Information processing device, parameter adjustment method and program

Abstract

To provide an information processing device capable of appropriately setting detection parameters.SOLUTION: An image data acquisition unit 2 acquires a plurality of image data. For example, the image data acquisition unit 2 acquires image data obtained by an imaging device such as a camera from the imaging device. A detection unit 4 detects, in an adjustment step, a person from one or more pieces of test image data using each combination of a plurality of detection parameter values. A determination unit 6 determines a combination of detection parameter values to be applied to person detection processing in an operation step, based on detection results in the adjustment step.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, a parameter adjustment method, and a program. [Background technology]

[0002] There are known techniques for detecting people from images captured using an imaging device such as a surveillance camera. In relation to such techniques, Patent Document 1 discloses a person recognition device that allows easy setting of a threshold for recognizing a person as a pre-registered person based on similarity. When a setting mode is selected, the person recognition device according to Patent Document 1 sets a threshold used when determining whether a captured person is a pre-registered person based on the similarity between facial feature information extracted by capturing an image of the registered person and stored facial feature information.

[0003] Furthermore, Patent Document 2 discloses a face matching device that performs efficient and stable face matching under various conditions. The face matching device disclosed in Patent Document 2 detects a face from an input image. The face matching device disclosed in Patent Document 2 also adjusts a score that indicates the degree of similarity between features extracted from the detected face and features of a registrant using a score adjustment parameter. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2014-063251 [Patent Document 2] Japanese Patent Application Laid-Open No. 2009-163555 Summary of the Invention [Problem to be solved by the invention]

[0005] The above-mentioned patent documents disclose adjusting parameters used when performing facial recognition of a specific person. However, the above-mentioned patent documents do not disclose adjusting detection parameters used in person detection, which is a step before facial recognition, that detects whether a person is captured in an image. Here, the detection parameters used when detecting a person may be adjusted by an operator at the site where the imaging device is installed. However, since the imaging conditions at the installation location of the imaging device vary depending on the installation location, it is difficult to appropriately set the detection parameters at the site.

[0006] The object of the present disclosure has been made to solve such problems, and is to provide an information processing device, a parameter adjustment method, and a program that enable appropriate setting of detection parameters. [Means for solving the problem]

[0007] The information processing device according to the present disclosure includes an image data acquisition means for acquiring multiple image data, a detection means for detecting a person from one or more test image data including an image of a person selected from the multiple image data using each of multiple combinations of multiple values ​​of multiple detection parameters used to detect a person during an adjustment stage, and a determination means for determining a combination of values ​​of the detection parameters to be applied in the person detection process during an operation stage based on the detection results during the adjustment stage.

[0008] In addition, the parameter adjustment method according to the present disclosure acquires a plurality of image data, and in an adjustment stage, detects a person from one or more test image data selected from the plurality of image data and including an image of a person using each of a plurality of combinations of values ​​of a plurality of detection parameters used to detect a person, and determines a combination of values ​​of the detection parameters to be applied in the person detection process in the operation stage based on the detection results in the adjustment stage.

[0009] In addition, the program according to the present disclosure causes a computer to execute the steps of acquiring multiple image data, detecting a person from one or more test image data selected from the multiple image data and including an image of a person using each of multiple combinations of multiple values ​​of multiple detection parameters used to detect a person in an adjustment phase, and determining, based on the detection results in the adjustment phase, a combination of values ​​of the detection parameters to be applied in the person detection process in the operation phase. [Effects of the Invention]

[0010] According to the present disclosure, it is possible to provide an information processing device, a parameter adjustment method, and a program that enable appropriate setting of detection parameters. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a diagram illustrating an overview of an information processing device according to an embodiment of the present invention. [Figure 2] 1 is a flowchart showing an outline of a parameter adjustment method executed by the information processing device according to the present embodiment. [Figure 3] FIG. 1 illustrates a human detection system according to a first embodiment. [Figure 4] FIG. 1 is a diagram illustrating a configuration of an information processing device according to a first embodiment. [Figure 5] 4 is a flowchart illustrating a parameter adjustment method executed by the information processing device according to the first embodiment. [Figure 6] FIG. 4 is a diagram for explaining a first detection process according to the first embodiment. [Figure 7] FIG. 10 is a diagram for explaining a second detection process according to the first embodiment. [Figure 8] FIG. 10 is a diagram illustrating a configuration of a face authentication system according to a second embodiment. [Figure 9] FIG. 10 is a diagram illustrating a configuration of a face authentication device according to a second embodiment. [Figure 10]FIG. 10 is a diagram illustrating an example of an operation screen displayed on a setting terminal according to the second embodiment. [Figure 11] FIG. 10 is a diagram for explaining switching of detection parameters according to the second embodiment. [Figure 12] FIG. 10 is a diagram for explaining switching of detection parameters according to the second embodiment. [Figure 13] FIG. 10 is a diagram for explaining switching of detection parameters according to the second embodiment. [Figure 14] FIG. 10 is a diagram illustrating a human detection system according to a third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0012] (Outline of this embodiment) Before describing the present embodiment, an outline of the present embodiment will be described. FIG. 1 is a diagram showing an outline of an information processing device 1 according to the present embodiment. The information processing device 1 according to the present embodiment can be realized by, for example, a computer. The information processing device 1 may be realized by a plurality of computers. Furthermore, the information processing device 1 may be realized by cloud computing.

[0013] The information processing device 1 has an image data acquisition unit 2, a detection unit 4, and a determination unit 6. The image data acquisition unit 2 functions as an image data acquisition means. The detection unit 4 functions as a detection means. The determination unit 6 functions as a determination means.

[0014] 2 is a flowchart showing an outline of a parameter adjustment method executed by the information processing device 1 according to this embodiment. The image data acquisition unit 2 acquires a plurality of image data (step S12). For example, the image data acquisition unit 2 acquires image data obtained by an imaging device such as a camera from the imaging device.

[0015] In the adjustment stage, the detection unit 4 detects people from one or more test image data using each combination of values ​​of a plurality of detection parameters (step S14). Here, "detection parameters" are parameters used to detect people. By appropriately setting the values ​​of the detection parameters, the accuracy of person detection can be improved. Specific examples of detection parameters will be described later.

[0016] Furthermore, the "adjustment stage" is a stage in which detection parameters are adjusted. The adjustment stage can also be referred to as a search stage in which values ​​of detection parameters to be used in the operation stage described below are searched for. Furthermore, "person detection" is a process of determining whether a person is present in an image represented by image data and at what position in this image the person is present. Person detection is, for example, face detection processing, but is not limited to this configuration. Person detection may be performed, for example, using a trained model that has been trained to input image data and output a person detection result. The trained model may be trained, for example, by machine learning such as a neural network.

[0017] Furthermore, "test image data" is used to adjust the values ​​of the detection parameters in the adjustment stage. The test image data is image data including an image of a person. The test image data is selected from a plurality of image data acquired by the image data acquisition unit 2. That is, the detection unit 4 detects a person from one or more test image data including an image of a person selected from the plurality of image data using each of a plurality of combinations of a plurality of values ​​of each of a plurality of detection parameters. Note that the test image data may be selected by a user, for example, but is not limited to such a configuration.

[0018] The determination unit 6 determines a combination of detection parameter values ​​to be applied in the person detection process in the operation stage based on the detection results in the adjustment stage (step S16). Here, the "operation stage" refers to a stage in which a person is actually detected from image data obtained using an imaging device using the detection parameter values ​​adjusted in the adjustment stage, and processing such as authentication processing or matching processing is performed on the detected person. The authentication processing and matching processing are face authentication processing and face matching processing, respectively, but are not limited to such configurations. In the operation stage, a specific person can be identified from people detected in image data.

[0019] There are various detection parameters used when performing person detection processing. Each detection parameter can take on various values. Furthermore, some detection parameters can affect the accuracy of person detection depending on the shooting conditions. The detection parameters may be adjusted by an operator at the site where the imaging device is installed. However, since the adjustment of the detection parameters is often performed based on the operator's experience and know-how, it is difficult to appropriately adjust the detection parameters.

[0020] In contrast, as described above, the information processing device 1 according to this embodiment is configured to determine a combination of detection parameter values ​​based on the detection results in the adjustment stage. Therefore, the detection parameter values ​​can be determined without relying on the experience and know-how of an operator. Therefore, the information processing device 1 according to this embodiment can appropriately set the detection parameters. Furthermore, the parameter adjustment method executed by the information processing device 1 according to this embodiment can also appropriately set the detection parameters. Furthermore, the program realizing the parameter adjustment method can also appropriately set the detection parameters.

[0021] (Embodiment 1) Hereinafter, embodiments will be described with reference to the drawings. For clarity of explanation, the following description and drawings have been omitted and simplified as appropriate. In addition, the same elements in each drawing are designated by the same reference numerals, and duplicate explanations are omitted as necessary.

[0022] FIG. 3 is a diagram illustrating a human detection system 20 according to the first embodiment. The human detection system 20 includes an imaging device 30 and an information processing device 100. The imaging device 30 and the information processing device 100 are communicably connected to each other via a network 22. The network 22 may be a wired network, a wireless network, or a combination of a wired network and a wireless network. The network 22 may be the Internet or a local area network (LAN). The wireless network 22 may be, for example, a network using a communication line standard such as LTE (Long Term Evolution), or may be a network used in a specific area such as WiFi (registered trademark) or local 5G.

[0023] The imaging device 30 captures images of the surroundings of the location where the imaging device 30 is installed. The imaging device 30 generates image data representing the captured image of the surroundings. The imaging device 30 transmits the generated image data to the information processing device 100. The imaging device 30 is, for example, a camera. The imaging device 30 may also be a surveillance camera. The imaging device 30 may also be, for example, an IP (Internet Protocol) camera or a network camera. In this case, the imaging device 30 may have computer functions. In this case, the imaging device 30 may have the hardware configuration of the information processing device 100, as described below. In this embodiment, the term "image" may also mean "image data representing an image" as a processing target in information processing. The "image" may be a still image or a moving image (video).

