WO2024183130A1 - 旷场偏好区域确定方法、装置、设备及存储介质 - Google Patents

旷场偏好区域确定方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2024183130A1
WO2024183130A1 PCT/CN2023/088453 CN2023088453W WO2024183130A1 WO 2024183130 A1 WO2024183130 A1 WO 2024183130A1 CN 2023088453 W CN2023088453 W CN 2023088453W WO 2024183130 A1 WO2024183130 A1 WO 2024183130A1
Authority
WO
WIPO (PCT)
Prior art keywords
region
area
sub
target object
segmented
Prior art date
Application number
PCT/CN2023/088453
Other languages
English (en)
French (fr)
Inventor
徐阳
王枫
Original Assignee
中国科学院深圳先进技术研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国科学院深圳先进技术研究院 filed Critical 中国科学院深圳先进技术研究院
Publication of WO2024183130A1 publication Critical patent/WO2024183130A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q90/00—Systems or methods specially adapted for administrative, commercial, financial, managerial or supervisory purposes, not involving significant data processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items

Definitions

  • the present application relates to the field of data processing technology, and for example, to a method, device, equipment and storage medium for determining an open field preference area.
  • the open field experiment is a classic animal behavior paradigm. By detecting the spontaneous activity of animals in a specific specification behavior experiment box, the spontaneous activity, anxiety level and other behavioral parameters of rodents are evaluated.
  • the open field experiment usually uses a square or round box, in which rodents can move freely. The animal's activities are recorded by a camera above the box, and the behavioral experiment data can be obtained by analyzing these video data.
  • the classic method of dividing the activity areas of experimental animals is based on the spatial structure of the experimental box, dividing the open field into a central area and an edge area, or into a central area, four boundary areas and four corner areas, or into square rings with the same internal width, etc.
  • the above-mentioned structural open field area division method has strong subjectivity and variability in the area division of the central area, which is prone to misleading experimental results and leads to unreasonable determination of the preferred area.
  • the present application provides a method, device, equipment and storage medium for determining an open field preference area to solve the technical problem that the determination of the preference area is unreasonable due to artificial division of the area, eliminate the interference of human subjective factors on the results, and make the division of the preference area more reasonable and accurate.
  • a method for determining an open field preference area comprising:
  • a target area boundary is determined based on the relationship curve, and a preferred area of the target object is determined based on the target area boundary.
  • an open field preference region device comprising:
  • An activity position information acquisition module configured to acquire multiple activity position information of a target object in an open field
  • a segmented region residence time determination module is configured to determine the segmented region residence time of the target object in each segmented region based on sub-region iteration through the multiple activity position information, and to construct a relationship curve between the segmented region residence time in each segmented region and the change of the corresponding segmented region boundary, wherein the segmented region is obtained based on the iterative sub-region segmentation in the sub-region iteration;
  • the preferred region determination module is configured to determine a target region boundary based on the relationship curve, and determine a preferred region of the target object based on the target region boundary.
  • an electronic device comprising:
  • the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the open field preference area method described in any embodiment of the present application.
  • a computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the open field preference area method described in any embodiment of the present application when executed.
  • FIG1 is a flow chart of a method for determining an open field preference region provided in Example 1 of the present application.
  • FIG2a is a flow chart of an open field preference area method provided in Example 2 of the present application.
  • FIG2b is a schematic diagram of a preferred area division result provided in Example 2 of the present application.
  • FIG2c is a schematic diagram of another preferred area division result provided in Example 2 of the present application.
  • FIG2d is a schematic diagram of another preferred area division result provided in Example 2 of the present application.
  • FIG3 is a schematic diagram of the structure of an open field preference area determination device provided in Example 3 of the present application.
  • FIG4 is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present application.
  • FIG1 is a flow chart of an open field preference region method provided in Embodiment 1 of the present application. This embodiment is applicable to the case of determining the preference region of a target object in an open field.
  • the method can be executed by an open field preference region device, which can be implemented in the form of hardware and/or software, and can be configured in an electronic device. As shown in FIG1 , the method includes:
  • the target object may be an active object in an open field, such as a rodent or the like.
  • this embodiment proposes a data-driven open field preference regional division method to obtain more accurate evaluation results.
  • the embodiment of the present application generates spontaneous position preference regions based on large-scale data to determine the range and boundaries of different preference regions of the target object in the open field box.
  • the activity analysis of the target object in the open field is usually based on the activity video of the target object.
  • the activity of the target object in the open field can be recorded in real time by a camera in the open field. After recording for a certain period of time, the video information shot in the period of time is obtained, and the activity position information of the target object in the open field at different time points is determined based on the shot video information.
  • cameras at multiple angles can be set up to shoot the open field area, and continuous activity videos of the target object at multiple angles in the open field shot by cameras at different angles can be obtained. Then, time alignment is performed based on the activity videos at multiple angles. For multiple activity videos corresponding to the same time point, a multi-eye vision three-dimensional reconstruction method is used to reconstruct the activity position information of the target object corresponding to each activity frame in the activity video.
  • the activity position information of the target object may be the activity position information of the feature points of the target object.
  • the feature points of the target object may be determined based on the target object. For example, assuming that the target object is a rodent, the back points of the rodent are usually used as its feature points. That is, the back points of the target object are obtained.
  • the location information of the part is used as the activity location information of the target object.
  • the activity position information of the target object in all the activity frames may be obtained to divide the preferred area, or the activity position information of the target object in some of the activity frames may be obtained to divide the preferred area.
  • the step of obtaining multiple activity position information of the target object in the open field includes:
  • Acquire an activity video of the target object in an open field perform posture estimation on a plurality of activity frames in the activity video, and determine posture estimation results of the plurality of activity frames;
  • the multiple credible activity frames are three-dimensionally reconstructed to determine activity position information of the target object in the multiple credible activity frames.
  • the posture estimation of the active frame in the active video can be performed before determining the active position information of the target object. If the credibility of the posture estimation is high, it can be determined that the data of the current active frame is accurate. Otherwise, or the credibility of the posture estimation is low, it can be determined that the data of the current active frame is inaccurate.
  • posture estimation is a computer vision technology widely used in the field of behavioral research, which mainly uses tools such as cameras, sensors and computer algorithms to identify the posture of animals.
  • the posture estimation method of the target object can be implemented based on the posture estimation method of the relevant technology.
  • the posture estimation of the target object can be performed by a monocular vision posture estimation method, a binocular vision posture estimation method, a depth sensor posture estimation method, a deep learning posture estimation method and the like.
  • a posture estimation technology based on deep learning can be used to determine the position information of the feature points of the target object to improve the accuracy of the feature point position information.
  • four cameras with 90 degrees (°) to each other can be used to simultaneously record the activities of the target object in the open field, so that images of four different perspectives can be obtained at the same time, and a multi-view 3D reconstruction method is used to perform 3D reconstruction of the target object's feature points.
  • the 3D reconstruction method can be selected based on the behavioral characteristics of the target object.
  • a method based on feature point 3D reconstruction can be used. This method uses the internal and external parameters of each camera and the image coordinates of the registered feature points to perform 3D reconstruction on the feature points.
  • the coordinate information of a large number of target object feature points in three-dimensional space can provide low-error and reliable data support for subsequent boundary fitting.
  • S120 Based on sub-region iteration, determine the segmented region residence time of the target object in each segmented region through the multiple activity position information, and construct a relationship curve between the segmented region residence time in each segmented region and the change of the corresponding segmented region boundary, wherein the segmented region is obtained based on the iterative sub-region segmentation in the sub-region iteration.
  • sub-region iteration can be understood as an operation of iteratively calculating the target object's residence time in the segmented region corresponding to different sub-regions based on the initial sub-region.
  • the initial sub-region is sub-region 1 (a circle with a radius of 18 cm)
  • the target object's residence time in the segmented region in sub-region 1 is calculated
