Person detection device, person detection system, person detection method, and person detection program

Abstract

A person detection device (100) comprising an acquisition unit (110) which acquires a video captured by an imaging device installed in a vehicle, a first detection unit (120) which detects a person image region estimated to include an image of a person from the video, a measurement unit (130) which measures the number of pixels of this person image region, an image processing unit (140) which applies resolution increasing processing to the person image region when the number of pixels of the person image region is less than a predetermined threshold value, and a second detection unit (150) which detects the person from the person image region to which the resolution increasing processing has been applied.

Description

[Technical Field]

[0001] The present disclosure relates to a person detection apparatus, a person detection system, a person detection method, and a non-transitory computer-readable medium. [Background Art]

[0002] When a person or the like appears in a video acquired by an imaging device such as a drive recorder mounted on a vehicle, masking processing may be performed on the region including the person from the perspective of personal information protection.

[0003] Patent Document 1 describes a technique for performing masking processing on regions such as faces of persons other than the person to be monitored in an image captured by a surveillance camera. [Prior Art Literature] [Patent Literature]

[0004] [Patent Document 1] Japanese Patent Laid-Open No. 2008-017093 [Summary of the Invention] [Problem to be Solved by the Invention]

[0005] In order to determine a region to be subjected to masking processing in a video acquired by an imaging device, a person is detected from the video. If masking processing is performed on all detected person regions, this is not preferable because there is a possibility that masking processing may be performed even on regions that do not require masking. For this reason, for example, even for a region including a person, if the size (number of pixels) of the region is small, an individual cannot normally be identified, so masking processing is not performed. However, there is a problem that a person may be identified even in a region having a small size (number of pixels). Therefore, there is a demand for more accurate detection of a person region to be subjected to masking processing from a video.

[0006] This disclosure is made to solve these problems and aims to provide a person detection device, a person detection system, a person detection method, and a non-temporary computer-readable medium that can accurately detect the area of ​​a person from video. [Means for solving the problem]

[0007] A person detection device according to a first aspect of the present disclosure includes: acquisition means for acquiring video footage captured by a camera installed in a vehicle; a first person detection means for detecting a person image region from the video footage, which is a region estimated to contain an image of a person; measurement means for measuring the number of pixels in the person image region; high-resolution means for performing high-resolution processing on the person image region when the number of pixels in the person image region is less than a predetermined threshold; and a second person detection means for detecting a person from the person image region on which the high-resolution processing has been performed.

[0008] A person detection system according to a second aspect of the present disclosure comprises a shooting device installed in a vehicle and shooting images of the area around the vehicle, and a person detection device capable of communicating with the shooting device, wherein the person detection device comprises an acquisition means for acquiring images captured by the shooting device, a first person detection means for detecting a person image region which is an area estimated to contain an image of a person from the images, a measurement means for measuring the number of pixels in the person image region, a high-resolution means for performing a high-resolution processing on the person image region when the number of pixels in the person image region is less than a predetermined threshold, and a second person detection means for detecting a person from the person image region on which the high-resolution processing has been performed.

[0009] A third aspect of the present disclosure relates to a person detection method in which a computer acquires video footage captured by a camera installed in a vehicle, performs a first person detection process to detect a person image region from the video footage which is an area estimated to contain an image of a person, measures the number of pixels in the person image region, performs a high-resolution processing on the person image region if the number of pixels in the person image region is less than a predetermined threshold, and performs a second person detection process to detect a person from the person image region on which the high-resolution processing has been performed.

[0010] A non-temporary computer-readable medium according to a fourth aspect of this disclosure stores a person detection program that causes a computer to perform the following: a process for acquiring video footage captured by a camera installed in a vehicle; a first person detection process for detecting a person image region, which is an area estimated to contain an image of a person, from the video footage; a process for measuring the number of pixels in the person image region; a high-resolution processing process for increasing the resolution of the person image region if the number of pixels in the person image region is less than a predetermined threshold; and a second person detection process for detecting a person from the person image region after the high-resolution processing has been performed. [Effects of the Invention]

[0011] This invention provides a person detection device, a person detection system, a person detection method, and a non-temporary computer-readable medium that can accurately detect the area of ​​a person from video footage. [Brief explanation of the drawing]