[0024] The information processing device 100 corresponds to the information processing device 1 shown in FIG. 1. The information processing device 100 can be realized by, for example, a computer. The information processing device 100 may be realized by multiple computers or by cloud computing. In this case, each component of the information processing device 100 described below may be realized by cloud computing or multiple computers. In other words, each component may be realized by physically different computers.

[0025] 4 is a diagram showing the configuration of the information processing device 100 according to the first embodiment. The information processing device 100 has, as its main hardware components, a control unit 102, a storage unit 104, a communication unit 106, and an interface unit 108. The control unit 102, the storage unit 104, the communication unit 106, and the interface unit 108 are connected to each other via a data bus or the like.

[0026] The control unit 102 is a processor such as a CPU (Central Processing Unit). The control unit 102 functions as a calculation device that performs analysis processing, control processing, calculation processing, etc. The control unit 102 may have multiple processors. The storage unit 104 is a storage device such as a memory or a hard disk. The storage unit 104 is, for example, a ROM (Read Only Memory) or a RAM (Random Access Memory). The storage unit 104 has a function to store control programs, calculation programs, etc. executed by the control unit 102. In other words, the storage unit 104 (memory) stores one or more instructions. The storage unit 104 also has a function to temporarily store processing data, etc. The storage unit 104 may include a database. The storage unit 104 may have multiple memories.

[0027] The communication unit 106 performs processing necessary for communicating with other devices, such as the imaging device 30, via the network 22. The communication unit 106 may include a communication port, a router, a firewall, etc. The interface unit 108 (IF; Interface) is, for example, a user interface (UI). The interface unit 108 has an input device, such as a keyboard, a touch panel, or a mouse, and an output device, such as a display or a speaker. The interface unit 108 may be configured such that the input device and the output device are integrated, such as a touch screen (touch panel). The interface unit 108 accepts data input operations by a user, such as an operator or worker, and outputs information to the user. The interface unit 108 may output information regarding detection parameters.

[0028] The information processing device 100 according to the first embodiment has, as its components, an image collection mode processing unit 110, an adjustment mode processing unit 120, and an operation mode processing unit 140. The image collection mode processing unit 110 has an image data acquisition unit 112, a test image selection processing unit 114, and a test image storage unit 116. The adjustment mode processing unit 120 has a detection parameter storage unit 122, a human detection unit 124, a detection parameter value determination unit 126, a detection parameter value storage unit 128, and an inapplicable detection parameter value storage unit 130. The operation mode processing unit 140 has a detection parameter value selection processing unit 142 and a human detection process execution unit 144.

[0029] As described above, the information processing device 100 does not have to be physically configured as a single device. In this case, each of the above-described components may be realized by a plurality of physically separate devices.

[0030] The image collection mode processing unit 110 functions as an image collection mode processing means. The image data acquisition unit 112 functions as an image data acquisition means. The test image selection processing unit 114 functions as a test image selection processing means. The test image storage unit 116 functions as a test image storage means.

[0031] The adjustment mode processing unit 120 functions as an adjustment mode processing means. The detection parameter storage unit 122 functions as a detection parameter storage means. The person detection unit 124 functions as a person detection means (detection means). The detection parameter value determination unit 126 functions as a detection parameter value determination means (determination means). The detection parameter value storage unit 128 functions as a detection parameter value storage means. The non-applicable detection parameter value storage unit 130 functions as a non-applicable detection parameter value storage means.

[0032] The operation mode processing unit 140 functions as an operation mode processing unit. The detection parameter value selection processing unit 142 functions as a detection parameter value selection processing unit. The person detection processing execution unit 144 functions as a person detection processing execution unit.

[0033] Each of the above-described components can be realized, for example, by executing a program under the control of the control unit 102. More specifically, each component can be realized by the control unit 102 executing a program stored in the storage unit 104. Alternatively, each component may be realized by recording the necessary program on an arbitrary non-volatile recording medium and installing it as needed. Each component may not necessarily be realized by software programs, but may be realized by any combination of hardware, firmware, and software. Each component may also be realized using a user-programmable integrated circuit, such as an FPGA (field-programmable gate array) or a microcomputer. In this case, a program consisting of each of the above-described components may be realized using this integrated circuit. The same applies to other embodiments described below.

[0034] The image collection mode processing unit 110 performs processing so that the information processing device 100 operates in the image collection mode. Here, the "image collection mode" refers to an operating mode in the image collection stage. The "image collection stage" is a stage before the adjustment stage. In the image collection mode, test image data to be used in the adjustment stage is collected. Note that in the image collection mode, detection parameter values ​​that allow the detection of not only people but also objects other than people can be applied in the detection process. In the image collection mode, continuous image data for several seconds before and after the timing at which the detection process is performed can be recorded.

[0035] The image collection mode processing unit 110 may set the time periods for collecting image data and the upper limit number of detections for each time period by the user operating the interface unit 108. The user interface for setting these will be described later. The "time periods" are set appropriately so that the image capture device 30 can capture sufficient images of pedestrians passing through an area corresponding to the angle of view of the image capture device 30.

[0036] The "detection limit number" corresponds to the number of times the detection process is performed. The detection limit number can be set for each time period. This allows test image data to be collected taking into account the operating conditions of the system user. In other words, the detection limit number is set taking into account the environment in which the imaging device 30 is installed. For example, if the imaging device 30 is installed in an environment where most pedestrians are concentrated in the early morning hours of the day, the accuracy of the detection parameters can be improved by increasing the number of samples taken in the early morning hours. In addition, the memory space for saving test image data is limited, and detection processes cannot be performed unlimitedly. Therefore, setting an upper limit by setting the detection limit number can save data capacity in the memory space.

[0037] The image data acquisition unit 112 corresponds to the image data acquisition unit 2 shown in FIG. 1. The image data acquisition unit 112 acquires a plurality of image data. Specifically, the image data acquisition unit 112 acquires a plurality of image data generated by the image capture device 30 from the image capture device 30 via the communication unit 106. Note that the image represented by the image data acquired by the image data acquisition unit 112 may include an image of a person, but does not necessarily include an image of a person. In other words, as described above, at the image collection stage, the image capture device 30 does not need to be adjusted to perform person detection with high accuracy. Therefore, at the image collection stage, the detection parameters used for person detection do not need to be set to perform person detection with high accuracy.

[0038] The test image selection processing unit 114 performs processing for selecting test image data. Specifically, the test image selection processing unit 114 may perform processing so that the user can select one or more test image data from the plurality of image data acquired by the image data acquisition unit 112. More specifically, the test image selection processing unit 114 may cause the interface unit 108 to display the image data acquired by the image data acquisition unit 112. Then, the user may operate the interface unit 108 to select test image data from the plurality of image data displayed.

[0039] For example, the user may visually check the displayed plurality of image data and select test image data. The user may select image data in which a person is correctly detected from the displayed plurality of image data as test image data. Specifically, the user excludes from the plurality of image data images images in which a person cannot be detected (undetected images) and images in which an object other than a person is mistakenly detected as a person (falsely detected images), and selects only image data in which a person is correctly detected as test image data. Therefore, the test image data includes an image of a person.

[0040] The test image storage unit 116 stores the selected test image data. Specifically, the test image storage unit 116 stores the test image data using the memory unit 104. The test image data stored in the test image storage unit 116 is used in the adjustment mode. The test image storage unit 116 may store correct answer data corresponding to each piece of test image data. The user may also select test image data for each specific person. That is, the user may separately select test image data in which person A is detected and test image data in which person B is detected. In this case, the test image storage unit 116 may separately store the test image data in which person A is detected and the test image data in which person B is detected. In this case, the test image storage unit 116 may store feature amount data indicating the feature amount of the image of each person as a registered person list.

[0041] The adjustment mode processing unit 120 performs processing so that the information processing device 100 operates in the adjustment mode. Here, the "adjustment mode" is an operation mode in the adjustment stage. The "adjustment stage" is a stage before the operation stage. As described above, in the adjustment mode, the values ​​of the detection parameters used in the operation stage are adjusted. Specific examples of the detection parameters will be described later.

[0042] The detection parameter storage unit 122 stores a plurality of detection parameters used in the detection process. The detection parameter storage unit 122 may also store candidate values ​​that may be applied in the operation stage for each of the plurality of detection parameters. The detection parameter storage unit 122 stores the detection parameters and the like using the memory unit 104. The detection parameter storage unit 122 may store, for example, information such as that illustrated in FIG. 11, which will be described later.

[0043] The person detection unit 124 corresponds to the detection unit 4 shown in FIG. 1. The person detection unit 124 detects people from the test image data. An existing person detection method may be adopted as the person detection processing method. For example, an existing face detection processing method may be adopted as the person detection processing method. For example, in the face detection processing, the position of a person's face is identified from the image data, and a reliability indicating the likelihood that the image of the face at the identified position is actually a face is calculated.

[0044] Furthermore, the person detection unit 124 detects people from one or more test image data using each of a plurality of combinations of a plurality of values ​​of a plurality of detection parameters. Specifically, the person detection unit 124 performs a person detection process on the test image data for each of a plurality of combinations of a plurality of values ​​of a plurality of detection parameters. For example, the person detection unit 124 performs a person detection process on the test image data by switching between combinations of values ​​of the plurality of detection parameters. Note that the number of detection parameter values ​​and candidate detection parameter values ​​used in the detection process in the person detection unit 124 can be appropriately set by the user operating the interface unit 108.

