reducing the number of times people come into close contact with each other. Physical distancing typically involves keeping people at a certain distance from each other and avoiding gathering together in large groups.
There is currently no system that enables real-time monitoring of physical distancing in building and other spaces.
Embodiments of the invention provide methods and systems for real-time monitoring of physical distancing in building and other spaces, e.g., by determining occupants' interactions in the space, substantially in real-time, and possibly generating an alarm based on the interactions, e.g., when non-compliance with physical distancing rules, is detected.
In one embodiment, a method for determining occupants' interactions in a space includes applying a computer vision algorithm to obtain locations of an occupant, in the space, over time and using a processor to compare locations of a plurality of occupants, to extract interaction points between the plurality of occupants. Typically, the interaction points are points that fulfill a pre-determined condition. For example, the interaction points may include locations in the space in which at least two occupants are situated at a distance within a pre-determined range, from each other, possibly for a time period over a specified threshold. In another example, the interaction points may include locations in the space in which a number of occupants above a specified threshold are located within a defined time period.
A system, according to an embodiment of the invention may include a camera to obtain images of a space and a processor in communication with the camera, to obtain locations of an occupant in the space, over time, from the images. The processor may then compare locations of a plurality of occupants to extract interaction points between the plurality of occupants. The interaction points may fulfill a pre-determined condition, such as, locations in the space in which at least two occupants are situated at a distance within a pre-determined range, from each other (possibly for a time period over a specified threshold) and/or locations in the space in which a number of occupants above a specified threshold are located within a defined time period. The processor may generate an alarm based on the extracted interaction points. The alarm or another signal generated by the processor, based on the extracted interaction points, may cause a change in a display, e.g., to show degree of compliance with physical distancing rules.
In another embodiment, a system includes a processor that can receive images of a space and can create a history log of an occupant based on tracking of the occupant through the images of the space, to obtain locations in the space of the occupant over time. A map of occupant interaction in the space can be created, based on comparison of history logs, that includes locations in which at least two occupants are situated at a distance, which is within a pre-determined range (e.g., below a specified threshold), from each other.
In one embodiment the map of occupant interactions includes locations in the space in which the at least two occupants are located for a time period over a specified threshold.
In another embodiment the map of occupant interactions includes locations in the space in which a number of occupants above a specified threshold are located.
For example, a system may include a processor to obtain locations of occupants in the space, by applying a computer vision algorithm to images of the space, and create a history log of an occupant based on tracking of the occupant through the images of the space to obtain locations in the space of the occupant over time. The processor may create a map of occupant interaction in the space based on comparison of history logs. The map may include locations in the space in which a number of occupants above a specified threshold are located.
The processor can generate a signal indicative of compliance with physical distancing rules. For example, the processor can generate a signal to issue an alarm when the number of occupants in the location is above a specified threshold.
The invention will now be described in relation to certain examples and embodiments with reference to the following illustrative drawing figures so that it may be more fully understood. In the drawings:
Embodiments of the invention provide methods and systems for automatically assigning space related resources to occupants. The space may be an in-door space (such as a building or parking lot space) or out-door space.
Although examples described herein refer to space related resources such as work stations and allocation of work stations to human occupants, it should be appreciated that embodiments of the invention relate to all types of stations, facilities or specific spaces for work purposes or other uses and for any type of occupant.
For example, a work station according to embodiments of the invention may include a desk and the occupant a person. In other embodiments of the invention a work station includes a stall and the occupant an animal. In yet other embodiments a work station includes a parking spot and the occupant a vehicle. Other stations and occupants are included in embodiments of the invention.
In the following description, various aspects of the present invention will be described. For purposes of explanation, specific configurations and details are set forth in order to provide a thorough understanding of the present invention. However, it will also be apparent to one skilled in the art that the present invention may be practiced without the specific details presented herein. Furthermore, well known features may be omitted or simplified in order not to obscure the present invention.
Unless specifically stated otherwise, as apparent from the following discussions, it is appreciated that throughout the specification discussions utilizing terms such as “analyzing”, “processing,” “computing,” “calculating,” “determining,” “detecting”, “identifying” or the like, refer to the action and/or processes of a computer or computing system, or similar electronic computing device, that manipulates and/or transforms data represented as physical, such as electronic, quantities within the computing system's registers and/or memories into other data similarly represented as physical quantities within the computing system's memories, registers or other such information storage, transmission or display devices.
