Detection system, detection method and program

Abstract

To provide a detection system, a detection method and a program that can estimate a state of a person to be detected.SOLUTION: A detection system 10 comprises: a foot detection part 1 which detects a position of a foot of a person to be detected; an upper half body detection part 2 which detects a position of at least a part of an upper half body of the person to be detected; and an estimation part 3 which estimates a state of the person to be detected based upon a movement distance of the foot and a movement distance of at least the part of the upper half body calculated using the detected position of the foot and the detected position of at least the part of the upper half body.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a detection system, a detection method, and a program. [Background technology]

[0002] The foot contact position tracking device disclosed in Patent Document 1 detects the position where a person's foot touches the ground by receiving and processing an image captured by a camera. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6892393 Summary of the Invention [Problem to be solved by the invention]

[0004] Such technology for detecting the positions of human body parts using image processing is sometimes used in the field of AI (artificial intelligence), such as machine learning and deep learning, for example, to generate training data.

[0005] The present inventors have discovered the following problems. The state of the person to be detected changes, for example, when the person's feet are hidden by an obstacle or the person to be detected changes his / her posture. In such a foot contact position tracking device, the accuracy of the detection result of the person's foot position may change depending on the state of the person to be detected. Therefore, if the state of the person to be detected can be estimated, the detection accuracy of the person's foot position can be improved. Therefore, there is a demand for the state of the person to be detected to be estimated.

[0006] The present disclosure has been made in consideration of the above-mentioned problems, and provides a detection system, a detection method, and a program capable of estimating the state of a person to be detected. [Means for solving the problem]

[0007] The detection system according to the present disclosure comprises: a foot detection unit for detecting the position of the feet of a person to be detected; an upper body detection unit that detects a position of at least a part of an upper body of the detection subject; and an estimation unit that estimates the state of the person to be detected based on a movement distance of the feet and a movement distance of at least a part of the upper body calculated using the detected position of the feet and the detected position of at least a part of the upper body.

[0008] In addition, in the above-mentioned detection system, when only the distance traveled by the feet exceeds a predetermined value and only the feet move in an upward or downward direction, the estimation unit may estimate that at least a part of the lower body of the person to be detected is hidden by an obstacle.

[0009] In addition, in the above-mentioned detection system, when the distance traveled by the feet and the distance traveled by at least a part of the upper body exceed a predetermined value, and the feet and at least a part of the upper body move in an upward or downward direction, the estimation unit may estimate that the person to be detected is going up or down stairs or that the person to be detected is jumping.

[0010] In addition, in the above-mentioned detection system, when the distance of movement of at least a part of the upper body exceeds a predetermined value, the estimation unit may estimate that the posture of the person to be detected has changed or that at least a part of the upper body of the person to be detected is hidden by an obstacle.

[0011] In addition, the above-mentioned detection system further includes a correction unit, When the estimation unit estimates that at least a part of the lower body of the person to be detected is hidden by an obstacle, the correction unit may correct the detected foot position to a foot position based on the skeleton of the person to be detected.

[0012] The detection method according to the present disclosure is a detection method using a detection system, Detecting a position of a foot of a detection subject; detecting a position of at least a part of an upper body of the detection subject; and estimating a state of the person to be detected based on a distance traveled by the feet and a distance traveled by at least a part of the upper body calculated using the detected position of the feet and the detected position of at least a part of the upper body.

[0013] The program according to the present disclosure is provided to a computer operating in a detection system, Detecting a position of a foot of a detection subject; detecting a position of at least a part of an upper body of the detection subject; and estimating a state of the person to be detected based on a distance traveled by the feet and a distance traveled by at least a part of the upper body calculated using the detected positions of the feet and the detected position of at least a part of the upper body. Effect of the Invention

[0014] According to the present disclosure, the state of a person to be detected can be estimated. [Brief description of the drawings]