[0012] [Figure 1] This is a block diagram showing the configuration of the person detection device according to Embodiment 1. [Figure 2] This figure shows an example of video footage acquired by the acquisition unit according to Embodiment 1. [Figure 3] This flowchart shows the person detection method according to Embodiment 1. [Figure 4] This is a block diagram showing the configuration of the person detection system according to Embodiment 2. [Figure 5]This is a block diagram showing the configuration of the person detection device according to Embodiment 2. [Figure 6] This flowchart shows the person detection method according to Embodiment 2. [Modes for carrying out the invention]

[0013] Embodiments of the present disclosure will be described in detail below with reference to the drawings. In each drawing, the same or corresponding elements are denoted by the same reference numerals, and redundant explanations will be omitted where necessary for clarity. <Embodiment 1> Figure 1 is a block diagram showing the configuration of a person detection device 100 according to Embodiment 1. The person detection device 100 comprises an acquisition unit 110 as an acquisition means, a first detection unit 120 as a first person detection means, a measurement unit 130 as a measurement means, an image processing unit 140 as a high-resolution means, and a second detection unit 150 as a second person detection means. The person detection device 100 is connected to a network 500 (not shown). The network 500 may be wired or wireless. A camera 310 (not shown) and the like are connected to the network 500. The camera 310 is installed on a vehicle 300 (not shown) and is a device that photographs the area around the vehicle 300. The video captured by the camera 310 is a video and includes a plurality of consecutive frames arranged in order of the progression of shooting time. Here, a frame is still image data captured by the camera 310.

[0014] The acquisition unit 110 acquires video footage captured by the camera 310 installed on the vehicle 300. The video footage includes multiple frames. The video footage captured by the camera 310 is transmitted from the camera 310 to the person detection device 100 via the network 500.

[0015] The first detection unit 120 detects a person image region, which is an area estimated to contain an image of a person, from the video acquired by the acquisition unit 110. Specifically, the first detection unit 120 detects a person image region for each frame included in the video. For example, the first detection unit 120 calculates a person detection score from each frame using a trained person detection model (not shown) for each detection frame region of a predetermined size (e.g., M × N pixels (M and N are integers of 2 or more)), and defines a person image region as an area where the person detection score is equal to or greater than a predetermined threshold for the person detection score. The person detection score is higher in areas where there is a high probability of a person being present, compared to other areas. The predetermined threshold for the person detection score is a preset value and is used when detecting a person from frames that make up the video. The threshold for the person detection score may be set to a different value depending on the area within the frame. In this specification, machine learning may be deep learning, but is not particularly limited. Also, the method for calculating the person detection score is not limited to the above, and other existing technologies can be applied.

[0016] The measurement unit 130 measures the number of pixels in the human image region detected by the first detection unit 120.

[0017] The image processing unit 140 performs high-resolution processing when the number of pixels in the person image region measured by the measurement unit 130 is less than a predetermined threshold. Here, the predetermined threshold is generally the lower limit of the number of pixels in the person image region that can identify the person as a specific individual, for example, 40 pixels. Furthermore, the high-resolution processing involves, for example, increasing the resolution of the person image region using a trained high-resolution processing model (not shown). The high-resolution processing model is, for example, a VDSR (Very-Deep Super-Resolution) neural network. Note that the method of high-resolution processing is not limited to the above, and other existing technologies can be applied.

[0018] The second detection unit 150 detects a person from the person image region that has been subjected to high-resolution processing by the image processing unit 140. For example, the second detection unit 150 calculates a person detection score for said person image region by the same method as the first detection unit 120. Then, when the calculated person detection score is equal to or higher than a predetermined threshold for the person detection score, the second detection unit 150 determines that a person is captured in said person image region.

[0019] FIG. 2 shows an example of a frame 10 included in the video acquired by the acquisition unit 110. The frame 10 shown in FIG. 2 includes persons 30A and 30B walking on a sidewalk, and a vehicle 40 traveling on a roadway 50 toward the front. In the example shown in FIG. 2, the first detection unit 120 detects person image regions R1 and R2, which are regions including images of the persons 30A and 30B. In FIG. 6, said person image regions R1 and R2 are indicated by broken lines. Next, the measurement unit 130 measures the number of pixels of said person image regions R1 and R2 respectively. Next, when the number of pixels of the person image regions R1 and R2 is less than a predetermined threshold, the image processing unit 140 performs high-resolution processing on said person image regions R1 and R2. For example, in FIG. 2, when the number of pixels of the person image region R1 is equal to or greater than the predetermined threshold, and the number of pixels of the person image region R2 is less than the predetermined threshold, the image processing unit 140 performs high-resolution processing on the person image region R2. Next, the second detection unit 150 detects a person from the person image region R2 that has been subjected to the high-resolution processing.