  • sub-region 2 a circle with a radius of 28 cm
  • the target object's residence time in the segmented region the ring between sub-region 1 and sub-region 2 in sub-region 2 is calculated.
  • the above iterative operation is repeated until the iterated sub-region can cover the open field area, the iterative operation is terminated, and the target object's residence time in the segmented region in each segmented region can be obtained.
  • the target object's activity preference area can be analyzed according to the difference of the segmented area residence time of adjacent segmented areas, so as to divide the preference area. It can be understood that, assuming that the difference of the segmented area residence time of adjacent segmented areas is not large, it can be determined that the target object has the same preference for the two adjacent segmented areas. If the difference of the segmented area residence time of adjacent segmented areas is large, it can be determined that the target object prefers the segmented area with a long segmented area residence time. Based on this, the preference of the target object can be judged according to the segmented area residence time of each segmented area, so as to divide the preference area.
  • the step of determining the segmented region residence time of the target object in each segmented region through the multiple activity position information based on the sub-region iteration includes:
  • a current segmented region corresponding to the current sub-region is determined, and a segmented region residence time of the target object in the current segmented region is determined based on the activity position information of the target object in the current sub-region.
  • an iteration distance may be pre-set, and when performing sub-region iteration, the sub-region iteration is performed based on the iteration distance.
  • the iteration distance r may be set, and the radius of the sub-region after each iteration is set to be r larger than the radius of the sub-region before the iteration, and the segmented region residence time corresponding to each iteration is calculated based on the above iteration method.
  • determining the current The segmented area determining the segmented area residence time of the target object in the current segmented area based on the activity position information of the target object in the current sub-area, comprises:
  • the difference between the current sub-region residence time and the previous sub-region residence time of the target object in the previous iteration sub-region is used as the segmented region residence time.
  • the preference of the target object in the iteration process is analyzed by the residence time difference of adjacent iteration sub-regions. Based on this, the region difference between the current iterated sub-region and the previous iterated sub-region is used as the current segmented region, and the difference between the current sub-region residence time of the target object in the current sub-region and the previous sub-region residence time of the target object in the previous iterated sub-region is used as the segmented region residence time.
  • the residence time of the target object in the current sub-region can be obtained by counting the activity position information of the target object in the current sub-region.
  • the current sub-region is a circle with a radius of R
  • the current sub-region residence time is K
  • the previous iterative sub-region is a circle with a radius of R-r
  • the previous sub-region residence time is M
  • the ring with an inner diameter of R-r and an outer diameter of R is used as the current segmentation region
  • K-M is used as the segmentation region residence time of the current segmentation region.
  • a relationship curve can be drawn based on the segmented region residence time of each segmented region.
  • the boundary of the segmented region can be used as the horizontal coordinate
  • the segmented region residence time of the segmented region can be used as the corresponding vertical coordinate to draw a curve to obtain a relationship curve of the segmented region residence time changing with the segmented region boundary.
  • the shape of the sub-region can be set according to actual needs.
  • the shape of the sub-region is circular.
  • the relationship curve drawn by the circular sub-region is in a double-peaked form, and the drawn preference area is more accurate.
  • S130 Determine a target area boundary based on the relationship curve, and determine a preferred area of the target object based on the target area boundary.
  • the obtained relationship curve is a segmented straight line with multiple inflection points.
  • the range and boundaries of the target object in different preferred areas of the open field box can be determined.
  • the segmentation curve boundary corresponding to the inflection point can be used as the target area boundary to divide the preferred area.
  • determining a target area boundary based on the relationship curve, and determining a preferred area of the target object based on the target area boundary includes:
  • the reference area is divided based on the target boundary, and the divided area is used as the preferred area of the target object.
  • the boundary of the segmentation area corresponding to the inflection point of the relationship curve can be used as the target boundary, and then the reference area can be divided based on the target boundary to obtain multiple preference areas.
  • the preference degree of each preference area can be determined according to the position and direction of the segmentation boundary corresponding to the preference area in the relationship curve. After dividing multiple preference areas and determining the preference degree of each preference area, the multiple preference areas can be reversely sorted according to the preference degree, and the order of preference areas with lower preference degrees can be obtained.
  • the reference area is a divided open field area or an original open field area. That is, the original open field area can be taken as an example, and the target boundary can be used to divide the original open field area to obtain the preferred area; the above division method can also be combined with the division method of the related technology, and the original open field area is first divided by artificial division (such as dividing it into a central area and an edge area), and then the target boundary is used to divide the open field area after artificial division to obtain the preferred area.
  • artificial division such as dividing it into a central area and an edge area
  • the original open field area is used as the reference area for division, so that the obtained preference area is completely determined based on the activity information of the target object, and the divided open field area is used as the reference area for division, so that the obtained preference area can take into account other factors besides the activity data. Any of the above division methods can be selected according to actual needs.
  • the technical solution of this embodiment is to obtain multiple activity position information of the target object in the open field; based on sub-region iteration, determine the segmented region residence time of the target object in each segmented region through the multiple activity position information, and construct a relationship curve between the segmented region residence time in each segmented region and the change of the corresponding segmented region boundary, wherein the segmented region is obtained based on the iterative sub-region segmentation in the sub-region iteration; determine the target region boundary based on the relationship curve, and determine the target object's preferred region based on the target region boundary.
  • Fig. 2a is a flow chart of an open field preference region method provided in Example 2 of the present application.
  • This embodiment provides an optional embodiment based on the above embodiment.
  • This embodiment takes the target object as a rodent as an example to exemplify the division of the rodent's preference region in the open field.
  • the embodiment of the present application first estimates the position of the back point of the rodent in the open field using a posture estimation method, and uses multi-eye vision to reconstruct the position of the back point of the rodent in three-dimensional space and output its coordinates; after obtaining a large number of coordinates of the back points of the rodent, after some preprocessing operations, sub-region iterative calculations are performed to generate the residence time of the rodent in different sub-regions; finally, a piecewise straight line (relationship curve) of the residence time of the rodent in the open field is fitted, and the inflection point of the straight line is calculated. Analyze and determine their spontaneous location preference areas in the open field.
  • the method includes:
  • a high frame rate and high resolution camera can be used to obtain better results.
  • the images recorded by the high frame rate camera also provide convenient conditions for collecting a large number of pictures of rodents in the open field.
  • the posture estimation method based on deep learning can be used for posture estimation.
  • the data source obtained by the embodiment of the present application is not limited to the behavioral data of animals, but can also include behavioral data of other model animals, as long as it is collected in a specific open field box.
  • any collected behavioral data can be analyzed using the method provided in the embodiment of the present application.
  • the use of four cameras can simultaneously obtain images from four different perspectives, which can help reconstruct the dorsal points of rodents with high accuracy.
  • the location preference areas of rodents in an open field are generated by fitting a large amount of data, and the boundary fitting is performed using geometric analysis and statistical analysis methods.
  • the present application uses the coordinates of a large number of back points to iteratively generate the residence time of rodents in different segmented areas in an open field, and iterative calculations are performed through different sub-areas such as square sub-areas, circular sub-areas, and diamond sub-areas. Compared with sub-areas of other shapes, the results calculated from the circular area are more reliable, and the fitting curve of the residence time of rodents in the open field is in the form of a double peak, which is more suitable for determining the boundary of the preferred area.
  • the sub-areas can also be defined as some specific shapes in the three-dimensional space.
  • S240 Determine the boundaries of different location preference areas of rodents in an open field.
  • a segmented straight line graph can be generated in which the residence time changes as the length of the iteration area increases.
  • Figure 2b is a schematic diagram of a preference area division result provided in Example 2 of the present application.
  • Figure 2c is another schematic diagram of a preference area division result provided in Example 2 of the present application.