[0015] [Figure 1] 1 is a block diagram showing an example of a configuration of a detection system according to an embodiment; [Diagram 2] 4 is a flowchart illustrating an example of a detection method according to an embodiment. [Diagram 3] 11A to 11C are diagrams showing images in an example of a detection method according to an embodiment. [Figure 4] 11A to 11C are diagrams showing images in an example of a detection method according to an embodiment. [Diagram 5] 11A to 11C are diagrams showing images in an example of a detection method according to an embodiment. [Figure 6] 11A to 11C are diagrams showing images in an example of a detection method according to an embodiment. [Figure 7] 11A to 11C are diagrams showing images in an example of a detection method according to an embodiment. [Figure 8]11A to 11C are diagrams showing images in an example of a detection method according to an embodiment. [Figure 9] FIG. 2 is a diagram illustrating an example of a hardware configuration included in the detection system. [Figure 10] 10 is a flowchart showing another example of a detection method according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings. However, the present invention is not limited to the following embodiments. In addition, the following description and drawings are appropriately simplified for clarity of explanation.

[0017] (Embodiment) <Configuration> An example of the configuration of a detection system according to an embodiment will be described with reference to FIG.

[0018] 1, the detection system 10 includes a foot detection unit 1, an upper body detection unit 2, and an estimation unit 3. The detection system 10 may further include a correction unit 4. Each unit of the detection system 10 may be realized by cooperation of one or more programs installed in a computer device constituting the detection system 10 and hardware such as a processor and memory of the computer device.

[0019] The foot detection unit 1 detects the position of the feet of the detection target person from the image. The foot detection unit 1 may detect the positions of the legs, heels, toes, etc. of the detection target person. The upper body detection unit 2 detects the position of at least a part of the upper body of the detection target person from the image. The at least a part of the upper body of the detection target person is, for example, the head, torso, the entire upper body, etc.

[0020] Specifically, foot detection unit 1 and upper body detection unit 2 acquire an image in which the subject is a person to be detected. Foot detection unit 1 and upper body detection unit 2 identify the person to be detected within the acquired image and generate information indicating the area of ​​the person to be detected. This information is, for example, a bounty box. Foot detection unit 1 detects, for example, the midpoint of the lower side of the rectangle indicated by the bounty box as the position of the feet of the person to be detected. Upper body detection unit 2 detects, for example, the midpoint of the upper side of the rectangle indicated by the bounty box as the position of the head of the person to be detected.

[0021] The estimation unit 3 estimates the state of the subject based on the moving distance of the feet and the moving distance of the upper body. Specifically, the estimation unit 3 calculates the moving distance of the feet using the positions of the feet detected by the foot detection unit 1. The estimation unit 3 also calculates the moving distance of the upper body using the position of at least a part of the upper body detected by the upper body detection unit 2.

[0022] The correction unit 4 corrects the foot position detected by the foot detection unit 1 to a foot position based on the skeleton of the detection target. Specifically, the correction unit 4 identifies the positions of the joints of the detection target. Furthermore, the correction unit 4 estimates the foot position of the detection target based on the joint positions. Furthermore, the correction unit 4 corrects the foot position detected by the foot detection unit 1 to the estimated foot position. More specifically, the correction unit 4 identifies the joint positions from the head to the ankle of the detection target, or the joint positions of the thigh and lower leg of the detection target. Furthermore, the correction unit 4 estimates the toe position of the detection target based on the identified joint positions. Furthermore, the correction unit 4 corrects the foot position detected by the foot detection unit 1 to the estimated toe position.

[0023] The detection system 10 may include a generating unit that captures an image that may include the detection target as a subject and generates data representing the captured image. The generating unit is, for example, a camera device. The detection system 10 may also include a storage unit that stores various data including information that associates the data representing the captured image with the time of the capture, etc. The time of the capture may be determined, for example, from the frame rate. The data representing the captured image may be associated with the information representing the area of ​​the detection target, the position of the feet detected by the foot detection unit 1, the position of at least a part of the upper body detected by the upper body detection unit 2, the moving distance of the feet calculated by the estimation unit 3, the moving distance of the upper body calculated by the estimation unit 3, etc.