[0020] Next, the person detection method according to Embodiment 1 will be described with reference to FIG. 3. First, the acquisition unit 110 acquires a video captured by an imaging device installed on a vehicle (step S101). Next, the first detection unit 120 detects a person image area from a frame included in the video based on a predetermined person detection score threshold (step S102). Next, the measurement unit 130 measures the number of pixels in the person image area (step S103). Next, the image processing unit 140 determines whether the number of pixels in the person image area is less than a predetermined threshold (step S104). In step S104, if the number of pixels in the person image area is equal to or greater than the predetermined threshold (step S104; No), it is determined that the person image area is an image area in which an individual person can be identified, and this process ends. In step S104, if the number of pixels in the person image area is less than the predetermined threshold (step S104; Yes), the image processing unit 140 performs super-resolution processing on the person image area (step S105). Next, the second detection unit 150 detects a person from the person image area that has been super-resolved in step S105 based on a predetermined person detection score threshold (step S106). If the person detection score calculated in step S106 is equal to or greater than the predetermined person detection score threshold, the second detection unit 150 determines that the person image area is an image area in which an individual person can be identified.

[0021] As described above, the person detection device 100 according to this embodiment performs high-resolution processing on areas of a person's image where the number of pixels is less than a predetermined threshold, which are not normally subject to masking, and then performs person detection again. By performing high-resolution processing on areas of a person's image with a small number of pixels, if a person is depicted in that area or if a person is depicted clearly enough to identify an individual in that area, the person detection score for that area will increase. On the other hand, if no person is depicted in that area, performing high-resolution processing on that area will not increase or will decrease the person detection score. Therefore, by performing high-resolution processing on areas of a person's image with a small number of pixels and then performing person detection again, it becomes possible to perform masking processing on areas of a person's image where a person is depicted clearly enough to identify an individual. Thus, the person detection device 100 according to this embodiment can accurately detect areas of a person that are subject to masking processing from video.

[0022] The person detection device 100 includes a processor, memory, and storage device (not shown in the figures). The storage device stores a computer program that implements the processing of the person detection method according to this embodiment. The processor loads the computer program from the storage device into the memory and executes the computer program. As a result, the processor functions as an acquisition unit 110, a first detection unit 120, a measurement unit 130, an image processing unit 140, and a second detection unit 150.

[0023] Furthermore, the acquisition unit 110, the first detection unit 120, the measurement unit 130, the image processing unit 140, and the second detection unit 150 may each be implemented with dedicated hardware. Also, some or all of the components of each device may be implemented by general-purpose or dedicated circuits, processors, etc., or combinations thereof. These may be configured by a single chip or by multiple chips connected via a bus. Some or all of the components of each device may be implemented by a combination of the above-mentioned circuits, etc., and programs. In addition, a CPU (Central Processing Unit), GPU (Graphics Processing Unit), FPGA (field-programmable gate array), etc. can be used as the processor.

[0024] Furthermore, if some or all of the components of the person detection device 100 are implemented by multiple information processing devices or circuits, these devices may be centrally located or distributed. For example, the information processing devices or circuits may be implemented in a form in which each is connected via a communication network, such as a client-server system or a cloud computing system. Also, the functions of the person detection device 100 may be provided in SaaS (Software as a Service) format.

[0025] <Embodiment 2> Figure 4 is a block diagram showing the configuration of the person detection system 200 according to Embodiment 2. The person detection system 200 comprises at least a shooting device 310 and a person detection device 400, and may further comprise a recording device 320. The shooting device 310 and the recording device 320 are each connected to the person detection device 400 via a network 500. Descriptions that overlap with Embodiment 1 will be omitted as appropriate.