[0045] Furthermore, the human detection unit 124 may perform a first detection process using a combination of values ​​of one or more first detection parameters among the plurality of detection parameters. Then, depending on the detection result obtained in the first detection process, the human detection unit 124 may perform a second detection process using a combination of values ​​of a second detection parameter other than at least the first detection parameter among the plurality of detection parameters. This will be described in detail later. After the second detection process, the human detection unit 124 may perform a third detection process using a third detection parameter.

[0046] In this way, by the person detection unit 124 performing the second detection process after the first detection process, the number of combinations of detection parameter values ​​in each detection process can be reduced. Therefore, it is possible to perform the detection processes efficiently. For example, assume that there are four detection parameters, each with three values ​​set. In this case, if the detection processes are performed using combinations of all the values ​​of the four detection parameters in a series of detection processes, it is necessary to perform the detection processes for each of the 81 (=3×3×3×3) combinations.

[0047] In contrast, suppose a first detection process is performed using three detection parameters as first detection parameters, and a second detection process is performed using the remaining detection parameter as a second detection parameter. In this case, the first detection process performs detection processing for each of 27 (=3×3×3) combinations. Then, for example, the second detection process performs detection processing by combining some of the 27 combinations set according to the detection results of the first detection process with combinations of three values ​​of the second detection parameters. For example, the second detection process performs detection processing for each of nine (=3×3) combinations, combining three of the 27 combinations in the first detection process with combinations of three values ​​of the second detection parameters. Therefore, the first and second detection processes perform 36 (=27+9) detection processes, which is fewer than 81. Therefore, by performing the second detection process after the first detection process, the number of combinations of detection parameter values ​​to be considered can be reduced. This allows for efficient detection processing.

[0048] The detection result may also include at least one accuracy parameter indicating the accuracy of human detection. That is, the human detection unit 124 may acquire a detection result including an accuracy parameter for each combination of values ​​of a plurality of detection parameters through a detection process using each combination of values ​​of a plurality of detection parameters. Examples of the accuracy parameter include an accuracy rate, a non-detection rate, and a false detection rate. However, the accuracy parameter is not limited to these.

[0049] The "accuracy rate" corresponds to the ratio of the number of data judged to be correct to the number of all data for which the person detection was judged to be correct. The "non-detection rate" corresponds to the ratio of the number of data judged to be non-detected to the number of all data for which the person detection was judged to be correct. "Non-detection" corresponds to a case where a person included in the test image data could not be detected. The "false detection rate" corresponds to the ratio of the number of data judged to be falsely detected to the number of all data for which the person detection was judged to be correct. "False detection" corresponds to a case where an object other than a person included in the test image data was detected as a person.

[0050] Furthermore, the person detection unit 124 may perform the first detection process using a combination of first detection parameter values ​​that is set according to the influence that the multiple detection parameters have on the accuracy of person detection. For example, the person detection unit 124 may perform the first detection process by setting a detection parameter that has a greater influence on the accuracy of person detection as the first detection parameter. Then, the person detection unit 124 may perform the second detection process using a combination of the value of the first detection parameter and the value of the second detection parameter that is selected according to the detection accuracy in the detection result obtained in the first detection process. This will be described in detail later.

[0051] The detection parameter value determination unit 126 corresponds to the determination unit 6 shown in FIG. 1. The detection parameter value determination unit 126 determines a combination of detection parameter values ​​to be applied in the person detection process in the operational phase, based on the detection result executed by the person detection unit 124. Specifically, the detection parameter value determination unit 126 may determine a combination of detection parameter values ​​to be applied in the person detection process in the operational phase, based on the detection result including at least one accuracy parameter as described above. This will be described in detail later. Note that by determining a combination of detection parameter values ​​to be applied in the operational phase based on the detection result including the accuracy parameter, it is possible to determine a combination of detection parameter values ​​that will improve detection accuracy, without relying on the experience and know-how of the operator.

[0052] Furthermore, the detection parameter value determination unit 126 may determine multiple combinations of detection parameter values ​​that can be applied in the operation phase, depending on the detection result including multiple accuracy parameters. This will be described in detail later. That is, the detection parameter value determination unit 126 may determine multiple combinations of detection parameter values ​​that can be applied in the operation phase, depending on each of the multiple accuracy parameters. With this configuration, it is possible to determine an appropriate combination of detection parameter values ​​depending on the operation mode of the system user.

[0053] The detection parameter value determination unit 126 may then determine, from among the determined combinations, a combination of detection parameter values ​​corresponding to the accuracy parameter set according to the operation mode as the combination of detection parameter values ​​to be applied in the operation stage. This will be described in detail later. For example, if it is desirable for the "accuracy rate" to be the highest in a certain operation stage, an accuracy parameter for the "accuracy rate" may be set for that operation stage. In this case, the detection parameter value determination unit 126 may determine, as the combination of detection parameter values ​​to be applied in that operation stage, a combination of detection parameter values ​​that maximizes the accuracy parameter for the "accuracy rate." This configuration makes it possible to appropriately determine, according to the operation mode of the system user, a combination of detection parameter values ​​that maximizes the accuracy parameter that is important in the detection process.

[0054] The detection parameter value storage unit 128 stores the combination of detection parameter values ​​determined by the detection parameter value determination unit 126. That is, the detection parameter value storage unit 128 stores the combination of detection parameter values ​​to be applied in the operation stage. At this time, the detection parameter value storage unit 128 may store the combination of detection parameter values ​​for each accuracy parameter set according to the operation mode.

[0055] For example, the detection parameter value storage unit 128 may store a combination of detection parameter values ​​that results in the best accuracy rate. Also, for example, the detection parameter value storage unit 128 may store a combination of detection parameter values ​​that results in the best false positive rate. Also, for example, the detection parameter value storage unit 128 may store a combination of detection parameter values ​​that results in the best undetected rate.

[0056] Furthermore, for example, the detection parameter value storage unit 128 may store a combination of detection parameter values ​​that provides the best balance between the false detection rate and the undetected rate. The "combination that provides the best balance between the false detection rate and the undetected rate" may be, for example, a combination that provides the best accuracy rate among combinations that provide the smallest difference between the false detection rate and the undetected rate. Depending on the operating mode of a user (system user), the balance between the false detection rate and the undetected rate may be emphasized. Therefore, the detection parameter value storage unit 128 may also store a combination that provides the best balance between the false detection rate and the undetected rate so that the user can select it during operation. The detection parameter value storage unit 128 uses the memory unit 104 to store combinations of detection parameter values.

[0057] The non-applicable detection parameter value storage unit 130 stores one or more combinations of detection parameter values ​​other than the combinations of detection parameter values ​​applied in the person detection process in the operation phase. That is, the non-applicable detection parameter value storage unit 130 stores combinations of detection parameter values ​​that are not applied in the operation phase. In other words, the non-applicable detection parameter value storage unit 130 stores combinations of detection parameter values ​​that are not determined by the detection parameter value determination unit 126 to be applied in the operation phase. The non-applicable detection parameter value storage unit 130 stores the combinations of detection parameter values ​​using the memory unit 104. This makes it possible to change the combinations of detection parameter values ​​in the operation phase. That is, if the accuracy of person detection is not satisfactory when actual person detection is performed in the operation phase, the detection parameter values ​​can be changed to a combination of detection parameter values ​​that is not applied in the current operation. Therefore, it is possible to flexibly change the detection parameter values ​​in the operation phase.

[0058] The operation mode processing unit 140 performs processing so that the information processing device 100 operates in the operation mode. Here, the "operation mode" refers to an operating mode in the operation stage. The "operation stage" is a stage after the adjustment stage. In the operation mode, the original operation of the system, such as person detection processing, is performed using the detection parameter values ​​determined in the adjustment stage. Furthermore, in the operation mode, person authentication processing (face authentication processing) or person matching processing (face matching processing) may be performed on a person detected in the person detection processing.

[0059] The detection parameter value selection processing unit 142 selects detection parameter values ​​that will actually be used in the operation phase. Specifically, the detection parameter value selection processing unit 142 may perform processing to allow the user to select a combination of detection parameter values ​​from among multiple combinations of detection parameter values ​​determined in the adjustment phase. More specifically, the detection parameter value selection processing unit 142 may display multiple combinations of detection parameter values ​​determined in the adjustment phase. The user may then operate the interface unit 108 to select a combination of detection parameter values ​​to be used in the operation phase from the multiple displayed combinations. This sets the combination of detection parameter values ​​to be used for human detection in the operation phase. Note that the detection parameter value selection processing unit 142 may not be necessary. In this case, after the adjustment phase is completed, the operation phase may automatically start, in which the combination of detection parameter values ​​determined by the detection parameter value determination unit 126 is applied.

[0060] The person detection process execution unit 144 controls the execution of person detection process in the operation stage. Specifically, the person detection process execution unit 144 controls the image capture device 30 to perform person detection process using a set combination of detection parameter values. Furthermore, the person detection process execution unit 144 performs person detection on image data obtained by the image capture device 30 capturing an image of a person, using the set combination of detection parameter values. Then, the person detection process execution unit 144 may perform processing to add a bounding box (rectangle) to a position where a person detected in the image data exists. The person detection process execution unit 144 may calculate a reliability indicating the likelihood of a person detected in the image data. Furthermore, the person detection process execution unit 144 may perform person authentication process and person matching process.

[0061] 5 is a flowchart showing a parameter adjustment method executed by the information processing device 100 according to the first embodiment. The image collection mode processing unit 110 acquires image data (step S102). Specifically, as described above, the image collection mode processing unit 110 sets the time periods for acquiring image data and the upper limit number of detections for each time period. Then, the image data acquisition unit 112 acquires image data under the set conditions.

[0062] The image collection mode processing unit 110 performs a test image selection process (step S104). Specifically, as described above, the test image selection processing unit 114 performs a process so that a user (an operator, etc.) can select one or more test image data from the plurality of image data acquired by the image data acquisition unit 112.