[0024] <Detection method> Next, an example of a detection method by the detection system 10 according to the embodiment will be described with reference to Fig. 2. Fig. 2 is a flowchart showing an example of the detection method according to the embodiment.

[0025] The positions of the feet and head of the subject in the current frame image and the previous frame image are detected (step ST1).

[0026] Next, the positions of the feet and head of the person to be detected in the current frame image and the previous frame image are compared to calculate the moving distance and moving direction of the feet and head (step ST2).

[0027] Next, it is determined whether or not only the moving distance of the feet exceeds a predetermined value and only the feet move in the vertical direction (step ST3). This predetermined value is, for example, the distance moved by the target person walking during a period t that has elapsed from the time when the current frame image is captured to the time when the previous frame image is captured. The period t may be calculated from the frame rate of the camera device described above. The distance moved by the target person walking during the period t may be calculated by multiplying the walking speed of the target person by the period t. The walking speed of the target person may be obtained in advance from publicly known data.

[0028] If only the foot movement distance exceeds a predetermined value and only the feet move in the vertical direction (step ST3: YES), it is estimated that at least a part of the lower body of the person to be detected is hidden by an obstacle (step ST4). Therefore, the foot position in the current frame image detected in step ST1 is actually the same position as the upper part of the obstacle and is different from the actual foot position. Therefore, the foot position in the current frame image detected in step ST1 is incorrect. Furthermore, the foot position in the current frame image detected in step ST1 is corrected to a foot position based on the skeleton of the person to be detected (step ST5). This correction increases the detection accuracy of the foot position in the current frame image. After performing step ST5, the detection method is terminated.

[0029] If not (step ST3: NO), it is determined whether the moving distance of the feet and the moving distance of the head exceed a predetermined value and whether the feet and the head have moved in the up and down direction (step ST6).

[0030] When the moving distance of the feet and the moving distance of the head exceed a predetermined value and the feet and the head move in the vertical direction (step ST6: YES), it is estimated that the subject is ascending or descending stairs or jumping (step ST7).

[0031] If not (step ST6: NO), it is determined whether or not only the moving distance of the head has exceeded a predetermined value (step ST8).

[0032] If only the head movement distance exceeds a predetermined value (step ST8: YES), it is estimated that the posture of the person to be detected has changed or that at least a part of the upper body of the person to be detected is hidden by an obstacle (step ST9). This detection method is terminated.

[0033] If not (step ST8: NO), this detection method ends. Similarly, for steps ST7, ST9 and after, this detection method ends. Note that if the posture of the person to be detected has changed, or if at least a part of the upper body of the person to be detected is hidden by an obstacle, the position of the feet is assumed to be correct, and therefore correction of the detected foot position is not necessary.

[0034] From the above, the state of the person to be detected can be estimated. Furthermore, depending on the estimation result, it is determined whether or not the detected foot position needs to be corrected, and the detected foot position is corrected as necessary. As a result, the foot position of the person to be detected can be correctly detected. Furthermore, by continuing to detect the foot position of the person to be detected and connecting the points indicating the detected foot positions, the path along which the feet of the person to be detected have moved can be obtained.

[0035] <First specific example of detection method> Next, a first specific example of a detection method using the detection system 10 will be described with reference to FIGS.

[0036] Naturally, the right-handed XYZ coordinate system shown in Figure 3 and other drawings is for the convenience of explaining the positional relationship of the components. Usually, the positive direction of the Z axis is vertically upward, and the XY plane is a horizontal plane, which is common among the drawings.

[0037] In this specific example, as shown in FIG. 3, the detection target P1 walking on the floor surface F1 on which obstacles OB1 and OB2 are placed is imaged from moment to moment, and the detection method by the detection system 10 is applied to the captured images. The obstacle OB1 is a shelf with wheels, and the obstacle OB2 is a chair. The floor surface F1 extends along the XY plane, that is, the horizontal plane. The detection target P1 moves on the floor surface F1 along the path DLL shown in FIG. 4 and FIG. 5. That is, the detection target P1 moves on the floor surface F1 in the positive direction of the Y axis, and then moves in the positive direction of the X axis. As shown in FIG. 3, the skeletal model SK1 is estimated and created from the skeleton of the detection target P1. The skeletal model SK1 is superimposed on the part of the image showing the detection target P1. FIG. 5 shows the current frame image FR2. FIG. 4 shows the previous frame image FR1 captured immediately before the current frame image FR2 shown in FIG. 5.