Examples of systems operable according to embodiments of the invention are schematically illustrated in
In one embodiment, which is schematically illustrated in
For example, processor 102 may apply shape detection algorithms on images obtained from image sensor 103 to detect an occupant by its shape in the image(s) and to track the shape of the occupant.
In some embodiments different occupant characterization groups are created by processor 102 or another processor, and the occupant is added to an occupant characterization group based his characterization. The occupant is then assigned a work station based on his characterization group.
“Characterization of an occupant” may include behavior patterns of the occupant, as detected from the tracking of the occupant. For example, behavior patterns may include amount and frequency of motion paths taken by the occupant, destination and timing of motion paths, location of the paths in the space, amount of time an occupant occupies a location, etc.
Thus, for example, if, based on tracking an occupant it is detected that the occupant frequently walks to location X in an office building and stays at that location long periods, then the occupant may be characterized as “interacting with location X” and based on this characterization the occupant may be assigned a work station in the vicinity of location X.
In one embodiment characterization of an occupant is based on the shape of the occupant in the images. For example, tracking the occupant in images of the space may include tracking the shape of the occupant.
In another embodiment characterization of the occupant is based on a body position of the occupant (e.g., standing vs. sitting or reclining). The body position may be detected, for example, from the shape of the occupant in the images.
In some embodiments processor 102 may create a characterization of an occupant based on input from a sensor in the space that is not an imager. For example, an audio sensor may monitor sound in the space. Audio input corresponding to images of an occupant can be used in characterizing the occupant. Thus, an occupant characterization can be created based on audio information related to the occupant and, for example, an occupant characterized as “noisy” may be assigned to an area of the office where noise is less disturbing.
In some embodiments an occupant characterization can be created based on a map of the occupant's paths through the space and the occupant may be assigned a work station in the space so as to minimize the occupant's paths in the space.
According to embodiments of the invention different image-based information and possibly non image-based information can be calculated by processor 102 to provide an occupant characterization.
The occupant characterizations, as well as other signals generated by processor 102 may be output to an external device 105, for example, to be saved, further calculated, analyzed or displayed. In one embodiment a signal is output to external device 105, based on the assignment of the occupant to a work station.
The external device 105 may include a central server or the cloud. For example, external device 105 may include a processing unit that uses the output from processor 102 (or from a plurality of processors connected to a plurality of image sensors) to update seating arrangements of the space 104 or to update statistics relating to the space. In one example, space 104 may include part of an office building space in which occupants are seated in groups. The groups may be dynamic because occupants in each group can change based on past behavior or interactions of the occupant with his surroundings. Information relating to occupants' behavior may be added to statistics of the office building e.g., information regarding the number of people going to specific areas in the building at what hours, etc.
In some embodiments device 105 includes a display and assignment of occupants to work stations may be shown on the graphical interface of the display.
In some embodiments output from processor 102 is used by space related resources management system software (e.g., a smart building management system) to assign work stations in the space to occupants. For example, a system, such as a smart building management system, may use output from processor 102 to cause a visual indication to appear in vicinity of the work station assigned to an occupant. For example, once a work station is assigned to an occupant by processor 102 a signal may be sent to light up an LED or other visual indicator above the work station so that other occupants are advised of the “occupied” status of this work station. The visual indicator may be turned off once it is detected (e.g., by analysis of images of the space) that the occupant is occupying the assigned work station.
The processor 102 may be in wired or wireless communication with device 105 and/or with other devices and other processors. For example, a signal generated by processor 102 may activate a process within the processor 102 or may be transmitted to another processor or device to activate a process at the other processor or device. In one embodiment a signal generated by processor 102 may be indicative of compliance with physical distancing rules and may cause an alarm to be issued, for example, if non-compliance with physical distancing rules has been detected.
A counter to count occupants and/or occupied work stations in the space 104 may be included in the system 100. The counter may be part of processor 102 or may be part of another processor that accepts output, such as a signal, from processor 102.