[0026] The person detection system 200 is a system for detecting people from images captured in a vehicle 300. The vehicle 300 is, for example, an automobile, but may also be a vehicle other than an automobile, such as a motorcycle or bicycle. The vehicle 300 is equipped with a camera 310 and a recording device 320. The camera 310 is a device that captures the scenery around the vehicle 300, and is, for example, a drive recorder. The camera 310 comprises a camera unit 311 and a communication unit 312. The camera unit 311 is a camera. The camera unit 311 captures, for example, the scenery in front of the vehicle 300, that is, the scenery that a driver can see while seated in the driver's seat of the vehicle 300. The communication unit 312 is a communication interface with the network 500. The communication unit 312 transmits the images captured by the camera unit 311 to the person detection device 400 via the network 500.

[0027] The recording device 320 is a device for recording the travel speed of the vehicle 300. The recording device 320 comprises a measuring unit 321 and a communication unit 322. The measuring unit 321 measures the travel speed of the vehicle 300. The communication unit 322 is a communication interface with the network 500. The communication unit 322 transmits speed information, including the speed measured by the measuring unit 321, to the person detection device 400 via the network 500.

[0028] Next, the configuration of the person detection device 400 will be described in detail with reference to Figure 5. Figure 5 is a block diagram showing the configuration of the person detection device 400. The person detection device 400 comprises a memory 410, a communication unit 420, a storage unit 430, and a control unit 440.

[0029] Memory 410 is a storage area that temporarily stores the processing contents of the control unit 440, and is a volatile storage device such as RAM (Random Access Memory). Communication unit 420 is an interface for communicating with the outside of the person detection device 400. Storage unit 430 is a storage device that stores the first threshold 431 of the person detection score, the second threshold 432 of the person detection score, and the program 433, etc. The first threshold 431 of the person detection score is a numerical value used by the first detection unit 442 (described later) when detecting a person from a frame included in the video, and different values ​​may be set depending on the area within the frame. The second threshold 432 of the person detection score is a numerical value used by the second detection unit 445 (described later) when detecting a person from the person image area. The program 433 is a computer program on which the person detection process according to this embodiment is implemented.

[0030] The control unit 440 comprises an acquisition unit 441, a first detection unit 442, a measurement unit 443, an image processing unit 444, a second detection unit 445, and a masking unit 446. The control unit 440 is a control device that controls the operation of the person detection device 400, and is, for example, a processor such as a CPU. The control unit 440 loads the program 433 from the storage unit 430 into the memory 410 and executes it. As a result, the control unit 440 realizes the functions of the acquisition unit 441, the first detection unit 442, the measurement unit 443, the image processing unit 444, the second detection unit 445, and the masking unit 446.

[0031] The acquisition unit 441 acquires video transmitted from the camera 310. The video contains multiple consecutive frames. The video may also contain identification information and time information. The identification information is information for identifying the vehicle 300 on which the camera 310 that captured the video is installed. The time information is information about the time the video was captured. Furthermore, the acquisition unit 441 may acquire speed information transmitted from the recording device 320. The speed information includes at least information about the vehicle 300's travel speed, and may also include identification information and time information. The time information included in the speed information is information about the time the travel speed was recorded.

[0032] The first detection unit 442 detects human image regions for each frame included in the video acquired by the acquisition unit 110. Specifically, the first detection unit 442 of Embodiment 1 calculates a human detection score using the same method as the first detection unit 120, and defines regions where the human detection score is equal to or greater than the first threshold 431 of the human detection score as human image regions.

[0033] The measurement unit 443 measures the number of pixels in the person image region detected by the first detection unit 442, similar to the measurement unit 130 in Embodiment 1.

[0034] The image processing unit 444, similar to the image processing unit 140 in Embodiment 1, performs high-resolution processing when the number of pixels in the human image region measured by the measurement unit 443 is less than a predetermined threshold.

[0035] The second detection unit 445 detects a person from the person image region that has been processed for high resolution by the image processing unit 444. For example, the second detection unit 445 calculates a person detection score for the person image region using the same method as the first detection unit 442. The second detection unit 445 then determines that a person is present in the person image region if the calculated person detection score is equal to or greater than the second threshold 432 for person detection scores.

[0036] Here, the second threshold 432 is a higher value than the first threshold 431. The second detection unit 445 detects a person from the person image region that has undergone high-resolution processing. Therefore, even if a person is depicted in the person image region in a blurry manner that makes it impossible to identify the individual, the person detection score for that person image region will be high. Furthermore, if a person is depicted in the person image region in a clear manner that makes it possible to identify the individual, the person detection score for that person image region will be even higher. Therefore, by setting the second threshold 432 to a higher value than the first threshold 431, the second detection unit 445 does not need to detect a person who is depicted in a blurry manner that makes it impossible to identify the individual, and can detect a person who is depicted in a clear manner that makes it possible to identify the individual.