[0038] In step ST1, a bounty box BB1, which is an area indicating the subject P1, is generated in the previous frame image FR1 shown in FIG. 4. The foot detection unit 1 detects the midpoint of the lower side of the rectangle indicated by the bounty box BB1 as the position DL1 of the foot PL1 of the subject P1. The upper body detection unit 2 detects the midpoint of the upper side of the rectangle indicated by the bounty box BB1 as the position DH1 of the head PH1 of the subject P1. This detected position DL1 is almost the same as the actual position of the foot PL1, and this detected position DH1 is almost the same as the actual position of the head PH1. Therefore, the positions of the foot PL1 and head PH1 of the subject P1 can be detected with good accuracy.

[0039] Similarly, in the current frame image FR2 shown in FIG. 5, a bounty box BB2, which is an area indicating the target person P1, is generated. The foot detection unit 1 detects the midpoint of the lower side of the rectangle indicated by the bounty box BB2 as the position DL2 of the foot PL1 of the target person P1. The upper body detection unit 2 detects the midpoint of the upper side of the rectangle indicated by the bounty box BB2 as the position DH2 of the head PH1 of the target person P1. This detected position DH2 is almost the same as the actual position of the head PH1. On the other hand, this detected position DL2 is separated from the actual position of the foot PL1. Therefore, although the position of the head PH1 of the target person P1 can be detected with good accuracy, the position of the foot PL1 cannot be detected with good accuracy.

[0040] In step ST2, the position DL1 is compared with the position DL2 to determine the movement distance and movement direction of the foot PL1. The distance between the position DL1 and the position DL2 is the movement distance of the foot PL1. The movement direction of the foot PL1 is a diagonal direction extending in the positive direction of the X-axis and the positive direction of the Z-axis, and includes the up-down direction and the horizontal direction.

[0041] Similarly, the position DH1 is compared with the position DH2 to determine the movement distance and movement direction of the head PH1. The distance between the position DH1 and the position DH2 is the movement distance of the head PH1. The movement direction of the head PH1 is the horizontal direction extending approximately in the positive direction of the X-axis.

[0042] In step ST3, it is determined whether only the movement distance of the foot PL1 exceeds a predetermined value and only the foot PL1 moves in the vertical direction. Also, the movement distance of the foot PL1 is the distance between the position DL1 and the position DL2, and exceeds the predetermined value, while the movement distance of the head PH1 is the distance between the position DH1 and the position DH2, and does not exceed the predetermined value. Therefore, only the movement distance of the foot PL1 exceeds the predetermined value. As described above, while the foot PL1 also moves in the vertical direction, the head PH1 moves in the horizontal direction, so only the foot PL1 moves in the vertical direction. Therefore, it is determined that only the movement distance of the foot PL1 exceeds the predetermined value and only the foot PL1 moves in the vertical direction (step ST3: YES).

[0043] In step ST4, it is estimated that the foot PL1 is hidden by an obstacle. Note that the actual foot PL1 is hidden by obstacles OB1 and OB2.

[0044] In step ST5, the position DL2 of the foot PL1 in the current frame image FR2 detected in step ST1 is corrected to the position DL21 of the foot PL1 based on the skeleton of the detection target person P1. Specifically, a skeleton model SK1 is estimated from the skeleton of the detection target person P1 from the head to above the knees contained in the bounty box BB1. The skeleton model SK1 is superimposed on the portion indicating the detection target person P1 in the current frame image FR2. The midpoint of both feet of the skeleton model SK1 is corrected to the position DL21 of the foot PL1 of the detection target person P1. As a result, the position DL2 is corrected to the position DL21. The position DL21 is almost the same as the actual position of the foot PL1. Therefore, the position of the foot PL1 of the detection target person P1 can be corrected and detected with good accuracy.