[0063] The adjustment mode processing unit 120 performs a person detection process (step S110). Specifically, as described above, the person detection unit 124 performs the person detection process on the test image data selected in the process of S104, using each of a plurality of combinations of a plurality of values ​​of a plurality of detection parameters. Also, as described above, the person detection unit 124 switches the combination of values ​​of the plurality of detection parameters to perform the person detection process on the test image data. That is, the person detection unit 124 performs a tuning test on the selected test image data by automatically switching the combination of values ​​of the detection parameters used in person detection.

[0064] More specifically, the person detection unit 124 performs a first detection process using a combination of values ​​of first detection parameters among the plurality of detection parameters (step S112). More specifically, as described above, the person detection unit 124 detects a person using the first detection parameters that have a large effect on the accuracy of person detection, and acquires a first detection result.

[0065] Furthermore, the person detection unit 124 performs a second detection process using a combination of values ​​of at least a second detection parameter among the plurality of detection parameters, according to the first detection result obtained in the first detection process (step S114). More specifically, as described above, the person detection unit 124 detects a person using the value of the first detection parameter that provided a good accuracy in the first detection result and the second detection parameter that has a smaller effect on detection accuracy than the first detection parameter. In this way, the person detection unit 124 obtains a second detection result.

[0066] FIG. 6 is a diagram for explaining a first detection process according to the first embodiment. FIG. 7 is a diagram for explaining a second detection process according to the first embodiment. In the examples of FIGS. 6 and 7, detection parameters A, B, and C are set as first detection parameters. In the examples of FIGS. 6 and 7, detection parameter D is set as a second detection parameter. In the examples of FIGS. 6 and 7, accuracy parameters X, Y, and Z are used as the above-mentioned accuracy parameters. The accuracy parameter X may correspond to, for example, a "correct answer rate." The accuracy parameter Y may correspond to, for example, a "non-detection rate." The accuracy parameter Z may correspond to, for example, a "false detection rate."

[0067] In the example of FIG. 6, the candidates for the setting value of detection parameter A are "Va1," "Va2," and "Va3." The candidates for the setting value of detection parameter B are "Vb1," "Vb2," and "Vb3." The candidates for the setting value of detection parameter C are "Vc1," "Vc2," and "Vc3." In the example of FIG. 7, the candidates for the setting value of detection parameter D are "Vd1," "Vd2," and "Vd3." Note that, although the number of candidates for the setting value of the detection parameter is three in the examples of FIGS. 6 and 7, the number of candidates for the setting value of the detection parameter is arbitrary.

[0068] As illustrated in FIG. 6, the person detection unit 124 performs person detection on the test image data for each combination of the values ​​of detection parameter A, detection parameter B, and detection parameter C. In the example of FIG. 6, for example, the person detection unit 124 performs person detection using combination #1 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc1." Furthermore, for example, the person detection unit 124 performs person detection using combination #2 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2." Furthermore, for example, the person detection unit 124 performs person detection using combination #3 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc3." Furthermore, for example, the person detection unit 124 performs person detection using combination #4 "detection parameter A: Va1, detection parameter B: Vb2, detection parameter C: Vc1." Furthermore, for example, the person detection unit 124 performs person detection using combination #10, "detection parameter A: Va2, detection parameter B: Vb1, detection parameter C: Vc1." Similarly, the person detection unit 124 performs person detection using each of the 27 combinations (combinations #1 to #27). Then, as a result of performing person detection using each combination, the person detection unit 124 obtains values ​​of accuracy parameter X, accuracy parameter Y, and accuracy parameter Z for each combination.

[0069] Then, the person detection unit 124 searches for the combination with the best value of the accuracy parameter X among the detection results of combinations #1 to #27. In the example of Fig. 6, when person detection is performed using combination #2 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2", the value of the accuracy parameter X in the detection result is the best "Vx11".

[0070] Furthermore, the person detection unit 124 searches for the combination with the best value of the accuracy parameter Z among the detection results of combinations #1 to #27. In the example of Fig. 6, when person detection is performed using combination #9 "detection parameter A: Va1, detection parameter B: Vb3, detection parameter C: Vc3", the value of the accuracy parameter Z in the detection result is the best "Vz11".

[0071] Furthermore, the person detection unit 124 searches for a combination that provides the best balance between the value of the accuracy parameter Y and the value of the accuracy parameter Z from the detection results of combinations #1 to #27. In the example of FIG. 6, when person detection is performed using combination #20 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2", the value of the accuracy parameter Y is "Vy12". At this time, the value of the accuracy parameter Z is "Vz12". When person detection is performed using combination #21 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc3", the value of the accuracy parameter Y is "Vy13". At this time, the value of the accuracy parameter Z is "Vz13". It is assumed that the difference between "Vy12" and "Vz12" and the difference between "Vy13" and "Vz13" are both minimum. For example, suppose that the difference between "Vy12" and "Vz12" and the difference between "Vy13" and "Vz13" are both 0. Furthermore, suppose that the value of accuracy parameter X, "Vx12," obtained when person detection is performed using combination #20 is better than the value of accuracy parameter X, "Vx13," obtained when person detection is performed using combination #21. In this case, when person detection is performed using combination #20, the detection result will have the best balance between the value of accuracy parameter Y and the value of accuracy parameter Z.

[0072] The person detection unit 124 performs a second detection process according to the detection result of the first detection process. The person detection unit 124 uses, in the second detection process, the combination with the best value of the accuracy parameter X, the combination with the best value of the accuracy parameter Z, and the combination with the best balance between the values ​​of the accuracy parameters Y and Z. In the example of Fig. 6, combination #2, combination #9, and combination #20 are used in the second detection process.

[0073] The person detection unit 124 performs person detection using a combination of "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2" and candidate values ​​of detection parameter D "Vd1," "Vd2," and "Vd3." In the example of FIG. 7, these three combinations correspond to combinations #1 to #3. The person detection unit 124 performs person detection using combination #1 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd1." Furthermore, the person detection unit 124 performs person detection using combination #2 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd2." Furthermore, the person detection unit 124 performs person detection using combination #3 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd3."

[0074] Furthermore, the person detection unit 124 performs person detection using a combination of "detection parameter A: Va1, detection parameter B: Vb3, detection parameter C: Vc3" and candidate values ​​of detection parameter D: "Vd1," "Vd2," and "Vd3." In the example of FIG. 7, these three combinations correspond to combinations #4 to #6. The person detection unit 124 performs person detection using combination #4 "detection parameter A: Va1, detection parameter B: Vb3, detection parameter C: Vc3, detection parameter D: Vd1." The person detection unit 124 also performs person detection using combination #5 "detection parameter A: Va1, detection parameter B: Vb3, detection parameter C: Vc3, detection parameter D: Vd2." The person detection unit 124 also performs person detection using combination #6 "detection parameter A: Va1, detection parameter B: Vb3, detection parameter C: Vc3, detection parameter D: Vd3."

[0075] Furthermore, the person detection unit 124 performs person detection using a combination of "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2" and candidate values ​​of detection parameter D "Vd1," "Vd2," and "Vd3." In the example of FIG. 7, these three combinations correspond to combinations #7 to #9. The person detection unit 124 performs person detection using combination #7 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd1." The person detection unit 124 also performs person detection using combination #8 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd2." The person detection unit 124 also performs person detection using combination #9 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd3."

[0076] Then, the human detection unit 124 searches for the combination with the best value of the accuracy parameter X from among the detection results of combinations #1 to #9. In the example of Fig. 7, when human detection is performed using combination #2 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd2", the value of the accuracy parameter X in the detection result is the best "Vx21".

[0077] Furthermore, the person detection unit 124 searches for the combination with the best value of the accuracy parameter Z from the detection results of combinations #1 to #9. In the example of Fig. 7, when person detection is performed using combination #4 "detection parameter A: Va1, detection parameter B: Vb3, detection parameter C: Vc3, detection parameter D: Vd1", the value of the accuracy parameter Z in the detection result is the best "Vz21".

[0078] Furthermore, the person detection unit 124 searches for a combination that provides the best balance between the value of the accuracy parameter Y and the value of the accuracy parameter Z from the detection results of combinations #1 to #9. In the example of FIG. 7, when person detection is performed using combination #8 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd2," the value of the accuracy parameter Y is "Vy22." At this time, the value of the accuracy parameter Z is "Vz22." When person detection is performed using combination #9 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd3," the value of the accuracy parameter Y is "Vy23." At this time, the value of the accuracy parameter Z is "Vz23." It is assumed that the difference between "Vy22" and "Vz22" and the difference between "Vy23" and "Vz23" are both minimum. For example, suppose that the difference between "Vy22" and "Vz22" and the difference between "Vy23" and "Vz23" are both 0. Furthermore, suppose that the value of accuracy parameter X, "Vx22," obtained when person detection is performed using combination #8 is better than the value of accuracy parameter X, "Vx23," obtained when person detection is performed using combination #9. In this case, when person detection is performed using combination #8, the detection result will have the best balance between the value of accuracy parameter Y and the value of accuracy parameter Z.

[0079] In this way, the person detection unit 124 performs the first detection process using a combination of first detection parameter values ​​set according to the influence they have on the accuracy of person detection. Then, the person detection unit 124 performs the second detection process using a combination of the first detection parameter value and the second detection parameter value selected according to the detection accuracy in the detection result obtained in the first detection process. With this configuration, it is possible to efficiently determine a combination of detection parameter values ​​that improves the accuracy of person detection.

[0080] As described above, the person detection unit 124 may perform a third detection process using a third detection parameter after the second detection process. The third detection parameter may have a smaller effect on detection accuracy than the second detection parameter. In the third detection process, the value of the detection parameter may be fine-tuned.