Processor 102 may include, for example, one or more processors and may be a central processing unit (CPU), a digital signal processor (DSP), a microprocessor, a controller, a chip, a microchip, an integrated circuit (IC), or any other suitable multi-purpose or specific processor or controller.
Memory unit(s) 112 may include, for example, a random access memory (RAM), a dynamic RAM (DRAM), a flash memory, a volatile memory, a non-volatile memory, a cache memory, a buffer, a short term memory unit, a long term memory unit, or other suitable memory units or storage units.
According to some embodiments, images obtained by the image sensor 103 are stored in memory 112. In one embodiment images obtained by the image sensor 103 are 2D images. The 2D images may be analyzed by processor 102 using image analysis methods, such as color detection, shape detection and motion detection or a combination of these and/or other computer vision methods.
For example, shape detection (or recognition) algorithms may include known shape detection methods such as an algorithm which calculates features in a Viola-Jones object detection framework. In another example, the processor 102 may run shape detection algorithms which include machine learning processes. For example, a machine learning process used to detect an occupant and/or a work station and/or an occupied work station, according to embodiments of the invention, may run a set of algorithms that use multiple processing layers on an image to identify desired image features (image features may include any information obtainable from an image, e.g., the existence of objects or parts of objects, their location, their type and more). Each processing layer receives input from the layer below and produces output that is given to the layer above, until the highest layer produces the desired image features. Based on identification of the desired image features an object such as an occupied or unoccupied work station or an occupant may be detected or identified. Motion in images may be identified similarly using a machine learning process. Objects, such as occupants, may be tracked by processor 102 through a sequence of images of the space using known tracking techniques such as optical flow or other suitable methods. In one embodiment an occupant is tracked based on his shape in the image. For example, an occupant is identified in a first image from a sequence of images as an object having a shape of a human form. A selected feature from within the human form shaped object is tracked. Shape recognition algorithms are applied at a suspected location of the human form shaped object in a subsequent image from the sequence of images to detect a shape of a human form in the subsequent image and a new selected feature from within the detected shape of the human form is then tracked, thereby providing verification and updating of the location of the human form shaped object.
In one embodiment the image sensor 103 is designed to obtain a top view of a space. For example, the image sensor 103 may be located on a ceiling of space 104, typically in parallel to the floor of the space, to obtain a top view image of the space or of part of the space 104.
Processor 102 may run processes to enable identification of objects such as stations and/or of occupants, such as humans, from a top view, e.g., by using rotation invariant features to identify a shape of an object or person or by using learning examples for a machine learning process including images of top views of objects such as work stations or other types of stations and of people or other types of occupants.
Another example of a system for assigning space related resources operative according to embodiments of the invention, is schematically illustrated in
Tracking an occupant may include determining locations of the occupant in the space (e.g., in relation to a floorplan of the space). In some embodiments, an appearance model may be used to tag a specific occupant and to track the specific occupant throughout images of the space.
In one embodiment, typically when analyzing top view images of a space, processor 102 may determine locations of objects in an image on the floor of the space in the image. In one embodiment location of an occupant (or other object) in an image is calculated from the detected shape of the occupant. The location on the floor in the image may then be transformed to a real-world location by the processor 102. The shape of the occupant may be used to determine his location on the floor of the space in the image by, for example, determining a projection of the center of mass of the occupant which can be extracted from the occupant's shape in the image, to a location on the floor. In another embodiment the location of an occupant on the floor in the image may be determined by identifying the feet of the occupant based on the detected shape of the occupant. The location of the feet in the image is determined to be the location of the occupant on the floor in the image. A processor may then transform the location on the floor in the image to a real world location by using, for example, projective geometry.
Characterizing features are deduced from the tracking and are input to a characterization database 12 (which may be maintained, for example, on memory 112 or externally, e.g., on a remote server or on the cloud).
In some embodiments an occupant may have an identity attached to him, for example, based on image processing of the images 10 (e.g., using face recognition algorithms) and/or by an external identifying signal 13 (such as an RF ID signal received at a time correlating to an image of the occupant or by an occupant registering to the system). Characterizations may be linked to occupant identities, for example, by using a look-up table 14.