[0045] <Second specific example of detection method> Next, a second specific example of the detection method using the detection system 10 will be described with reference to FIG.

[0046] In this specific example, as shown in Fig. 7, an obstacle OB3 and a detection target P2 whose posture changes are continuously imaged, and the detection method by the detection system 10 is applied to the captured images. The obstacle OB3 is a shelf. Only a part of the upper part of the legs of the detection target P2 is hidden by the obstacle OB3. Fig. 7 shows a previous frame image FR11 and a current frame image FR12. The current frame image FR12 was captured when a period t had elapsed since the previous frame image FR11 was captured.

[0047] In step ST1, a bounty box BB11, which is an area indicating the person to be detected P2, is generated in the previous frame image FR11. The foot detection unit 1 detects the midpoint of the lower side of the rectangle indicated by the bounty box BB11 as the position DL11 of the foot PL2 of the person to be detected P2. The upper body detection unit 2 detects the midpoint of the upper side of the rectangle indicated by the bounty box BB11 as the position DH11 of the head PH2 of the person to be detected P2. This detected position DL11 is almost the same as the actual position of the foot PL2, and this detected position DH11 is almost the same as the actual position of the head PH2. Therefore, the positions of the feet PL2 and the head PH2 of the person to be detected P2 can be detected with good accuracy.

[0048] Similarly, in the current frame image FR12, a bounty box BB12, which is an area indicating the person to be detected P2, is generated. The foot detection unit 1 detects the midpoint of the lower side of the rectangle indicated by the bounty box BB12 as the position DL12 of the foot PL2 of the person to be detected P2. The upper body detection unit 2 detects the midpoint of the upper side of the rectangle indicated by the bounty box BB2 as the position DH12 of the head PH2 of the person to be detected P2. This detected position DH12 is almost the same as the actual position of the head PH2, and this detected position DL12 is almost the same as the actual position of the foot PL2. Therefore, the positions of the feet PL2 and head PH1 of the person to be detected P2 can be detected with good accuracy.

[0049] In step ST2, the position DL11 is compared with the position DL12 to determine the movement distance and movement direction of the foot PL2. The distance between the position DL11 and the position DL12 is the movement distance of the foot PL2, which is approximately zero. Therefore, the movement distance of the foot PL2 is approximately zero, and the movement direction of the foot PL2 cannot be identified.

[0050] Similarly, the position DH11 and the position DH12 are compared to determine the movement distance and movement direction of the head PH2. The distance between the position DH11 and the position DH12 is the movement distance of the head PH2, and has a predetermined length. The movement distance of the head PH2 exceeds a predetermined value. The movement direction of the head PH2 is the up-down direction extending approximately in the positive direction of the Z axis.

[0051] In step ST3, it is determined whether only the movement distance of the foot PL2 exceeds a predetermined value and only the foot PL2 has moved in the vertical direction. As described above, the movement distance of the foot PL2 is approximately zero, and therefore does not exceed the predetermined value. The movement direction of the foot PL2 cannot be identified, and the movement direction of the head PH2 is vertical. Therefore, only the foot PL2 has not moved in the vertical direction. Therefore, it is determined that only the movement distance of the foot PL2 has not exceeded the predetermined value and only the foot PL2 has not moved in the vertical direction (step ST3: NO).

[0052] In step ST6, it is determined whether the movement distance of the foot PL2 and the movement distance of the head PH2 exceed a predetermined value and whether the foot PL2 and the head PH2 have moved in the vertical direction. As described above, the movement distance of the foot PL2 is approximately zero, and therefore does not exceed the predetermined value. Furthermore, the movement direction of the foot PL2 cannot be specified. It is determined that the movement distance of the foot PL2 and the movement distance of the head PH2 exceed the predetermined value and that the foot PL2 and the head PH2 have not moved in the vertical direction (step ST6: NO).