[0081] Returning to the description of Fig. 5, the adjustment mode processing unit 120 determines detection parameter values ​​to be applied in the operation phase (step S120). Specifically, the detection parameter value determination unit 126 determines a combination of detection parameter values ​​to be applied in the person detection process in the operation phase, based on the detection result executed by the person detection unit 124. More specifically, the detection parameter value determination unit 126 determines a combination of detection parameter values ​​that results in a good accuracy parameter value in the detection result as the combination of detection parameter values ​​to be applied in the operation phase.

[0082] For example, the detection parameter value determination unit 126 may determine a combination that provides the best value for the accuracy parameter X as the combination of detection parameter values ​​to be applied in the operation phase. In the example of Fig. 7, the detection parameter value determination unit 126 may determine that combination #2 "detection parameter A: Va1, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd2" is to be applied in the operation phase.

[0083] Furthermore, for example, the detection parameter value determination unit 126 may determine a combination that provides the best value for accuracy parameter Z as the combination of detection parameter values ​​to be applied in the operation phase. In the example of Fig. 7, the detection parameter value determination unit 126 may determine that combination #4 "detection parameter A: Va1, detection parameter B: Vb3, detection parameter C: Vc3, detection parameter D: Vd1" is to be applied in the operation phase. Note that the detection parameter value determination unit 126 may determine a combination that provides the best value for accuracy parameter Y as the combination of detection parameter values ​​to be applied in the operation phase.

[0084] Furthermore, for example, the detection parameter value determination unit 126 may determine a combination that provides the best balance between the value of accuracy parameter Y and the value of accuracy parameter Z as the combination of detection parameter values ​​to be applied in the operation phase. In the example of Fig. 7, the detection parameter value determination unit 126 may determine that combination #8 "detection parameter A: Va3, detection parameter B: Vb1, detection parameter C: Vc2, detection parameter D: Vd2" is to be applied in the operation phase.

[0085] That is, the detection parameter value determination unit 126 determines detection parameter values ​​that can be applied in the operation phase, based on the detection results of tuning by the person detection unit 124. At this time, the detection parameter value determination unit 126 may determine a combination of detection parameter values ​​that results in the highest accuracy rate as the combination of detection parameter values ​​to be applied in the operation phase. Furthermore, the detection parameter value determination unit 126 may determine a combination of detection parameter values ​​that results in the lowest false detection rate as the combination of detection parameter values ​​to be applied in the operation phase. Furthermore, the detection parameter value determination unit 126 may determine a combination of detection parameter values ​​that results in the lowest undetected rate as the combination of detection parameter values ​​to be applied in the operation phase. Furthermore, the detection parameter value determination unit 126 may determine a combination of detection parameter values ​​that best balances the false detection rate and the undetected rate as the combination of detection parameter values ​​to be applied in the operation phase. Furthermore, the detection parameter value determination unit 126 may determine a combination of detection parameter values ​​that results in a high accuracy rate in a specific time period as the combination of detection parameter values ​​to be applied in the operation phase. In this case, the detection parameter value determination unit 126 may determine that a combination of detection parameter values ​​that will result in a high accuracy rate when performing person detection on test image data captured during a specific time period will be applied during the operational phase.

[0086] The operation mode processing unit 140 executes the operation mode (step S130). Specifically, the detection parameter value selection processing unit 142 sets the detection parameter values ​​that will actually be used in the operation stage from the combination of detection parameter values ​​determined in the process of S120. Furthermore, the person detection process execution unit 144 performs control so that the person detection process is executed using the set detection parameter values.

[0087] (Embodiment 2) Next, a second embodiment will be described with reference to the drawings. For clarity of explanation, the following description and drawings have been omitted and simplified as appropriate. Furthermore, in each drawing, the same elements are given the same reference numerals, and repeated explanations are omitted as necessary. The second embodiment differs from the first embodiment in that the functions of the information processing device 100 according to the first embodiment can be realized in a face recognition device. Furthermore, the second embodiment will describe an example in which the person detection processing in the first embodiment is performed by face detection processing that detects a person's face. Therefore, in the second embodiment, the person authentication processing is performed by face authentication processing that authenticates the person's face.

[0088] 8 is a diagram showing the configuration of a face authentication system 200 according to the second embodiment. The face authentication system 200 includes a face authentication device 201, a setting terminal 205, a camera 206, and a center 207. The face authentication device 201 is communicably connected to the setting terminal 205 and the camera 206 via a network such as a wired or wireless network. The face authentication device 201 is communicably connected to the center 207 via a network such as the Internet.

[0089] The camera 206 corresponds to the imaging device 30 shown in Fig. 3. The camera 206 captures an image of the environment in which the camera 206 is installed. As a result, the camera 206 generates image data obtained by capturing the image. The camera 206 transmits the image data to the face authentication device 201. Note that the image data may include an image of a person.

[0090] The face authentication device 201 has the hardware configuration of the information processing device 100 shown in FIG. 4. The face authentication device 201 performs face authentication processing. The face authentication device 201 has at least a parameter adjustment unit 230 and a face authentication processing unit 240. The face authentication processing unit 240 detects a person from image data obtained by the camera 206 and authenticates the detected person. The parameter adjustment unit 230 adjusts the values ​​of detection parameters used in the face authentication processing unit 240. Therefore, the face authentication processing unit 240 performs face detection processing using the detection parameters adjusted by the parameter adjustment unit 230. The detailed configuration and functions of the face authentication device 201 will be described later.

[0091] The setting terminal 205 is, for example, a computer such as a PC (Personal Computer). The setting terminal 205 has the hardware configuration of the information processing device 100 shown in Fig. 4. The setting terminal 205 functions as a user interface. When the parameter adjustment unit 230 adjusts the values ​​of the detection parameters, the setting terminal 205 outputs information required by the user (operator) and accepts user operations.

[0092] The center 207 has the hardware configuration of the information processing device 100 shown in Fig. 4. The center 207 uses the results of the face authentication process performed by the face authentication device 201. When authentication (matching) is performed for a specific person, the face authentication device 201 transmits a notification to the center 207. Upon receiving the notification, the center 207 performs the necessary processing for the specific person.

[0093] 9 is a diagram illustrating a configuration of a face authentication device 201 according to a second embodiment. The face authentication device 201 includes a camera image input unit 210, a parameter adjustment unit 230, a face authentication processing unit 240, and a test data storage unit 312. The camera image input unit 210 includes a temporary storage unit 304. The face authentication processing unit 240 includes a face detection unit 305, a detection parameter storage unit 307, a position information storage unit 308, a feature amount storage unit 303, a face matching unit 306, and a score / similarity output unit 309. The parameter adjustment unit 230 includes a parameter adjustment UI 310 and a mode processing unit 320. The mode processing unit 320 includes a collection mode processing unit 322, an adjustment mode processing unit 324, and an operation mode processing unit 326.

[0094] The camera image input unit 210 acquires image data from the camera 206. The camera image input unit 210 outputs the acquired image data as an input image to the face authentication processing unit 240. The camera image input unit 210 also stores input images from a certain period of time in the past (for example, about 5 seconds) in the temporary storage unit 304.

[0095] The face authentication processing unit 240 detects a person using the input image output from the camera image input unit 210 and authenticates the detected person. The detection parameter storage unit 307 stores the values ​​of the detection parameters adjusted by the parameter adjustment unit 230. The detection parameter storage unit 307 corresponds to the detection parameter value storage unit 128 according to the first embodiment.

[0096] The face detection unit 305 performs face detection processing to detect a human face from the input image using adjusted detection parameter values ​​stored in the detection parameter storage unit 307. The face detection processing method performed by the face detection unit 305 may be an existing face detection processing method such as those described above. The face detection unit 305 outputs position information indicating the position of the face in the input image and stores the position information in the position information storage unit 308. The position information may indicate, for example, the position of a bounding box (rectangle) surrounding the face detected in the input image.

[0097] The feature amount storage unit 303 stores in advance feature amounts indicating facial features of a specific person. The face matching unit 306 matches the detected face image with face images of the specific person registered in advance. Specifically, the face matching unit 306 calculates the similarity between the feature amounts of the detected face image and the feature amounts stored in the feature amount storage unit 303. If the similarity of the feature amounts is equal to or greater than a threshold, the face matching unit 306 determines that the person whose face is detected matches the specific person. The score / similarity output unit 309 transmits the score and similarity, which are the face matching results, to the center 207. Note that the face matching unit 306 may not be configured in this embodiment.

[0098] The parameter adjustment UI 310 performs processing to provide a UI (user interface) for adjusting detection parameters to the setting terminal 205. For example, the parameter adjustment UI 310 performs processing to display an operation screen such as that shown in FIG. 10 (described later) on the setting terminal 205. The mode processing unit 320 performs processing to operate the face authentication device 201 in a collection mode, an adjustment mode, or an operation mode. The collection mode corresponds to the image collection mode according to the first embodiment described above.

[0099] The collection mode processing unit 322 corresponds to the image collection mode processing unit 110 according to the first embodiment. The collection mode processing unit 322 has substantially the same functions as the image collection mode processing unit 110. Therefore, the collection mode processing unit 322 collects test image data in the collection mode by substantially the same method as the image collection mode processing unit 110. The test data storage unit 312 stores the test image data.

[0100] The adjustment mode processing unit 324 corresponds to the adjustment mode processing unit 120 according to the first embodiment described above. The adjustment mode processing unit 324 has substantially the same functions as the adjustment mode processing unit 120. The adjustment mode processing unit 324 switches between combinations of detection parameter values ​​in the adjustment mode and performs face detection processing on test image data stored in the test data storage unit 312. The adjustment mode processing unit 324 then determines a combination of detection parameter values ​​that maximizes the accuracy rate. The adjustment mode processing unit 324 also determines a combination of detection parameter values ​​that minimizes the false detection rate. The adjustment mode processing unit 324 also determines a combination of detection parameter values ​​that best balances the non-detection rate and the false detection rate. The adjustment mode processing unit 324 may also determine a combination of detection parameter values ​​that minimizes the non-detection rate. These determined combinations of detection parameter values ​​are used in the operation mode.