  • Figure 2b is the preference area division result of a 50 ⁇ 50 cm open field
  • Figure 2c is the preference area division result of a 40 ⁇ 40 cm open field.
  • the ratio of the radius of the central area to the side length of the open field is about 0.73 to 0.75.
  • the corner area is divided separately, the fan-shaped corner area is the primary preference area of the target object (such as a mouse), and the remaining part of the corner is the secondary preference area.
  • Figure 2d is another schematic diagram of the preferred area division result provided in Example 2 of the present application.
  • the open field is first divided into a central area and an edge area.
  • the area division is performed based on the target boundary, and the open field can be divided into 4 different areas: the primary preference area D, the secondary preference area C, the edge area B, and the central area A.
  • the technical solution of this embodiment uses the method of back point posture estimation and three-dimensional reconstruction to analyze behavioral data to determine the preferred area, retaining the most instinctive location preference characteristics of animals in the open field, and through automated data analysis, reducing the influence of the experimenter's subjective factors on the experimental results, ensuring the stability of the open field experiment evaluation results.
  • this method has high generalization and can be transferred and used in open fields of various sizes.
  • FIG3 is a schematic diagram of the structure of an open field preference region device provided in Example 3 of the present application. As shown in FIG3 , the device includes:
  • An activity position information acquisition module 310 is configured to acquire a plurality of activity position information of a target object in an open field
  • a segmented region residence time determination module 320 is configured to determine the segmented region residence time of the target object in each segmented region based on sub-region iteration through the multiple activity position information, and construct a relationship curve between the segmented region residence time in each segmented region and the change of the corresponding segmented region boundary, wherein the segmented region is obtained based on the iterative sub-region segmentation in the sub-region iteration;
  • the preference region determination module 330 is configured to determine a target region boundary based on the relationship curve, and determine a preference region of the target object based on the target region boundary.
  • the technical solution of this embodiment is to obtain multiple activity position information of the target object in the open field through the activity position information acquisition module; the segmented area residence time determination module determines the segmented area residence time of the target object in each segmented area through the multiple activity position information based on sub-area iteration, and constructs a relationship curve between the segmented area residence time in each segmented area and the change of the corresponding segmented area boundary, wherein the segmented area is obtained based on the iterative sub-area segmentation in the sub-area iteration; the preferred area determination module determines the target area boundary based on the relationship curve, and determines the preferred area of the target object based on the target area boundary.
  • the segmented region residence time determination module 320 includes:
  • the sub-region iteration unit is set to take the initial sub-region as the benchmark and perform the iteration based on the set iteration distance. Sub-region iteration;
  • the residence time determination unit is configured to determine the current segmented area corresponding to the current sub-area iterated to, and determine the segmented area residence time of the target object in the current segmented area based on the activity position information of the target object in the current sub-area.
  • the residence time determination unit is configured to determine the segmented region residence time of the target object in each segmented region in the following manner:
  • the difference between the current sub-region residence time and the previous sub-region residence time of the target object in the previous iteration sub-region is used as the segmented region residence time.
  • the preference region determination module 330 is configured to determine the preference region of the target object according to the relationship curve in the following manner:
  • the reference area is divided based on the target boundary, and the divided area is used as the preferred area of the target object.
  • the reference area is a divided open field area or an original open field area.
  • the sub-region is circular in shape.
  • the activity position information acquisition module 310 is configured to obtain the activity position information of the target object in the open field in the following manner:
  • Acquire an activity video of the target object in an open field perform posture estimation on a plurality of activity frames in the activity video, and determine posture estimation results of the plurality of activity frames;
  • the multiple credible activity frames are three-dimensionally reconstructed to determine activity position information of the target object in the multiple credible activity frames.
  • the open field preference area device provided in the embodiments of the present application can execute the open field preference area method provided in any embodiment of the present application, and has the corresponding functional modules and beneficial effects of the execution method.
  • FIG4 is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present application.
  • the electronic device 10 can be used to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers.
  • Electronic devices may also refer to various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices.
  • the components shown herein, their connections and relationships, and their functions are only examples.
  • the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11 in communication, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor 11, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the ROM 12 or the computer program loaded from the storage unit 18 to the RAM 13.
  • the RAM 13 various programs and data required for the operation of the electronic device 10 can also be stored.
  • the processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14.
  • An input/output (I/O) interface 15 is also connected to the bus 14.
  • a number of components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc.
  • the communication unit 19 allows the electronic device 10 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.
  • the processor 11 may be a variety of general and/or special processing components with processing and computing capabilities. Some examples of the processor 11 may include a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSP), and any appropriate processors, controllers, microcontrollers, etc.
  • the processor 11 executes the various methods described above, such as the open field preference area method.
  • the open field preference area method may be implemented as a computer program that is tangibly contained in a computer-readable storage medium, such as a storage unit 18.
  • part or all of the computer program may be loaded and/or installed on the electronic device 10 via the ROM 12 and/or the communication unit 19.
  • the processor 11 may be configured to perform the open field preference area method by other appropriate means (e.g., by means of firmware).
  • Various implementations of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard parts (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof.
  • FPGAs field-programmable gate arrays
  • ASICs application-specific integrated circuits
  • ASSPs application-specific standard parts
  • SOCs systems on chips
  • CPLDs complex programmable logic devices
  • implementations may include: being implemented in one or more computer programs that can be executed and/or interpreted on a programmable system that includes at least one programmable processor, which may be A special purpose or general purpose programmable processor may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
  • the computer program for implementing the open field preference region method of the present application can be written in a combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that when the computer program is executed by the processor, the functions/operations specified in the flow chart and/or block diagram are implemented.
  • the computer program can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.
  • Embodiment 5 of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a processor to execute an open field preference area method, the method comprising:
  • a target area boundary is determined based on the relationship curve, and a preferred area of the target object is determined based on the target area boundary.
  • a computer readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or apparatus.
  • a computer readable storage medium may include an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or a suitable combination of the above.
  • a computer readable storage medium may be a machine readable signal medium.
  • Examples of machine readable storage media may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (such as an electronic programmable read-only memory (EPROM) or a flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or a suitable combination of the above.
  • RAM random access memory
  • ROM read-only memory
  • EPROM electronic programmable read-only memory
  • CD-ROM portable compact disk read-only memory
  • CD-ROM compact disk read-only memory
  • magnetic storage device or a suitable combination of the above.
  • the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD), a monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the electronic device.
  • a display device e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD), a monitor
  • a keyboard and a pointing device e.g., a mouse or a trackball
  • Other types of devices may also be used to provide interaction with a user. interaction; for example, the feedback provided to the user can be any form of sensory feedback (for example, visual feedback, auditory feedback, or tactile feedback); and the input from the user can be received in any form (including acoustic input, voice input, or tactile input).
  • the systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components.
  • the components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: Local Area Network (LAN), Wide Area Network (WAN), blockchain network, and the Internet.
  • a computing system may include a client and a server.
  • the client and the server are generally remote from each other and usually interact through a communication network.
  • the client and server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other.
  • the server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and virtual private servers (VPS) services.
  • VPN virtual private servers