[0053] In step ST8, it is determined whether only the moving distance of the head PH2 has exceeded the predetermined value. As described above, while the moving distance of the feet PL2 has not exceeded the predetermined value, the moving distance of the head PH2 has exceeded the predetermined value. Therefore, it is determined that only the moving distance of the head PH2 has exceeded the predetermined value (step ST8: YES).

[0054] In step ST9, it is estimated that the posture of the person to be detected P2 has changed or that the head PH2 of the person to be detected P2 is hidden by an obstacle. As shown in FIG. 7, in reality, only a part of the upper part of the legs of the person to be detected P2 is hidden by an obstacle OB3, and the head PH2 is not hidden by the obstacle, while the posture of the person to be detected P2 has changed. Therefore, the estimation result described above has a certain degree of validity, and the estimation accuracy is good. As described in step ST1, since the positions of the feet PL2 and the head PH2 of the person to be detected P2 can be detected with good accuracy, correction of the detected foot positions is not necessary.

[0055] <Third specific example of detection method> Next, a third specific example of the detection method using the detection system 10 will be described with reference to FIG.

[0056] In this specific example, as shown in Fig. 8, a person to be detected P3 who repeatedly sits down and stands up from an obstacle OB4, which is a chair, is imaged from moment to moment, and the detection method by the detection system 10 is applied to the multiple images captured. The legs of the person to be detected P3 are hidden by the obstacle OB4.

[0057] In this specific example, similarly to the second specific example of the detection method using the detection system 10 described above, the process proceeds in the order of step ST1, step ST2, step ST3, step ST6, step ST8, and step ST9.

[0058] In step ST1, a bounty box BB2, which is an area indicating the target person P3, is generated in the current frame image FR22. The foot detection unit 1 detects the midpoint of the lower side of the rectangle indicated by the bounty box BB2 as the position DL22 of the foot PL3 ​​of the target person P3. The foot PL3 ​​of the target person P3 is hidden by an obstacle OB4, but is almost at the same position as the position DL22. The upper body detection unit 2 detects the midpoint of the upper side of the rectangle indicated by the bounty box BB2 as the position DH22 of the head PH3 of the target person P3. This detected position DH22 is almost the same as the actual position of the head PH3. This detected position DL22 is almost the same as the actual position of the foot PL3. Therefore, the positions of the foot PL3 ​​and head PH1 of the target person P3 can be detected with good accuracy.

[0059] In step ST9, it is estimated that the posture of the person to be detected P3 has changed or that the head PH3 of the person to be detected P3 is hidden by an obstacle. As shown in FIG. 8, the legs of the person to be detected P3 are actually hidden by an obstacle OB4, and the head PH3 is not hidden by the obstacle. On the other hand, the posture of the person to be detected P3 changes because the person to be detected P3 repeatedly sits and stands. Therefore, the estimation result described above has a certain degree of validity, and the estimation accuracy is good. As described in step ST1, the positions of the feet PL3 ​​and the head PH3 of the person to be detected P3 can be detected with good accuracy, so there is no need to correct the detected foot positions.

[0060] (Other embodiments, etc.) The detection system according to the above embodiment may have the following hardware configuration. Fig. 9 is a diagram showing an example of a hardware configuration included in the detection system. As described above in the various embodiments, the steps of the detection method in the detection system have been described, the present disclosure may also take the form of a detection method.

[0061] 9 includes a processor 301 and a memory 302, as well as an interface 303. Each component of the detection system 10 described in the above embodiment (see FIG. 1) is realized by the processor 301 reading and executing a program stored in the memory 302. In other words, this program is a control program for causing the processor 301 to function as the detection system 10 or a part thereof.

[0062] The above-mentioned program includes a set of instructions (or software code) for making the computer perform one or more functions described in the embodiment when the program is loaded into the computer. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, the computer-readable medium or the 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 disc (DVD), Blu-ray® disk or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may be transmitted on a transitory computer-readable medium or a communication medium. By way of example and not limitation, the transitory computer-readable medium or the communication medium includes electrical, optical, acoustic, or other forms of propagating signals.