[0101] The operation mode processing unit 326 corresponds to the operation mode processing unit 140 according to the above-described first embodiment. The operation mode processing unit 326 has substantially the same functions as the operation mode processing unit 140. The operation mode processing unit 326 performs processing such that the face authentication processing unit 240 performs face detection processing using a combination of detection parameter values ​​determined in the adjustment mode.

[0102] Fig. 10 is a diagram illustrating an example of an operation screen displayed on the setting terminal 205 according to the second embodiment. A user (an operator, etc.) can adjust the values ​​of the detection parameters using the operation screen illustrated in Fig. 10. The user can perform a desired setting operation by operating the operation screen using an input device such as a mouse.

[0103] The user operates the collection mode setting UI 401 on the collection mode operation screen 400A to set the time periods for collecting image data that are candidates for test image data and the upper limit number of detections for each time period. The user can then start the collection mode by operating the start button 402A. The user can also stop the collection mode by operating the stop button 402B. The parameter adjustment UI 310 causes the setting terminal 205 to display the collected image data. The user can view the displayed image data and select test image data to be used for parameter adjustment. The user may also visually check the image data and manually select test image data.

[0104] The user can start the adjustment mode by operating start button 403A on adjustment mode operation screen 400B, and can stop the adjustment mode by operating stop button 403B on adjustment mode operation screen 400B.

[0105] In the adjustment mode, the adjustment mode processing unit 324 performs face detection processing on the test image data while switching between combinations of preset detection parameter values. At this time, face matching processing and score output are not performed. Note that the detection parameter values ​​are switched by sequentially changing the values ​​starting from the detection parameter that has the greatest effect on detection accuracy. Details will be described later using Figures 11 to 13.

[0106] Furthermore, the user can use the adjustment mode setting UI 404 to set the depth of adjustment, that is, the fineness of tuning. This setting value corresponds to the number of times the adjustment mode processing unit 324 performs a series of detection processes. Therefore, the larger this setting value, the finer the adjustment of the detection parameter values ​​will be. Furthermore, the user can use the adjustment mode setting UI 404 to set "automatic operation switching." If "automatic operation switching" is set to "on," the operation of the face recognition device 201 automatically transitions to the operation mode when the adjustment mode ends. At that time, the combination of detection parameter values ​​to be applied may be set in advance in the operation mode setting UI 406, which will be described later. Furthermore, the user can check the progress of the adjustment (remaining time) by visually checking the time display 405.

[0107] On the operation mode operation screen 400C, the operation mode setting UI 406 displays candidate combinations of detection parameter values ​​to be applied in the operation mode. The "Maximum Accuracy Rate" field may display the combination of detection parameter values ​​that maximizes the accuracy rate in face detection processing in the adjustment mode. The "Minimum False Detection Rate" field may display the combination of detection parameter values ​​that minimizes the False Detection Rate in face detection processing in the adjustment mode. The "Balance" field may display the combination of detection parameter values ​​that optimizes the balance between the non-detection rate and the False Detection Rate in face detection processing in the adjustment mode. The "Non-Detected / False Detection" field indicates the ratio of the number of non-detected data and the ratio of the number of False Detection data, assuming that the total number of incorrect data is 100%. In this case, "Non-Detected Rate" + "False Detection Rate" = 100. For example, if the number of incorrect data is 40, the number of non-detected data is 32, and the number of False Detection data is 8, the non-detected / False Detection ratio is (80 / 20).

[0108] The user can select from these combinations a combination of detection parameter values ​​to be actually applied in the operation mode by operating the operation mode setting UI 406. In the example of FIG. 10, a combination of detection parameter values ​​corresponding to the "minimum false detection rate" is selected. The user can start the operation mode by operating the start button 407A on the operation mode operation screen 400C. The user can stop the operation mode by operating the stop button 407B on the operation mode operation screen 400C. At this time, the user may operate the operation mode setting UI 406 to select another combination of detection parameter values. For example, although the combination of detection parameter values ​​corresponding to the "minimum false detection rate" is selected in the example of FIG. 10, a combination of detection parameter values ​​corresponding to the "maximum accuracy rate" may be selected instead.

[0109] Furthermore, the combination of detection parameter values ​​that was not applied can be saved by the function of the non-applied detection parameter value storage unit 130 according to the first embodiment described above. Then, the user can apply the combination of detection parameter values ​​that was not applied in the operation mode by operating the parameter save / restore button 408.

[0110] 11 to 13 are diagrams for explaining switching of detection parameters according to the second embodiment. FIG. 11 shows specific examples of first and second detection parameters. The first detection parameter 501 is a detection parameter used in the first detection process (first round of detection process). Therefore, the first detection parameter 501 is a detection parameter used in adjusting "depth 1." Note that the first detection parameter 501 is a detection parameter that has a greater influence on the accuracy of face detection than the second detection parameter 502 and the third detection parameter 503, which will be described later.

[0111] In the example of FIG. 11, the first detection parameters 501 are a score threshold, a maximum eye distance, and a minimum eye distance. The "score threshold" corresponds to a threshold for the score of face reliability in face detection processing. When the score of the face reliability (an index indicating face-likeness) of an object detected in face detection processing is equal to or greater than this score threshold, the object is detected as a face. The "maximum eye distance" corresponds to the maximum distance (e.g., the number of pixels) between two objects that can be determined to be eyes in face detection processing. The "minimum eye distance" corresponds to the minimum distance (e.g., the number of pixels) between two objects that can be determined to be eyes in face detection processing.

[0112] In the example of FIG. 11, the range (number) of values ​​for each of the score threshold, maximum inter-eye distance, and minimum inter-eye distance is three. The candidate setting values ​​for the score threshold are "0.2," "0.25," and "0.3." The candidate setting values ​​for the maximum inter-eye distance are "10," "20," and "30." The candidate setting values ​​for the minimum inter-eye distance are "50," "60," and "70."

[0113] In the example of FIG. 11, the second detection parameter 502 is a face angle range. The "face angle range" corresponds to the range of tilt angles of an object that can be determined as a face in face detection processing. In the example of FIG. 11, the range (number) of values ​​of the face angle range is three. Candidate setting values ​​for the face angle range are "15," "30," and "45."

[0114] 11, the third detection parameter 503 is a parameter used for fine adjustment. The third detection parameter 503 may be, for example, a maximum number of detected faces. The "maximum number of detected faces" corresponds to the upper limit of the number of faces that can be detected in one input image.

[0115] FIG. 12 is a diagram illustrating the first detection process according to the second embodiment. The adjustment mode processing unit 324 performs face detection processing on the test image data for each combination of the score threshold, maximum inter-eye distance, and minimum inter-eye distance. For example, the adjustment mode processing unit 324 performs face detection using combination #1 “score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 50.” Furthermore, for example, the adjustment mode processing unit 324 performs face detection using combination #2 “score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 60.” Furthermore, for example, the adjustment mode processing unit 324 performs face detection using combination #3 “score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 70.” Furthermore, for example, the adjustment mode processing unit 324 performs face detection using combination #4 “score threshold: 0.2, maximum inter-eye distance: 20, minimum inter-eye distance: 50.” Also, for example, the adjustment mode processing unit 324 performs face detection using combination #10 "score threshold: 0.25, maximum inter-eye distance: 10, minimum inter-eye distance: 50." Similarly, the adjustment mode processing unit 324 performs face detection using each of the 27 combinations (combinations #1 to #27). Then, as a result of performing face detection using each combination, the adjustment mode processing unit 324 obtains values ​​for the accuracy rate, non-detection rate, and false detection rate for each combination.

[0116] Then, the adjustment mode processing unit 324 searches for the combination with the best accuracy rate value among the detection results of combinations #1 to #27. In the example of Fig. 12, when face detection is performed using combination #6 of the detection result 601, "score threshold: 0.2, maximum inter-eye distance: 20, minimum inter-eye distance: 70", the best accuracy rate value is "94.3%".

[0117] Furthermore, the adjustment mode processing unit 324 searches for the combination with the best false detection rate among the detection results of combinations #1 to #27. In the example of Fig. 12, when face detection is performed using combination #3 of detection result 602, "score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 70", the false detection rate is the best value of "2.0%".

[0118] Furthermore, the adjustment mode processing unit 324 searches for the combination that provides the best balance between the non-detection rate and the false detection rate among the detection results of combinations #1 to #27. In the example of FIG. 12, when face detection is performed using combination #26 (score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 60), the non-detection rate and the false detection rate are equal to each other, i.e., 3.0%. When face detection is performed using combination #27 (score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 70), the non-detection rate and the false detection rate are equal to each other, i.e., 3.7%. The accuracy rate of 94.0% obtained when face detection is performed using combination #26 is better than the accuracy rate of 92.7% obtained when face detection is performed using combination #27. Therefore, when face detection is performed using combination #26 of the detection result 603, the balance between the non-detection rate and the false detection rate is best.

[0119] The adjustment mode processing unit 324 performs a second detection processing according to the detection results of the first detection processing. The adjustment mode processing unit 324 uses, in the second detection processing, the combination with the best accuracy rate, the combination with the best false detection rate, and the combination with the best balance between the non-detection rate and the false detection rate. In the example of FIG. 12, combination #6 of the detection result 601, combination #3 of the detection result 602, and combination #26 of the detection result 603 are used in the second detection processing. That is, the adjustment mode processing unit 324 changes the values ​​of the second detection parameters based on the combination of detection parameter values ​​determined in the first detection processing.