Landscapes

  • Business, Economics & Management (AREA)
  • Economics (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

本申请公开了一种旷场偏好区域确定方法、装置、设备及存储介质。所述方法包括：获取目标对象在旷场内的多个活动位置信息；基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。

Description

旷场偏好区域确定方法、装置、设备及存储介质
本公开要求在2023年3月6日提交中国专利局、申请号为202310243859.4的中国专利申请的优先权，该申请的全部内容通过引用结合在本申请中。
技术领域
本申请涉及数据处理技术领域，例如涉及一种旷场偏好区域确定方法、装置、设备及存储介质。
背景技术
旷场实验，是一种经典的动物行为学范式。通过检测动物在特定规格行为实验箱体中的的自发活动，评估啮齿类动物的自发活动性、焦虑水平等行为参数。旷场实验通常采用方形或圆形的箱体，啮齿类动物可以在其中自由活动，通过箱体内上方的摄像头记录下动物的活动，通过分析这些视频数据可以获得行为实验数据。
经典的划分实验动物活动区域方法是以实验箱的空间结构为基础的划分方式，将旷场分为中央区域和边缘区域，或划分为中心区域、四个边界区域和四个角落区域，或划分为内部宽度相同的方形环等。
在实现本申请的过程中，发现相关技术中至少存在以下技术问题：上述提到的结构性旷场区域划分方法对中心区域的面积划分的人为主观性强、变动性大，容易出现误导性的实验结果，导致偏好区域的确定不合理。
发明内容
本申请提供了一种旷场偏好区域确定方法、装置、设备及存储介质，以解决人为划分区域导致偏好区域的确定不合理的技术问题，实现排除人为主观因素对结果的干扰，使得偏好区域的划分更加合理、准确。
根据本申请的一方面，提供了一种旷场偏好区域确定方法，包括：
获取目标对象在旷场内的多个活动位置信息；
基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；
基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。
根据本申请的另一方面，提供了一种旷场偏好区域装置，包括：
活动位置信息获取模块，设置为获取目标对象在旷场内的多个活动位置信息；
分割区域驻留时间确定模块，设置为基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；
偏好区域确定模块，设置为基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。
根据本申请的另一方面，提供了一种电子设备，所述电子设备包括：
至少一个处理器；以及
与所述至少一个处理器通信连接的存储器；其中，
所述存储器存储有可被所述至少一个处理器执行的计算机程序，所述计算机程序被所述至少一个处理器执行，以使所述至少一个处理器能够执行本申请任一实施例所述的旷场偏好区域方法。
根据本申请的另一方面，提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机指令，所述计算机指令用于使处理器执行时实现本申请任一实施例所述的旷场偏好区域方法。
附图说明
为了说明本申请实施例中的技术方案，下面将对实施例描述中所需要使用的附图作介绍，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本申请实施例一提供的一种旷场偏好区域确定方法的流程图；
图2a是本申请实施例二提供的一种旷场偏好区域方法的流程图；
图2b是本申请实施例二提供的一种偏好区域划分结果示意图；
图2c是本申请实施例二提供的又一种偏好区域划分结果示意图；
图2d是本申请实施例二提供的又一种偏好区域划分结果示意图；
图3是本申请实施例三提供的一种旷场偏好区域确定装置的结构示意图；
图4是本申请实施例四提供的一种电子设备的结构示意图。
具体实施方式
为了使本技术领域的人员更好地理解本申请方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行描述，所描述的实施例仅仅是本申请一部分的实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
需要说明的是，本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便这里描述的本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于列出的那些步骤或单元，而是可包括没有列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
实施例一
图1是本申请实施例一提供的一种旷场偏好区域方法的流程图，本实施例可适用于确定目标对象在旷场中的偏好区域的情况，该方法可以由旷场偏好区域装置来执行，该旷场偏好区域装置可以采用硬件和/或软件的形式实现，该旷场偏好区域装置可配置于电子设备中。如图1所示，该方法包括：
S110、获取目标对象在旷场内的多个活动位置信息。
在本实施例中，目标对象可以为旷场内的活动对象，如啮齿类动物等。
为了解决人为划分区域导致的区域划分不准确的技术问题，本实施例提出了一种数据驱动的旷场偏好区域划分方法，以获得更精准的评估结果。整体来说，本申请实施例基于大规模数据生成自发位置偏好区域，以确定目标对象在旷场箱体不同偏好区域的范围及边界。
一般来说，对旷场内目标对象的活动分析通常基于拍摄的目标对象的活动视频。可选的，可以通过旷场内摄像头实时记录目标对象在旷场内的活动。并在记录一定时间段后，获取该时间段所拍摄的视频信息，基于所拍摄的视频信息确定不同时间点目标对象在旷场内的活动位置信息。
可选的，可以设置多个角度的相机，对旷场区域进行拍摄，获取不同角度相机拍摄的目标对象在旷场内的多个角度的连续活动视频，然后基于多个角度的活动视频进行时间对齐，针对同一时间点对应的多个活动视频，采用多目视觉三维重建方法，重建出活动视频中每个活动帧对应的目标对象的活动位置信息。
其中，目标对象的活动位置信息可以为目标对象的特征点的活动位置信息。目标对象的特征点可以基于目标对象确定。示例性的，假设目标对象为啮齿类动物，通常将啮齿类动物的背部点作为其特征点，也就是说，获取目标对象背 部的位置信息作为目标对象的活动位置信息。
需要说明的是，可以获取所有活动帧中的目标对象的活动位置信息，以进行偏好区域的划分，也可以获取部分活动帧中的目标对象的活动位置信息，进行偏好区域的划分。
在上述方案的基础上，所述获取目标对象在旷场内的多个活动位置信息，包括：
获取所述目标对象在旷场内的活动视频，对所述活动视频中的多个活动帧进行姿态估计，确定所述多个活动帧的姿态估计结果；
基于所述多个活动帧的姿态估计结果对所述多个活动帧进行筛选，得到多个可信活动帧；
对所述多个可信活动帧进行三维重建，确定所述多个可信活动帧中所述目标对象的活动位置信息。
为了避免不准确的位置信息导致的区域划分不准确，在本实施例中，在确定目标对象的活动位置信息之前可以对活动视频中的活动帧进行姿态估计，若姿态估计的可信度较高，则可以认定当前活动帧的数据准确，反之，或姿态估计的可信度较低，则可以认定当前活动帧的数据不准确。