[0063] Moreover, the above-mentioned program can be said to be a control program for causing the detection system 10 to execute such a detection method.

[0064] The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit and scope of the present invention. In addition, the present invention may be implemented by appropriately combining the above-described embodiment and examples thereof.

[0065] In the example of the detection method according to the present embodiment, the position of the head of the detection subject is detected, but the position of at least a part of the upper body, for example the torso or the entire upper body, may be detected.

[0066] In addition, in an example of the detection method by the detection system 10 according to the embodiment, steps ST1 to ST9 are performed, but as shown in FIG. 10, steps ST1, ST2, ST31, ST32, ST33, ST4, ST5, ST7, and ST9 may be performed. In step ST31, it is determined whether at least one of the movement distance of the feet and the movement distance of the head exceeds a predetermined value. In step ST32, it is determined whether the movement distance of the feet exceeds a predetermined value and the feet have moved in the vertical direction. In step ST33, it is determined whether the head has moved the same amount in the same direction as the feet. In other words, it is determined whether the head has moved in the vertical direction and the movement distance of the head is approximately the same as the movement distance of the feet. [Explanation of symbols]

[0067] 10,300 Detection System 1 Foot detection unit 2 Upper body detection section 3 Estimation part 4. Correction section 301 Processor 302 Memory 303 Interface BB1, BB2, BB3, BB11, BB12 Bounty Box P1, P2, P3 detection targets PH1, PH2, PH3 head PL1, PL2, PL3 Foot BB1, BB2, BB3 Bounty Box DH1, DH2, DH11, DH12, DH22 (detected head) position DL1, DL2, DL11, DL12, DL22 (detected foot) positions OB1, OB2, OB3, OB4 Obstacles F1 Floor FR1, FR11 Previous frame image FR2, FR12, FR22 Current frame image SK1 skeleton model ST1, ST2, ST3, ST31, ST32, ST33, ST4, ST5, ST6, ST7, ST8, ST9 steps t period

Claims

1. a foot detection unit for detecting the position of the feet of a person to be detected; an upper body detection unit that detects a position of at least a part of an upper body of the detection subject; and an estimation unit that estimates a state of the detection subject based on a moving distance of the feet and a moving distance of at least a part of the upper body that are calculated using the detected positions of the feet and the detected position of at least a part of the upper body. Detection system.

2. When only the moving distance of the feet exceeds a predetermined value and only the feet move in an up-down direction, the estimation unit estimates that at least a part of the lower body of the detection subject is hidden by an obstacle. The detection system of claim 1 .

3. When a moving distance of the feet and a moving distance of at least a part of the upper body exceed a predetermined value and the feet and at least a part of the upper body move in an up-down direction, the estimation unit estimates that the detection subject is ascending or descending stairs or jumping.

3. A detection system according to claim 1 or 2.

4. When only a moving distance of at least a part of the upper body exceeds a predetermined value, the estimation unit estimates that a posture of the person to be detected has changed or that at least a part of the upper body of the person to be detected is hidden by an obstacle.

3. A detection system according to claim 1 or 2.

5. A correction unit is further provided, When the estimation unit estimates that at least a part of the lower body of the detection target is hidden by an obstacle, the correction unit corrects the detected foot positions to foot positions based on a skeleton of the detection target. The detection system of claim 2 .

6. A method of detection by a detection system, comprising: Detecting a position of a foot of a detection subject; detecting a position of at least a part of an upper body of the detection subject; and estimating a state of the detection subject based on a moving distance of the feet and a moving distance of at least a part of the upper body calculated using the detected positions of the feet and the detected position of at least a part of the upper body. Detection methods.

7. A computer operating in the detection system includes: Detecting a position of a foot of a detection subject; detecting a position of at least a part of an upper body of the detection subject; and estimating a state of the detection subject based on a moving distance of the feet and a moving distance of at least a part of the upper body calculated using the detected positions of the feet and the detected position of at least a part of the upper body. program.

Images