Processor 102 may run a process 15 to assign an identified occupant a work station by matching characterizing features from database 12 to a work station from facility database 16 (which may be maintained, for example, on memory 112 or externally, e.g., on the cloud).
A signal output from processor 102 can be sent, for example to a display 17 to show which work station was assigned to which occupant. In some embodiments a signal output from processor 102 is indicative of compliance with physical distancing rules and can cause a change on display 17 to indicate compliance (or non-compliance) with physical distancing rules.
In some embodiments processor 102 may create a map of occupant interactions in a space from images of the space 104 received from image sensor 103. The processor 102 may, by tracking an occupant throughout the images, obtain locations of an occupant in the space over time and may create a history log of the occupant from these locations. The created history logs and/or updated history logs of a plurality of occupants may be maintained in a database of the history logs. Processor 102 may then create a map of occupant interactions in the space based on comparison of history logs. The map typically includes locations fulfilling predetermined conditions, as further described below.
The map of occupant interactions may be displayed on display 17 or may be otherwise transmitted to an external device for further analysis and processing.
An example of a method, run by processor 102, for automatically assigning space related resources, such as work stations in an office building, is schematically illustrated in
In one embodiment the method includes detecting a shape of the occupant in the images and tracking the shape of the occupant (e.g., as described above).
In some embodiments the method includes determining a body position (e.g., standing vs. sitting or reclining) of the occupant in one or more of the images and creating an occupant characterization based on the body position of the occupant.
A body position of an occupant may be determined based on the shape of the occupant. In one embodiment the visual surrounding of the shape of the occupant in an image may be used to assist in determining the body position of the occupant. For example, the shape of an occupant in a 2D top view image may be similar to the shape of a standing occupant however based on the visual surrounding of the shape of the occupant it may be determined that the person is sitting, not standing.
Thus, in one embodiment the method may include tracking an occupant through images of the space and detecting the occupant's body position to deduce, for example, if the occupant is sitting, where he is sitting and for how long is he sitting, and creating a characterization which includes this information.
In some embodiment the characterization is created based on additional, not image-based information, such as audio information as described above.
In one embodiment, which is schematically illustrated in
A map may include, for example, a list of coordinates based on the tracking of the occupant. In some embodiments one or more of the coordinates may be associated with a time. In some embodiments a map may include a graphical presentation of the coordinates.
In one example, the occupant is assigned a work station in the space so as to minimize the occupant's paths in the space. For example, if the map created for an occupant shows most paths of the occupant to be within a certain area of the space, then the occupant is assigned a work station within that area to minimize the occupant's paths in the office.
In a setting where allocation of work stations is required, such as in an office, there are typically more than one occupant in need of a work station. Thus, attaching an identity (e.g., an identifying tag) to an occupant may assist in differentiating between occupants and connecting between an occupant and the occupant characterization and/or map of paths.
In one embodiment the method includes tracking an occupant in a first set of images and attaching an identity to the tracked occupant. The method further includes detecting the identified occupant in a second, later, set of images and assigning the occupant detected in the second set of images a work station based on the identity of the occupant.
In one example, which is schematically illustrated in
In one embodiment different occupant characterization groups are created. An occupant is assigned to an occupant characterization group based on his characterization and the occupant is then assigned a workstation in the space based on the occupant characterization group of the occupant.
In one embodiment which is schematically illustrated in
The characterizations and/or group may be maintained in a database, e.g., database 12. Typically an identification is attached to the occupant and saved, e.g., in look-up table 14.
If the occupant is identified in a second set of images of the space (512), then the occupant is assigned a work station according to his group (514). If the occupant is not identified in the second set of images (512) then he is tracked and assigned to a group as described above.
A group may be created, for example, based on an occupant's behavior or characteristics, e.g., based on how noisy the occupant is. In another example, similar maps of occupant paths may be included in the same group.
In one embodiment, a history log of the occupant is created based on the tracking of the occupant. When the occupant is subsequently identified (in a second, later, set of images), the occupant is assigned a work station in the space based on the history log.
In one example, which is schematically illustrated in
The history log may be maintained in a database, e.g., database 12 and the identification attached to the occupant may be saved, e.g., in look-up table 14.