基于此，可以首先对多个活动帧进行姿态估计，获得每个活动帧中目标对象的姿态估计结果(包括姿态估计可信度)，将可信度高于设定阈值的活动帧作为可信活动帧，然后对可信活动帧进行三维重建，得到可信活动帧中目标对象的位置信息作为目标对象的活动位置信息。通过对目标对象特征点进行姿态估计，可以获得目标对象在不同相机中的位置坐标信息，并利用多目相机三维重建技术可以确定目标对象特征点在三维空间中的精确位置。
其中，姿态估计是一种广泛应用于行为学研究领域的计算机视觉技术，主要通过摄像机、传感器和计算机算法等工具来识别动物的姿势。目标对象的姿态估计方法可以基于相关技术的姿态估计方法实现。示例性的，可以通过单目视觉姿态估计方法、双目视觉姿态估计方法、深度传感器姿态估计方法、深度学习姿态估计方法等方法对目标对象进行姿态估计。在本实施例中，可以采用基于深度学习的姿态估计技术来确定目标对象特征点的位置信息，以提高特征点位置信息的准确度。
一个实施方式中，可以采用四个互为90度(°)的摄像机同时记录目标对象在旷场中的活动，以在同一时间能够获得四种不同视角的图像，并使用多目视觉三维重建方法对目标对象特征点进行三维重建。其中，三维重建的方法可以根据目标对象的行为学特征选取。
一个实施例中，考虑到系统复杂度、算法效率、解算问题和环境光线对重建效果的影响，可以采用基于特征点三维重建的方法，该方法利用每个摄像机的内外参数以及配准的特征点的图像坐标，对特征点进行三维重建，通过获取 大量目标对象特征点在三维空间中的坐标信息，能够为后续边界拟合提供低误差、可靠的数据支持。
S120、基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到。
在本实施例中，获得目标对象随时间变化的活动位置信息的大量数据后，基于上述数据进行迭代计算，以确定目标对象在旷场箱体不同区域的驻留时间，从而得到一个准确的拟合曲线，以确定啮齿类动物在旷场箱体的位置偏好区域的边界。
整体来说，子区域迭代可以理解为是以初始子区域为基准，迭代计算目标对象在不同子区域对应分割区域中的分割区域驻留时间的操作。示例性的，如图2b所示，假设初始子区域为子区域1(半径为18cm的圆)，则计算目标对象在子区域1中的分割区域驻留时间，然后基于迭代距离得到子区域2(半径为28cm的圆)，并计算目标对象在子区域2中分割区域(子区域1和子区域2之间的圆环)的分割区域驻留时间，重复执行上述迭代操作，直到迭代到的子区域能够覆盖旷场区域，结束迭代操作，并且能够得到目标对象在每个分割区域中的分割区域驻留时间。
基于上述迭代操作，可以根据相邻分割区域的分割区域驻留时间的差值分析目标对象的活动偏好区域，从而进行偏好区域的划分。可以理解的是，假设相邻分割区域的分割区域驻留时间差别不大，则可以认定目标对象对该两个相邻分割区域的偏好一致，假设相邻分割区域的分割区域驻留时间的差值较大，则可以认定目标对象偏好分割区域驻留时间长的分割区域。基于此，可以根据每个分割区域的分割区域驻留时间判断目标对象的偏好情况，从而进行偏好区域的划分。
在本申请的一种实施方式中，所述基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，包括：
以初始子区域为基准，基于设定迭代距离，进行子区域迭代；
针对迭代到的当前子区域，确定所述当前子区域对应的当前分割区域，基于所述目标对象在所述当前子区域中的活动位置信息确定所述目标对象在所述当前分割区域中的分割区域驻留时间。
可选的，可以预先设置迭代距离，在进行子区域迭代时，基于迭代距离进行子区域的迭代。假设子区域为圆形，则可以设定迭代距离r，设置每次迭代时迭代后的子区域半径比迭代前的子区域半径大r，基于上述迭代方式计算每次迭代对应的分割区域的分割区域驻留时间。
可选的，所述针对迭代到的当前子区域，确定所述当前子区域对应的当前 分割区域，基于所述目标对象在所述当前子区域中的活动位置信息确定所述目标对象在所述当前分割区域中的分割区域驻留时间，包括：
将所述当前子区域与前一迭代子区域之间的区域差作为所述当前分割区域；
基于所述目标对象在所述当前子区域中的活动位置信息，确定当前子区域驻留时间；
将所述当前子区域驻留时间和所述目标对象在前一迭代子区域中的前一子区域驻留时间之间的差值作为所述分割区域驻留时间。
在本实施例中，通过相邻迭代子区域的驻留时间差值对迭代过程中目标对象的偏好进行分析。基于此，将迭代到的当前子区域与前一迭代子区域之间的区域差作为当前分割区域，将目标对象在当前子区域中的当前子区域驻留时间和所述目标对象在前一迭代子区域中的前一子区域驻留时间之间的差值作为所述分割区域驻留时间。
其中，目标对象在当前子区域中的当前子区域驻留时间可以通过目标对象位于当前子区域的活动位置信息统计得到。
示例性的，假设当前子区域为半径R的圆，当前子区域驻留时间为K，前一迭代子区域为半径R-r的圆，前一子区域驻留时间为M，则将内径R-r，外径R的圆环作为当前分割区域，将K-M作为当前分割区域的分割区域驻留时间。
得到每个分割区域的分割区域驻留时间后，可以基于每个分割区域的分割区域驻留时间进行关系曲线的绘制。示例性的，可以将分割区域的边界作为横坐标，将分割区域的分割区域驻留时间作为对应纵坐标，进行曲线绘制，得到分割区域驻留时间随分割区域边界变化的关系曲线。
在本实施例中，子区域的形状可以根据实际需求设置。例如，子区域的形状为圆形。圆形子区域绘制出的关系曲线呈双峰形式，绘制出的偏好区域更加精准。
S130、基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。
在本实施例中，得到的关系曲线为一条具有多个拐点的分段直线，通过分析分段直线拐点前后斜率的关系，可以确定目标对象在旷场箱体不同偏好区域的范围及边界。示例性的，可以将拐点对应的分割曲线边界作为目标区域边界，以划分出偏好区域。
一个实现方式中，基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域，包括：
根据所述关系曲线的拐点确定所述目标区域边界的目标边界；
基于所述目标边界对基准区域进行划分，将划分得到的区域作为所述目标对象的偏好区域。
可选的，可以将关系曲线的拐点对应的分割区域的边界作为目标边界，然后基于目标边界对基准区域划分，得到多个偏好区域。其中，每个偏好区域的偏好程度可以根据该偏好区域对应的分割边界在关系曲线中的位置及走向确定。划分出多个偏好区域，并确定每个偏好区域的偏好程度后，可以根据偏好程度，对多个偏好区域进行逆向排序，可以得到排序越靠后，偏好程度越低的偏好区域顺序。
如图2b所示，以子区域为以旷场中心为圆心的圆形为例，假设关系曲线的拐点对应的分割边界分别为18厘米(cm)、28cm，则可以将18cm、28cm作为目标边界，在基准区域，以基准区域的中心为圆心，分别绘制半径为18cm和28cm的圆，将绘制的圆对基准区域划分得到的区域作为目标对象的偏好区域。
在上述方案的基础上，所述基准区域为已划分的旷场区域或原始旷场区域。也就是说，可以以原始旷场区域为例，采用目标边界对原始旷场区域进行划分，得到偏好区域；也可以将上述划分方法与相关技术的划分方法相结合，先通过人为划分方式对原始旷场区域进行划分(如划分为中心区域和边缘区域)，然后采用目标边界对人为划分后的旷场区域进行划分，得到偏好区域。
以原始旷场区域为基准区域进行划分，使得划分得到的偏好区域完全基于目标对象的活动信息确定，以已划分的旷场区域为基准区域进行划分，使得划分得到的偏好区域能够考虑到除活动数据之外的其他因素。可以根据实际需求选择上述任一划分方式。