[0037] The masking unit 446 performs masking on the person image region where the number of pixels detected by the first detection unit 442 is equal to or greater than a predetermined threshold, and on the person image region where a person has been detected by the second detection unit 445. Here, masking refers to image processing performed on the region so that the person cannot be identified, and includes solid fill processing, filtering processing, etc. The masking unit 446 may also perform masking on a part of the person image region (for example, the part corresponding to the face).

[0038] Next, with reference to Figure 7, the person detection method according to Embodiment 2 will be described. First, the acquisition unit 441 acquires the video transmitted from the shooting device 310 (step S201). Next, the first detection unit 442 detects the person image region from the frames constituting the video based on the first threshold 431 of the person detection score (step S202). Next, the measurement unit 443 measures the number of pixels in the person image region (step S203). Next, the image processing unit 444 determines whether the number of pixels in the person image region is less than a predetermined threshold (step S204). In step S204, if the number of pixels in the person image region is greater than or equal to the predetermined threshold (step S204; No), the process proceeds to step S207. In step S204, if the number of pixels in the person image region is less than a predetermined threshold (step S204; Yes), the image processing unit 444 performs a high-resolution processing on the person image region (step S205). Next, the second detection unit 445 detects a person from the person image region that has been re-uploaded in step S205 based on the second threshold 432 of the person detection score (step S206). Next, the masking unit 446 performs masking on the person image region that was determined to have a number of pixels equal to or greater than a predetermined threshold in step S204 and on the person image region where a person was detected in step S206 (step S207).

[0039] Thus, in the person detection device 400 according to this embodiment, the second threshold 432 is set to a higher value than the first threshold 431. As a result, the second detection unit 445 can eliminate areas of the person image where the person is blurred to the extent that the individual cannot be identified, and detect areas of the person image where the person is clearly visible enough to identify the individual. Therefore, by performing a high-resolution processing on areas of the person image with a small number of pixels and then performing person detection again, it becomes possible to perform masking on the areas of the person image where the person is clearly visible enough to identify the individual. Consequently, the person detection device 400 according to this embodiment can accurately detect areas of people that are subject to masking from video.

[0040] Although the above embodiments were described in terms of hardware configuration, the invention is not limited thereto. The processes described in Figures 3 and 6 can also be implemented by having a CPU execute a computer program.

[0041] In the above example, the program can be stored and supplied to the computer using various types of non-transitory computer-readable medium. Non-transitory computer-readable medium includes various types of tangible storage medium. Examples of non-transitory computer-readable medium include magnetic storage media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical storage media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, DVDs (Digital Versatile Discs), and semiconductor memory (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, RAMs (Random Access Memory)). Alternatively, the program may be supplied to the computer by various types of transient computer-readable medium. Examples of transient computer-readable medium include electrical signals, optical signals, and electromagnetic waves. Transitory computer-readable medium can be supplied to the computer via wired communication channels such as electric wires and optical fibers, or via wireless communication channels.

[0042] Furthermore, this disclosure is not limited to the embodiments described above, and can be modified as appropriate without departing from the spirit of the invention. Also, the embodiments of this disclosure may be combined as appropriate. Additionally, the imaging device 310 of each vehicle 300 may be equipped with the functions of the first detection unit 120, measurement unit 130, image processing unit 140, and second detection unit 150 of the person detection device 100. Similarly, the imaging device 310 of each vehicle 300 may be equipped with the functions of the first detection unit 442, measurement unit 443, image processing unit 444, second detection unit 445, and masking unit 446 of the person detection device 400. This allows for individual person detection and masking processing in the imaging device 310 of each vehicle 300.