[0120] The adjustment mode processing unit 324 performs face detection using a combination of "score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 70" and candidate values ​​of the face angle range: "15," "30," and "45." In the example of FIG. 13, these three combinations correspond to combinations #1 to #3. The adjustment mode processing unit 324 performs face detection using combination #1 "score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 70, face angle range: 15." The adjustment mode processing unit 324 also performs face detection using combination #2 "score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 70, face angle range: 30." The adjustment mode processing unit 324 also performs face detection using combination #3 "score threshold: 0.2, maximum inter-eye distance: 10, minimum inter-eye distance: 70, face angle range: 45."

[0121] The adjustment mode processing unit 324 also performs face detection using a combination of "score threshold: 0.2, maximum inter-eye distance: 20, minimum inter-eye distance: 70" and candidate values ​​for the face angle range: "15," "30," and "45." In the example of FIG. 13, these three combinations correspond to combinations #4 to #6. The adjustment mode processing unit 324 performs face detection using combination #4 "score threshold: 0.2, maximum inter-eye distance: 20, minimum inter-eye distance: 70, face angle range: 15." The adjustment mode processing unit 324 also performs face detection using combination #5 "score threshold: 0.2, maximum inter-eye distance: 20, minimum inter-eye distance: 70, face angle range: 30." The adjustment mode processing unit 324 also performs face detection using combination #6 "score threshold: 0.2, maximum inter-eye distance: 20, minimum inter-eye distance: 70, face angle range: 45."

[0122] The adjustment mode processing unit 324 also performs face detection using a combination of "score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 60" and candidate values ​​of the face angle range: "15," "30," and "45." In the example of FIG. 13, these three combinations correspond to combinations #7 to #9. The adjustment mode processing unit 324 performs face detection using combination #7 "score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 60, face angle range: 15." The adjustment mode processing unit 324 also performs face detection using combination #8 "score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 60, face angle range: 30." The adjustment mode processing unit 324 also performs face detection using combination #9 "score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 60, face angle range: 45."

[0123] Then, the adjustment mode processing unit 324 searches for the combination with the best accuracy rate value among the detection results of combinations #1 to #9. In the example of Fig. 13, when face detection is performed using combination #4 of the detection result 701, "score threshold: 0.2, maximum inter-eye distance: 20, minimum inter-eye distance: 70, face angle range: 15", the accuracy rate value becomes the best "94.4%".

[0124] Furthermore, the adjustment mode processing unit 324 searches for the combination with the best false detection rate among the detection results of combinations #1 to #9. In the example of Fig. 13, when face detection is performed using combination #3 of detection result 702, "score threshold: 0.2, maximum eye distance: 10, minimum eye distance: 70, face angle range: 45", the false detection rate is the best value of "2.0%".

[0125] Furthermore, the adjustment mode processing unit 324 searches for the combination that provides the best balance between the non-detection rate and the false detection rate among the detection results of combinations #1 to #9. In the example of FIG. 13, when face detection is performed using combination #8 (score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 60, face angle range: 30), the non-detection rate and the false detection rate are equal to each other, 2.9%. When face detection is performed using combination #9 (score threshold: 0.3, maximum inter-eye distance: 30, minimum inter-eye distance: 60, face angle range: 45), the non-detection rate and the false detection rate are equal to each other, 3.0%. The accuracy rate of 94.2% obtained when face detection is performed using combination #8 is better than the accuracy rate of 94.0% obtained when face detection is performed using combination #9. Therefore, when face detection is performed using combination #8 of the detection result 703, the balance between the non-detection rate and the false detection rate is best. Note that the combination that best balances the detection rate and the false detection rate does not necessarily have to have the highest accuracy rate.

[0126] The adjustment mode processing unit 324 sets the combination #4 of the detection results 701, the combination #3 of the detection results 702, and the combination #8 of the detection results 703 obtained as described above as combinations of detection parameter values ​​that can be applied in the operation mode. Note that if it is desired to further adjust the detection parameter values, the adjustment mode processing unit 324 may perform fine adjustment using the third detection parameter 503.

[0127] (Embodiment 3) Next, a third embodiment will be described with reference to the drawings. For clarity of explanation, the following description and drawings have been omitted and simplified as appropriate. In addition, in each drawing, the same elements are given the same reference numerals, and duplicate explanations are omitted as necessary. The third embodiment differs from the other embodiments described above in that image data is acquired from multiple cameras (imaging devices).

[0128] FIG. 14 is a diagram illustrating a human detection system 20A according to the third embodiment. The human detection system 20A includes a plurality of imaging devices 30A and 30B and an information processing device 100. The imaging devices 30A and 30B and the information processing device 100 are communicably connected to each other via a network 22. Although two imaging devices 30 are illustrated in FIG. 14, the number of imaging devices 30 is arbitrary. The functions of the imaging devices 30A and 30B are substantially similar to those of the imaging device 30 according to the first embodiment described above, and therefore detailed descriptions thereof will be omitted except for the following description. The configuration of the information processing device 100 is also substantially similar to that of the information processing device 100 according to the first embodiment described above, and therefore detailed descriptions thereof will be omitted except for the following description.

[0129] The imaging devices 30A and 30B are installed at different positions. Therefore, the imaging devices 30A and 30B capture images from a plurality of different viewpoints. Therefore, the imaging devices 30A and 30B generate image data obtained from the plurality of different viewpoints.

[0130] In the information processing device 100, the image data acquisition unit 112 (image collection mode processing unit 110) acquires a plurality of image data from each of the imaging devices 30A and 30B. That is, the image data acquisition unit 112 acquires a plurality of image data obtained from a plurality of different viewpoints.

[0131] The test image selection processing unit 114 performs processing to select test image data for each of the plurality of image data obtained from each of the plurality of viewpoints. For example, the test image selection processing unit 114 may perform processing to select test image data from the image data obtained by the image capture device 30A, and then perform processing to select test image data from the image data obtained by the image capture device 30B. In this way, test image data is selected separately for each of the plurality of image data obtained from each of the plurality of viewpoints. The test image storage unit 116 may store test image data separately for each of the plurality of viewpoints.

[0132] The person detection unit 124 detects people from each of the test image data selected for each of the plurality of image data obtained from each of the plurality of viewpoints. Note that the person detection method is substantially the same as the method according to the first embodiment described above, and therefore a description thereof will be omitted. For example, the person detection unit 124 performs person detection on the test image data selected from the image data obtained by the image pickup device 30A. Then, the person detection unit 124 performs person detection on the test image data selected from the image data obtained by the image pickup device 30B.

[0133] Then, the person detection unit 124 acquires detection results including at least one accuracy parameter for each test image data selected for each of the multiple image data obtained from the multiple viewpoints. For example, the person detection unit 124 acquires detection results obtained by performing person detection on the test image data related to the image capture device 30A. The person detection unit 124 also acquires detection results obtained by performing person detection on the test image data related to the image capture device 30B. This makes it possible to determine which of the detection results using the test image data related to the image capture device 30A and the detection results using the test image data related to the image capture device 30B has higher detection accuracy. That is, it is possible to determine which of the installation positions of the image capture device 30A and the image capture device 30B has higher detection accuracy. In other words, it is possible to determine from which of the multiple viewpoints shooting will result in higher detection accuracy. This makes it possible to efficiently determine the installation position that will result in higher detection accuracy.

[0134] (Variation) The present invention is not limited to the above-described embodiments, and can be modified as appropriate without departing from the spirit of the present invention. For example, the above-described embodiments can be mutually applied. For example, the function of the person detection system 20A according to the third embodiment may be realized by the face authentication system 200 according to the second embodiment.

[0135] In addition, in the above-described flowchart, the order of each process (step) can be changed as appropriate. Furthermore, one or more of the multiple processes (steps) may be omitted. For example, in the flowchart of FIG. 5, the process of S130 may be omitted.

[0136] Furthermore, in the third embodiment described above, the number of imaging devices 30 is plural, but the present invention is not limited to this configuration. One imaging device 30 may be relocated to a different position. Even with this configuration, imaging can be performed from a plurality of different viewpoints.

[0137] In the above examples, the program includes instructions (or software code) that, when loaded into a computer, cause the computer to perform one or more functions described in the embodiments. The program may be stored on a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disk (DVD), Blu-ray® disk or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may also be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, transitory computer-readable medium or communication medium includes electrical, optical, acoustic, or other forms of propagated signals.