本实施例的技术方案，通过获取目标对象在旷场内的多个活动位置信息；基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。通过基于目标对象的活动位置信息进行偏好区域的划分，排除了人为主观因素对结果的干扰，使得偏好区域的划分更加合理、准确，进而使得偏好区域的确定更加准确。
实施例二
图2a是本申请实施例二提供的一种旷场偏好区域方法的流程图，本实施例在上述实施例的基础上，提供了一种可选实施例。本实施例以目标对象为啮齿类动物为例，对啮齿类动物在旷场中的偏好区域划分进行示例性说明。
整体来说，本申请实施例首先利用姿态估计方法估计出啮齿类动物背部点在旷场中的位置，使用多目视觉三维重建出啮齿类背部点在三维空间中的位置并输出其坐标；在获取大量啮齿类动物背部点坐标后，经过部分预处理操作，进行子区域迭代计算生成啮齿类动物在不同子区域内的驻留时间；最后通过拟合啮齿类动物在旷场中的驻留时间分段直线(关系曲线)，对该直线的拐点进 行分析，确定其在旷场中自发的位置偏好区域。
如图2a所示，该方法包括：
S210、啮齿类动物背部点姿态估计。
在本实施例中，可以使用高帧率和高分辨率的相机，以获得更好的效果。同时，高帧率相机记录的图像也为收集大量啮齿类动物在旷场中活动图片提供了便利条件。可以使用基于深度学习的姿态估计方法进行姿态估计。
需要说明的是，本申请实施例获取的数据源并不仅限于动物的行为学数据，还可以包括其他模式动物的行为数据，只要是在一个特定旷场箱体中采集到的即可。理论上，任何采集到的行为学数据都可以采用本申请实施例提供的方法进行分析。
S220、啮齿类动物背部点三维重建。
四个相机的使用可同时获得四个不同的视角的图像，能够帮助重建出高准确性的啮齿类动物背部点。首先，评估啮齿类动物背部点姿态估计的可信度，去除姿态估计中可信度低的视角，并选择超过两个视角的图像进行三维重建；其次，以背部点为主要特征点进行重建(可以参考相关技术的行为分析方法)；最后，通过基于深度学习的姿态估计方法，将需要控制背部点重建的误差在小于3个像素(小于0.1厘米)范围内。
S230、啮齿类动物背部点边界拟合。
通过大量数据拟合生成啮齿类动物在旷场中的位置偏好区域，应用几何分析和统计分析的方法进行边界拟合。本申请利用大量背部点的坐标迭代生成了啮齿类动物在旷场中不同分割区域的驻留时间，通过方形子区域、圆形子区域、菱形子区域等不同子区域进行迭代计算。与其他形状子区域相比，圆形区域计算出来的结果更为可靠，啮齿类动物在旷场中驻留时间的拟合曲线呈双峰值形式，更加适合确定偏好区域边界。在一实施例中，如果计算三维空间内的驻留时间分布，也可以将子区域定义为三维空间内的一些特定的形状。
S240、确定啮齿类动物在旷场中不同位置偏好区域边界。
利用步骤S230迭代计算生成的驻留时间数据，可以生成一条驻留时间随着迭代区域边长增加而变化的分段直线图，通过观察分段直线的斜率的变化情况，判断其中是否存在拐点。如存在拐点，则分析该拐点前后的斜率变化情况，从而确定该拐点是否代表啮齿类动物在旷场中某个位置偏好区域的边界。借助这一泛化性误差(3％)很低的方法，可以划分不同尺寸旷场内部的中心区域。
需要说明的是，经实验，采用本实施例提供的方法对两种不同尺寸的旷场进行偏好区域划分。图2b是本申请实施例二提供的一种偏好区域划分结果示意图。图2c是本申请实施例二提供的又一种偏好区域划分结果示意图。图2b为50×50cm旷场的偏好区域划分结果，图2c为40×40cm旷场的偏好区域划分结果。参照图2b和图2c可以得到，在50×50cm(图2b)、40×40cm(图2c) 尺寸下，中心区域的半径与旷场边长的比例在0.73～0.75左右。并且，图2b和图2c中将角落区域单独划分，扇形的角落区域为目标对象(如小鼠)首要偏好区域、角落的剩余部分为次要偏好区。
在上述基础上，还可以将本实施例提供的区域划分方法与经验结合，进行偏好区域的划分。图2d是本申请实施例二提供的又一种偏好区域划分结果示意图。如图2d所示，在经验的基础上，先将旷场划分为中心区域和边缘区域，在此基础上，基于目标边界进行区域划分，可以将旷场划分成4种不同的区域：首要偏好区域D、次要偏好区域C、边缘区域B以及中心区域A。
本实施例的技术方案，利用动物背部点姿态估计和三维重建方法对行为学数据进行分析，以确定偏好区域，保留了动物在旷场中最本能的位置偏好特征，通过自动化的数据分析，降低了实验者主观因素对实验结果的影响，保证了旷场实验评价结果的稳定性。此外，这种方法具有较高的泛化性，能够在多种不同尺寸的旷场中进行迁移使用。
实施例三
图3是本申请实施例三提供的一种旷场偏好区域装置的结构示意图。如图3所示，该装置包括：
活动位置信息获取模块310，设置为获取目标对象在旷场内的多个活动位置信息；
分割区域驻留时间确定模块320，设置为基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；
偏好区域确定模块330，设置为基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。
本实施例的技术方案，通过活动位置信息获取模块获取目标对象在旷场内的多个活动位置信息；分割区域驻留时间确定模块基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；偏好区域确定模块基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。通过基于目标对象的活动位置信息进行偏好区域的划分，排除了人为主观因素对结果的干扰，使得偏好区域的划分更加合理、准确，进而使得偏好区域的确定更加准确。
在上述实施例的基础上，可选的，分割区域驻留时间确定模块320包括：
子区域迭代单元，设置为以初始子区域为基准，基于设定迭代距离，进行 子区域迭代；
驻留时间确定单元，设置为针对迭代到的当前子区域，确定所述当前子区域对应的当前分割区域，基于所述目标对象在所述当前子区域中的活动位置信息确定所述目标对象在所述当前分割区域中的分割区域驻留时间。
在上述实施例的基础上，可选的，驻留时间确定单元设置为通过以下方式确定目标对象在每个分割区域中的分割区域驻留时间：
将所述当前子区域与前一迭代子区域之间的区域差作为所述当前分割区域；
基于所述目标对象在所述当前子区域中的活动位置信息，确定当前子区域驻留时间；
将所述当前子区域驻留时间和所述目标对象在前一迭代子区域中的前一子区域驻留时间之间的差值作为所述分割区域驻留时间。
在上述实施例的基础上，可选的，偏好区域确定模块330设置为通过以下方式根据关系曲线确定目标对象的偏好区域：
根据所述关系曲线的拐点确定所述目标区域边界的目标边界；
基于所述目标边界对基准区域进行划分，将划分得到的区域作为所述目标对象的偏好区域。
在上述实施例的基础上，可选的，所述基准区域为已划分的旷场区域或原始旷场区域。
在上述实施例的基础上，可选的，所述子区域的形状为圆形。
在上述实施例的基础上，可选的，活动位置信息获取模块310设置为通过以下方式根据目标对象在旷场内的多个活动位置信息：
获取所述目标对象在旷场内的活动视频，对所述活动视频中的多个活动帧进行姿态估计，确定所述多个活动帧的姿态估计结果；
基于所述多个活动帧的姿态估计结果对所述多个活动帧进行筛选，得到多个可信活动帧；
对所述多个可信活动帧进行三维重建，确定所述多个可信活动帧中所述目标对象的活动位置信息。
本申请实施例所提供的旷场偏好区域装置可执行本申请任意实施例所提供的旷场偏好区域方法，具备执行方法相应的功能模块和有益效果。
实施例四