[0043] Although the present invention has been described above with reference to embodiments, the present invention is not limited to the above embodiments. Various modifications to the structure and details of the present invention can be made, as can be understood by those skilled in the art within the scope of the present invention. [Explanation of symbols]

[0044] 10 frames 30A,30B People 40 vehicles 50 Roadway 100,400 people detection device 410 memory 420 Communications Department 430 Storage section 431 First threshold 432 Second threshold 433 Programs 440 Control Unit 110,441 Acquisition unit (acquisition means) 120,442 First detection unit (first person detection means) 130,443 Measurement unit (measurement means) 140,444 Image processing unit (high-resolution enhancement means) 150,445 Second detection unit (second person detection means) 446 Masking section 200 Person Detection System 300 vehicles 310 Imaging device 311 Photography Department 312 Communications Department 320 Recording device 321 Measuring part 322 Communications Department 500 Networks R1,R2 area

Claims

1. An acquisition means for acquiring images captured by a camera installed in a vehicle, A first person detection means calculates a person detection score using a person detection model for each detection frame area of ​​a predetermined size from the aforementioned video, and detects a person image area which is an area where the calculated person detection score is equal to or greater than a predetermined threshold for the person detection score and is presumed to contain an image of a person. A measuring means for measuring the number of pixels in the aforementioned person image region, A high-resolution means that performs a high-resolution processing on the person image region to increase its resolution when the number of pixels in the person image region is less than a predetermined threshold, which is the lower limit of the number of pixels that can identify the person in the person image region as a specific individual. A second person detection means calculates a person detection score based on the person image region on which the aforementioned high-resolution processing has been performed, and if the calculated person detection score is equal to or greater than a predetermined threshold for person detection scores, it determines that a person is present in the person image region and detects the person. Equipped with, The threshold for the person detection score used by the second person detection means is higher than the threshold for the person detection score used by the first person detection means. Person detection device.

2. A camera installed on a vehicle to capture images of the area around the vehicle, The camera device and the person detection device are equipped with a person detection device that can communicate with the camera device, The aforementioned person detection device Acquisition means for acquiring images captured by the aforementioned camera, A first person detection means calculates a person detection score using a person detection model for each detection frame area of ​​a predetermined size from the aforementioned video, and detects a person image area which is an area where the person detection score is equal to or greater than a predetermined threshold for the person detection score and is presumed to contain an image of a person. A measuring means for measuring the number of pixels in the aforementioned person image region, A high-resolution means that performs a high-resolution processing on the person image region to increase its resolution when the number of pixels in the person image region is less than a predetermined threshold, which is the lower limit of the number of pixels that can identify the person in the person image region as a specific individual. The system includes a second person detection means that calculates a person detection score based on the person image region on which the aforementioned high-resolution processing has been performed, and if the calculated person detection score is equal to or greater than a predetermined threshold for person detection scores, it determines that a person is present in the person image region and detects the person. The threshold for the person detection score used by the second person detection means is higher than the threshold for the person detection score used by the first person detection means. Person detection system.

3. Computers The video footage captured by the camera installed in the vehicle is acquired. From the aforementioned video, a first person detection process is performed to calculate a person detection score using a person detection model for each detection frame area of ​​a predetermined size, and to detect a person image area that is estimated to contain a person's image and whose person detection score is equal to or greater than a predetermined threshold for the person detection score. The number of pixels in the aforementioned human image region is measured, If the number of pixels in the person image region is less than a predetermined threshold, which is the lower limit of the number of pixels that can identify the person in the person image region as a specific individual, a high-resolution processing is performed on the person image region to increase its resolution. Based on the person image region on which the resolution enhancement process described above has been performed, a person detection score is calculated, and if the calculated person detection score is equal to or greater than a predetermined threshold for person detection scores, a second person detection process is performed to determine that a person is present in the person image region and to detect the person. The threshold for the person detection score used in the second person detection process is higher than the threshold for the person detection score used in the first person detection process. Person detection method.

4. On the computer, The process of acquiring video footage captured by a camera installed in a vehicle, A first person detection process calculates a person detection score using a person detection model for each detection frame area of ​​a predetermined size from the aforementioned video, and detects a person image area which is an area where the person detection score is equal to or greater than a predetermined threshold for the person detection score and is presumed to contain an image of a person. A process to measure the number of pixels in the aforementioned person image region, If the number of pixels in the person image region is less than a predetermined threshold, which is the lower limit of the number of pixels that can identify the person in the person image region as a specific individual, a high-resolution processing is performed to increase the resolution of the person image region. A second person detection process calculates a person detection score based on the person image region on which the aforementioned high-resolution processing has been performed, and if the calculated person detection score is equal to or greater than a predetermined threshold for person detection scores, it determines that a person is present in the person image region and detects the person. Make it run, The threshold for the person detection score used in the second person detection process is higher than the threshold for the person detection score used in the first person detection process. Person detection program.

Images