[0138] A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes. (Appendix 1) image data acquisition means for acquiring a plurality of image data; a detection means for detecting a person from one or more test image data items selected from the plurality of image data items and including an image of a person, using each of a plurality of combinations of a plurality of values ​​of each of a plurality of detection parameters used for detecting a person in an adjustment stage; a determination means for determining a combination of values ​​of the detection parameters to be applied in the person detection process in the operation stage based on the detection result in the adjustment stage; An information processing device having the above. (Appendix 2) the determination means determines a combination of values ​​of the detection parameters to be applied in the person detection process in the operation phase based on the detection result obtained when the person is detected in the adjustment phase, the detection result including at least one accuracy parameter indicating accuracy of person detection. 10. The information processing device according to claim 1. (Appendix 3) The detection means performing a first detection process using a combination of values ​​of one or more first detection parameters among the plurality of detection parameters; performing a second detection process using a combination of values ​​of a second detection parameter other than at least the first detection parameter among the plurality of detection parameters according to the detection result obtained in the first detection process; 10. The information processing device according to claim 1. (Appendix 4) the detection result includes at least one accuracy parameter indicating accuracy of the person detection; The detection means performing the first detection process using a combination of values ​​of the first detection parameters, which are set according to the influence of the combination of values ​​on the accuracy of human detection among the plurality of detection parameters; performing the second detection process using a combination of the value of the first detection parameter and the value of the second detection parameter selected according to the detection accuracy in the detection result obtained in the first detection process; 4. The information processing device according to claim 3. (Appendix 5) the determining means determines a plurality of combinations of values ​​of the detection parameters that can be applied in an operation phase according to the detection result including a plurality of the accuracy parameters; 3. The information processing device according to claim 2. (Appendix 6) the determination means determines, from among a plurality of combinations of values ​​of the detection parameters that can be applied in the operation phase, which are determined in accordance with each of the plurality of accuracy parameters, a combination of values ​​of the detection parameters that corresponds to the accuracy parameter set in accordance with the operation mode, as the combination of values ​​of the detection parameters to be applied in the operation phase; 6. The information processing device according to claim 5. (Appendix 7) non-applicable detection parameter value storage means for storing one or more combinations of the detection parameter values ​​other than the combinations of the detection parameter values ​​applied in the person detection process in the operational stage; 2. The information processing device according to claim 1, further comprising: (Appendix 8) the image data acquisition means acquires a plurality of image data obtained from a plurality of different viewpoints, the detection means detects a person from each of the test image data selected for each of the plurality of image data obtained from each of the plurality of viewpoints, and obtains the detection result including at least one accuracy parameter indicating accuracy of person detection. 10. The information processing device according to claim 1. (Appendix 9) Acquire multiple image data, In the adjustment stage, detecting a person from one or more test image data selected from the plurality of image data and including an image of a person, using each of a plurality of combinations of a plurality of values ​​of each of a plurality of detection parameters used to detect a person; determining a combination of values ​​of the detection parameters to be applied in the person detection process in the operation stage based on the detection results in the adjustment stage; Parameter adjustment method. (Appendix 10) determining a combination of values ​​of the detection parameters to be applied in the person detection process in the operation phase based on the detection result including at least one accuracy parameter indicating the accuracy of person detection obtained when the person is detected in the adjustment phase; 10. The parameter adjustment method according to claim 9. (Appendix 11) performing a first detection process using a combination of values ​​of one or more first detection parameters among the plurality of detection parameters; performing a second detection process using a combination of values ​​of a second detection parameter other than at least the first detection parameter among the plurality of detection parameters according to the detection result obtained in the first detection process; 10. The parameter adjustment method according to claim 9. (Appendix 12) the detection result includes at least one accuracy parameter indicating accuracy of the person detection; performing the first detection process using a combination of values ​​of the first detection parameters, which are set according to the influence of the combination of values ​​on the accuracy of human detection among the plurality of detection parameters; performing the second detection process using a combination of the value of the first detection parameter and the value of the second detection parameter selected according to the detection accuracy in the detection result obtained in the first detection process; 12. The parameter adjustment method according to claim 11. (Appendix 13) determining a plurality of combinations of values ​​of the detection parameters that can be applied in an operational phase according to the detection result including a plurality of the accuracy parameters; 11. The parameter adjustment method according to claim 10. (Appendix 14) determining, from among a plurality of combinations of values ​​of the detection parameters that can be applied in the operation phase, which are determined in accordance with each of the plurality of accuracy parameters, a combination of values ​​of the detection parameters that corresponds to the accuracy parameters set in accordance with an operation mode, as a combination of values ​​of the detection parameters that will be applied in the operation phase; 14. The parameter adjustment method according to claim 13. (Appendix 15) storing one or more combinations of values ​​of the detection parameters other than the combinations of values ​​of the detection parameters applied in the person detection process in the operational phase; 10. The parameter adjustment method according to claim 9. (Appendix 16) acquiring a plurality of image data obtained from a plurality of different viewpoints, detecting a person from each of the test image data selected for each of the plurality of image data obtained from each of the plurality of viewpoints, and obtaining the detection result including at least one accuracy parameter indicating accuracy of person detection; 10. The parameter adjustment method according to claim 9. (Appendix 17) acquiring a plurality of image data; In the adjustment stage, detecting a person from one or more test image data selected from the plurality of image data and including an image of a person, using each of a plurality of combinations of a plurality of values ​​of each of a plurality of detection parameters used to detect a person; determining a combination of values ​​of the detection parameters to be applied in a person detection process in an operation stage based on the detection results in the adjustment stage; A program that causes a computer to execute the following. [Explanation of symbols]

[0139] 1. Information processing equipment 2 Image data acquisition unit 4. Detection unit 6 Decision Section 20 Person Detection System 22 Network 30 Imaging device 100 Information processing device 110 Image acquisition mode processing section 112 Image data acquisition unit 114 Test image selection processing unit 116 Test image storage section 120 Adjustment mode processing section 122 Detection parameter storage section 124 Person detection unit 126 Detection parameter value determination unit 128 Detection parameter value storage section 130 Non-applicable detection parameter value storage section 140 Operational mode processing unit 142 Detection parameter value selection processing unit 144 Person detection processing execution unit 200 Facial Recognition System 201 Facial Recognition Device 205 Setting terminal 206 Camera 207 Center 210 Camera image input unit 230 Parameter Adjustment Section 240 Face recognition processing unit 303 Feature storage unit 304 Temporary storage section 305 Face detection unit 306 Face matching unit 307 Detection parameter storage section 308 Location information storage section 309 Score / Similarity Output Unit 310 Parameter Adjustment UI 312 Test Data Storage 320 Mode Processing Section 322 Acquisition mode processing section 324 Adjustment mode processing section 326 Operational Mode Processing Unit 400A Acquisition Mode Operation Screen 400B Adjustment Mode Operation Screen 400C Operational Mode Operation Screen 401 Collection mode setting UI 402A Start button 402B Stop button 403A Start button 403B Stop button 404 Adjustment mode setting UI 405 Time Display 406 Operation Mode Setting UI 407A Start button 407B Stop button 408 Parameter save / restore button 501 first detection parameter 502 Second detection parameter 503 Third detection parameter 601 detection results 602 detection results 603 Detection Results 701 detection results 702 detection results 703 Detection Results

Claims

1. An image data acquisition means for acquiring multiple image data, In the adjustment phase, a detection means for detecting a person from one or more test image data containing an image of a person, selected from the multiple image data, using each of the multiple combinations of values ​​for each of the multiple detection parameters used to detect a person, A determination means for determining a combination of values ​​for the detection parameters to be applied in the person detection process during the operational phase, based on the detection results during the adjustment phase. It has, The image data acquisition means acquires a plurality of image data obtained from each of a plurality of different viewpoints, The detection means detects a person from each of the test image data selected for each of the multiple image data obtained from each of the multiple viewpoints, and obtains the detection result which includes at least one accuracy parameter indicating the accuracy of person detection. Information processing device.

2. The determination means determines a combination of values ​​for the detection parameters to be applied in the person detection process during the operation phase, based on the detection result obtained when a person is detected during the adjustment phase, which includes at least one accuracy parameter indicating the accuracy of person detection. The information processing apparatus according to claim 1.

3. The detection means is A first detection process is performed using a combination of values ​​for one or more first detection parameters from the plurality of detection parameters. In accordance with the detection results obtained in the first detection process, a second detection process is performed using a combination of values ​​of at least the second detection parameter other than the first detection parameter from among the plurality of detection parameters. The information processing apparatus according to claim 1.

4. The detection result includes at least one accuracy parameter indicating the accuracy of person detection. The detection means is Using a combination of values ​​of the first detection parameter, which is set according to its influence on the accuracy of person detection among the plurality of detection parameters, the first detection process is performed. The second detection process is performed using a combination of the value of the first detection parameter, selected according to the detection accuracy in the detection result obtained in the first detection process, and the value of the second detection parameter. The information processing apparatus according to claim 3.

5. The determination means determines, in accordance with the detection result which includes a plurality of accuracy parameters, a plurality of combinations of values ​​of the detection parameters that can be applied during the operational phase. The information processing apparatus according to claim 2.

6. The determination means determines, from among a plurality of combinations of detection parameter values ​​that can be applied in the operational stage, which are determined according to each of the plurality of precision parameters, the combination of detection parameter values ​​that corresponds to the precision parameter set according to the operational mode, as the combination of detection parameter values ​​to be applied in the operational stage. The information processing apparatus according to claim 5.

7. An unapplicable detection parameter value storage means for storing one or more combinations of detection parameter values ​​other than the combination of detection parameter values ​​applied in the person detection process during the operational phase, The information processing apparatus according to claim 1, further comprising:

8. Image data acquisition step to acquire multiple image data, In the adjustment phase, a detection step is performed to detect a person from one or more test image data containing an image of a person, selected from the multiple image data, using each of the multiple combinations of values ​​for each of the multiple detection parameters used to detect a person, A decision step in which, based on the detection results in the adjustment phase, a combination of values ​​for the detection parameters to be applied in the person detection process in the operation phase is determined, Includes, The image data acquisition step involves acquiring multiple image data obtained from each of several different viewpoints, The detection step involves detecting a person from each of the test image data selected for each of the multiple image data obtained from each of the multiple viewpoints, and obtaining the detection result which includes at least one accuracy parameter indicating the accuracy of person detection. Parameter adjustment method.

9. Image data acquisition step to acquire multiple image data, In the adjustment phase, a detection step is performed to detect a person from one or more test image data containing an image of a person, selected from the multiple image data, using each of the multiple combinations of values ​​for each of the multiple detection parameters used to detect a person, A decision step in which, based on the detection results in the adjustment phase, a combination of values ​​for the detection parameters to be applied in the person detection process in the operation phase is determined, Have the computer run it, The image data acquisition step involves acquiring multiple image data obtained from each of several different viewpoints, The detection step involves detecting a person from each of the test image data selected for each of the multiple image data obtained from each of the multiple viewpoints, and obtaining the detection result which includes at least one accuracy parameter indicating the accuracy of person detection. program.

Images