图4是本申请实施例四提供的一种电子设备的结构示意图。电子设备10可以用于表示各种形式的数字计算机，诸如，膝上型计算机、台式计算机、工作台、个人数字助理、服务器、刀片式服务器、大型计算机、和其它适合的计算 机。电子设备还可以表示各种形式的移动装置，诸如，个人数字处理、蜂窝电话、智能电话、可穿戴设备(如头盔、眼镜、手表等)和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例。
如图4所示，电子设备10包括至少一个处理器11，以及与至少一个处理器11通信连接的存储器，如只读存储器(Read Only Memory，ROM)12、随机访问存储器(Random Access Memory，RAM)13等，其中，存储器存储有可被至少一个处理器11执行的计算机程序，处理器11可以根据存储在只读存储器(ROM)12中的计算机程序或者从存储单元18加载到随机访问存储器(RAM)13中的计算机程序，来执行各种适当的动作和处理。在RAM 13中，还可存储电子设备10操作所需的各种程序和数据。处理器11、ROM 12以及RAM 13通过总线14彼此相连。输入/输出(Input/Output，I/O)接口15也连接至总线14。
电子设备10中的多个部件连接至I/O接口15，包括：输入单元16，例如键盘、鼠标等；输出单元17，例如各种类型的显示器、扬声器等；存储单元18，例如磁盘、光盘等；以及通信单元19，例如网卡、调制解调器、无线通信收发机等。通信单元19允许电子设备10通过诸如因特网的计算机网络和/或各种电信网络与其他设备交换信息/数据。
处理器11可以是各种具有处理和计算能力的通用和/或专用处理组件。处理器11的一些示例可以包括中央处理单元(Central Processing Unit，CPU)、图形处理单元(Graphic Process Unit，GPU)、各种专用的人工智能(Artificial Intelligence，AI)计算芯片、各种运行机器学习模型算法的处理器、数字信号处理器(Digital Signal Processing，DSP)、以及任何适当的处理器、控制器、微控制器等。处理器11执行上文所描述的各个方法，例如旷场偏好区域方法。
在一些实施例中，旷场偏好区域方法可被实现为计算机程序，其被有形地包含于计算机可读存储介质，例如存储单元18。在一些实施例中，计算机程序的部分或者全部可以经由ROM 12和/或通信单元19而被载入和/或安装到电子设备10上。当计算机程序加载到RAM 13并由处理器11执行时，可以执行上文描述的旷场偏好区域方法的一个或多个步骤。备选地，在其他实施例中，处理器11可以通过其他适当的方式(例如，借助于固件)而被配置为执行旷场偏好区域方法。
本文中以上描述的系统和技术的各种实施方式可以在数字电子电路系统、集成电路系统、场可编程门阵列(Field-Programmable Gate Array，FPGA)、专用集成电路(Application Specific Integrated Circuit，ASIC)、专用标准产品(Application Specific Standard Parts，ASSP)、芯片上系统的系统(System on Chip，SOC)、负载可编程逻辑设备(Complex Programmable Logic Device，CPLD)、计算机硬件、固件、软件、和/或它们的组合中实现。这些各种实施方式可以包括：实施在一个或者多个计算机程序中，该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释，该可编程处理器可以是 专用或者通用可编程处理器，可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令，并且将数据和指令传输至该存储系统、该至少一个输入装置、和该至少一个输出装置。
用于实施本申请的旷场偏好区域方法的计算机程序可以采用一个或多个编程语言的组合来编写。这些计算机程序可以提供给通用计算机、专用计算机或其他可编程数据处理装置的处理器，使得计算机程序当由处理器执行时使流程图和/或框图中所规定的功能/操作被实施。计算机程序可以完全在机器上执行、部分地在机器上执行，作为独立软件包部分地在机器上执行且部分地在远程机器上执行或完全在远程机器或服务器上执行。
实施例五
本申请实施例五还提供了一种计算机可读存储介质，计算机可读存储介质存储有计算机指令，计算机指令用于使处理器执行一种旷场偏好区域方法，该方法包括：
获取目标对象在旷场内的多个活动位置信息；
基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；
基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。
在本申请的上下文中，计算机可读存储介质可以是有形的介质，其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的计算机程序。计算机可读存储介质可以包括电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备，或者上述内容的合适组合。备选地，计算机可读存储介质可以是机器可读信号介质。机器可读存储介质的示例可以包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(如电子可编程只读存储器(Electronic Programable Read Only Memory，EPROM)或快闪存储器)、光纤、便捷式紧凑盘只读存储器(Compact Disc-Read Only Memory，CD-ROM)、光学储存设备、磁储存设备、或上述内容的合适组合。
为了提供与用户的交互，可以在电子设备上实施此处描述的系统和技术，该电子设备具有：用于向用户显示信息的显示装置(例如，阴极射线管(Cathode Ray Tube，CRT)或者液晶显示器(Liquid Crystal Display，LCD)、监视器)；以及键盘和指向装置(例如，鼠标或者轨迹球)，用户可以通过该键盘和该指向装置来将输入提供给电子设备。其它种类的装置还可以用于提供与用户的交 互；例如，提供给用户的反馈可以是任何形式的传感反馈(例如，视觉反馈、听觉反馈、或者触觉反馈)；并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如，作为数据服务器)、或者包括中间件部件的计算系统(例如，应用服务器)、或者包括前端部件的计算系统(例如，具有图形用户界面或者网络浏览器的用户计算机，用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如，通信网络)来将系统的部件相互连接。通信网络的示例包括：局域网(Local Area Network，LAN)、广域网(Wide Area Network，WAN)、区块链网络和互联网。
计算系统可以包括客户端和服务器。客户端和服务器一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客户端-服务器关系的计算机程序来产生客户端和服务器的关系。服务器可以是云服务器，又称为云计算服务器或云主机，是云计算服务体系中的一项主机产品，以解决传统物理主机与虚拟专用服务器(Virtual Private Server，VPS)服务中，存在的管理难度大，业务扩展性弱的缺陷。
应该理解，可以使用上面所示的各种形式的流程，重新排序、增加或删除步骤。例如，本申请中记载的各步骤可以并行地执行也可以顺序地执行也可以不同的次序执行，只要能够实现本申请的技术方案所期望的结果。