If the occupant is identified in a second, subsequent, set of images of the space (612), then the occupant is assigned a work station according to his history log (614). If the occupant is not identified in the second set of images (612) then he is tracked, identified and a history log is created for him, as described above.
In some embodiments the method includes detecting a shape of an occupant in the subsequent set of images (e.g., based image analysis techniques as described above) and identifying the shape of the occupant as the occupant, namely, attaching an identity to the shape of the occupant. For example, an identity may be attached to the shape of the occupant based on image analysis of the subsequent set of images (e.g., by applying face recognition algorithms on the subsequent set of images to identify a face of an occupant detected in the first set of images).
The history log may include occupant interactions in the space. For example, the history log may include occupant paths in the space, lengths and duration of the paths, locations frequently visited by the occupant, time of arrival/departure at work station, frequency of departures from the work station, etc.
In one embodiment the history log may include non-image-based information related to the occupant (e.g., audio information, as described above).
Assigning a work station based on information deduced or calculated from tracking of the occupant, such as described herein, ensures a better fit between the occupant's needs and the assigned work station, providing the occupant with a uniquely suitable work station.
In one embodiment, an example of which is schematically illustrated in
An occupant is tracked, e.g., by processor 102, in a set of images (e.g., obtained by image sensor 103) of the space to obtain locations of the occupant over time by applying a computer vision algorithm (702). Typically, the computer vision algorithm, such as an appearance model, enables differentiating between occupants, locating the same occupant in images of the space and creating an occupant specific history log (704). In one embodiment the history log of the occupant includes information such as locations of the occupant in the space over time. The method further includes the step of comparing the history logs of a plurality of occupants (706) to extract interaction points between the plurality of occupants. A map of occupant interactions is then created from the interaction points (708).
Interaction points typically include locations in the space in which predefined conditions are fulfilled.
For example, interaction points may include locations in the space in which a number of occupants, which is above a specified threshold, are located within a defined time period. In another embodiment interaction points include locations in the space in which at least two occupants are located for a time period over a specified threshold. In another embodiment, interaction points include locations in the space in which at least two occupants are situated at a distance within a pre-determined range, from each other, possibly, for a time period over a specified threshold. For example, a distance within a pre-determined range may include a distance that is above and/or below a threshold.
For example, building space management personnel may be interested in identifying locations in the building where 3 or more people convene for a time period of over 5 minutes, or, for example, a time period in between 5 and 10 minutes. Thus, a threshold of 3 occupants may be specified (e.g., through building management software) and a time period threshold of 5 minutes or a range of between 5 and 10 minutes may be specified to identify the interaction points of interest to the building management personnel.
In other embodiments interaction points may include locations in the space in which at least two occupants are located within a defined time period, in a repetitive pattern. For example, an interaction point may include a location where, for example, 100 or more people, convene every Sunday.
In other embodiments, interaction points may include locations within the space in which there is compliance or non-compliance with physical distancing rules, e.g., as defined by building space management. For example, interaction points may include locations in which at least two occupants are situated at a distance which is below 2 meters from each other. In another example, an interaction point may include a location in which a number of occupants is above a threshold (e.g., above 10 people).
In some embodiments, history logs, which include locations of occupants in the space over time, may include locations in the space where no occupants ever pass or stop, or for example, where no occupants pass or stop on Sundays. Thus, interaction points may be locations in which no occupants or a number of occupants below a predetermined threshold are detected.
In some embodiments processor 102 may use a shape detection algorithm to detect if a body position of the occupant in one or more images of the space, is sitting or standing and may detect interaction points as location in the space in which a sitting body position of an occupant is detected.
A summation or other aggregation of interaction points detected according to embodiments of the invention, may be used to create a map of occupants' interactions in the space.
Embodiments of the invention enable receiving information from a large number of interactions between anonymized occupants and their usage of a space and may enable discovering unknown trends and patterns to better manage and utilize space.
| Number | Date | Country | Kind |
|---|---|---|---|
| 249077 | Nov 2016 | IL | national |
This application is a Continuation-in-Part of U.S. patent application Ser. No. 15/810,475, filed on Nov. 13, 2017, the disclosure of which is incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 15810475 | Nov 2017 | US |
| Child | 16882560 | US |