Claims (10)

  1. 一种旷场偏好区域确定方法，包括：
    获取目标对象在旷场内的多个活动位置信息；
    基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；
    基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。
  2. 根据权利要求1所述的方法，其中，所述基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，包括：
    以初始子区域为基准，基于设定迭代距离，进行子区域迭代；
    针对迭代到的当前子区域，确定所述当前子区域对应的当前分割区域，基于所述目标对象在所述当前子区域中的活动位置信息确定所述目标对象在所述当前分割区域中的分割区域驻留时间。
  3. 根据权利要求2所述的方法，其中，所述针对迭代到的当前子区域，确定所述当前子区域对应的当前分割区域，基于所述目标对象在所述当前子区域中的活动位置信息确定所述目标对象在所述当前分割区域中的分割区域驻留时间，包括：
    将所述当前子区域与前一迭代子区域之间的区域差作为所述当前分割区域；
    基于所述目标对象在所述当前子区域中的活动位置信息，确定当前子区域驻留时间；
    将所述当前子区域驻留时间和所述目标对象在前一迭代子区域中的前一子区域驻留时间之间的差值作为所述分割区域驻留时间。
  4. 根据权利要求1所述的方法，其中，所述基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域，包括：
    根据所述关系曲线的拐点确定所述目标区域边界的目标边界；
    基于所述目标边界对基准区域进行划分，将划分得到的区域作为所述目标对象的偏好区域。
  5. 根据权利要求4所述的方法，其中，所述基准区域为已划分的旷场区域或原始旷场区域。
  6. 根据权利要求1所述的方法，其中，所述子区域的形状为圆形。
  7. 根据权利要求1所述的方法，其中，所述获取目标对象在旷场内的多个活动位置信息，包括：
    获取所述目标对象在旷场内的活动视频，对所述活动视频中的多个活动帧进行姿态估计，确定所述多个活动帧的姿态估计结果；
    基于所述多个活动帧的姿态估计结果对所述多个活动帧进行筛选，得到多个可信活动帧；
    对所述多个可信活动帧进行三维重建，确定所述多个可信活动帧中所述目标对象的活动位置信息。
  8. 一种旷场偏好区域装置，包括：
    活动位置信息获取模块，设置为获取目标对象在旷场内的多个活动位置信息；
    分割区域驻留时间确定模块，设置为基于子区域迭代，通过所述多个活动位置信息确定所述目标对象在每个分割区域中的分割区域驻留时间，构建每个分割区域中的分割区域驻留时间随对应分割区域边界变化的关系曲线，其中，所述分割区域基于子区域迭代中的迭代子区域分割得到；
    偏好区域确定模块，设置为基于所述关系曲线确定目标区域边界，基于所述目标区域边界确定所述目标对象的偏好区域。
  9. 一种电子设备，所述电子设备包括：
    至少一个处理器；以及
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述至少一个处理器执行的计算机程序，所述计算机程序被所述至少一个处理器执行，以使所述至少一个处理器能够执行如权利要求1-7中任一项所述的旷场偏好区域方法。
  10. 一种计算机可读存储介质，所述计算机可读存储介质存储有计算机指令，所述计算机指令用于使处理器执行时实现如权利要求1-7中任一项所述的旷场偏好区域方法。
PCT/CN2023/088453 2023-03-06 2023-04-14 旷场偏好区域确定方法、装置、设备及存储介质 WO2024183130A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202310243859.4A CN116664372A (zh) 2023-03-06 2023-03-06 一种旷场偏好区域确定方法、装置、设备及存储介质
CN202310243859.4 2023-03-06

Publications (1)

Publication Number Publication Date
WO2024183130A1 true WO2024183130A1 (zh) 2024-09-12

Family

ID=87717785

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/088453 WO2024183130A1 (zh) 2023-03-06 2023-04-14 旷场偏好区域确定方法、装置、设备及存储介质

Country Status (2)

Country Link
CN (1) CN116664372A (zh)
WO (1) WO2024183130A1 (zh)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150332469A1 (en) * 2014-05-14 2015-11-19 International Business Machines Corporation Static Image Segmentation
CN110599520A (zh) * 2019-08-30 2019-12-20 深圳先进技术研究院 一种旷场实验数据分析方法、系统及终端设备
CN115100231A (zh) * 2022-07-15 2022-09-23 京东城市(北京)数字科技有限公司 一种区域边界的确定方法和装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150332469A1 (en) * 2014-05-14 2015-11-19 International Business Machines Corporation Static Image Segmentation
CN110599520A (zh) * 2019-08-30 2019-12-20 深圳先进技术研究院 一种旷场实验数据分析方法、系统及终端设备
CN115100231A (zh) * 2022-07-15 2022-09-23 京东城市(北京)数字科技有限公司 一种区域边界的确定方法和装置

Also Published As

Publication number Publication date
CN116664372A (zh) 2023-08-29

Similar Documents

Publication Publication Date Title
US10122994B2 (en) Object reconstruction from dense light fields via depth from gradients
US11747898B2 (en) Method and apparatus with gaze estimation
WO2021082635A1 (zh) 一种关注区域检测方法、装置、可读存储介质及终端设备
WO2021088849A1 (zh) 一种超声成像方法、装置、可读存储介质及终端设备
KR20210075140A (ko) 이미지 처리 방법 및 장치, 프로세서, 전자 기기, 저장 매체
CN113887447B (zh) 一种针对密集群体目标的密度估计、分类预测模型的训练、推理方法及装置
Kluger et al. Cuboids revisited: Learning robust 3d shape fitting to single rgb images
WO2022099528A1 (zh) 点云法向量计算方法、装置、计算机设备和存储介质
CN113129352B (zh) 一种稀疏光场重建方法及装置
CN111192312B (zh) 基于深度学习的深度图像获取方法、装置、设备及介质
WO2022143366A1 (zh) 图像处理方法、装置、电子设备、介质及计算机程序产品
Xiao et al. Building segmentation and modeling from airborne LiDAR data
US20170185909A1 (en) Systems and methods for performing real-time convolution calculations of matrices indicating amounts of exposure
Li et al. Fitting boxes to Manhattan scenes using linear integer programming
WO2022160897A1 (zh) 双目视差估计方法、模型训练方法以及相关设备
Xie et al. Event-based stereo matching using semiglobal matching
CN115330940A (zh) 一种三维重建方法、装置、设备和介质
CN114565916A (zh) 目标检测模型训练方法、目标检测方法以及电子设备
WO2024183130A1 (zh) 旷场偏好区域确定方法、装置、设备及存储介质
Micheas et al. Random set modelling of three-dimensional objects in a hierarchical bayesian context
US20230104674A1 (en) Machine learning techniques for ground classification
CN114612544A (zh) 图像处理方法、装置、设备和存储介质
CN114529801A (zh) 一种目标检测的方法、装置、设备及存储介质
Chen et al. A 3-D point clouds scanning and registration methodology for automatic object digitization
Li et al. An FPGA-based tree crown detection approach for remote sensing images