CN-121999426-A - Information processing system, information processing method, and program product

Abstract

An information processing system includes a processor that continuously captures an image of a person including a monitoring object, performs skeleton estimation of the person in the captured image to detect presence or absence of a posture abnormality of the person, and when the posture abnormality is detected, acquires text information by inputting the image in which the posture abnormality of the person is detected into a large-scale language model that converts the content of the input image into text information and outputs the text information in association with the state of the person in the input image, and notifies a preset notification address of the presence or absence of the posture abnormality of the person of the monitoring object together with the acquired text information and the image in which the posture abnormality is detected.

Inventors

• KAWAKAMI SHOHEI

Assignees

• 富士胶片商业创新有限公司

Dates

Publication Date

20260508

Application Date

20250513

Priority Date

20241105

Claims (7)

1. 1. An information processing system, characterized in that, The device is provided with a processor and is provided with a control unit, The processor Continuously taking images of persons including the monitoring object, To detect whether or not a person is abnormal in posture by performing skeleton estimation of the person in a captured image, In the case where the posture abnormality is detected, text information is acquired by inputting an image in which the posture abnormality of the person is detected into a large-scale language model that converts the content of the input image into text information and outputs the text information, the text information being related to the state of the person in the input image, And notifying the notification address set in advance that the person to be monitored is in abnormal posture together with the acquired text information and the image in which the abnormal posture is detected.
2. 2. The information handling system of claim 1, wherein, The posture abnormality detected by the skeleton estimation is a fall of the person to be monitored.
3. 3. The information handling system of claim 2, wherein, The processor acquires text information including information on the state of the person of the monitoring object before and after the fall by inputting an image of the person of the monitoring object detected to fall and images before and after the fall into the large-scale language model.
4. 4. The information handling system of claim 1, wherein, The processor When the posture abnormality of the person is detected by skeleton estimation, it is determined whether the detected posture abnormality is a fall of the person to be monitored, In the case where it is determined that the detected posture abnormality is not a fall of the person to be monitored, text information relating to the state of the person in the input image is acquired by inputting the image in which the posture abnormality is detected into the large-scale language model, And notifying the notification address set in advance that the person to be monitored is in abnormal posture together with the acquired text information and the image in which the abnormal posture is detected.
5. 5. The information handling system of claim 4, wherein, The processor In the case where it is determined that the detected posture abnormality is a fall of a person to be monitored, text information including information on the state of the person to be monitored before and after the fall is acquired by inputting an image of the fall of the person detected and images before and after the fall into the large-scale language model, And notifying the person of the monitored object that falls to a preset notification address together with the acquired text information and the image for detecting the falling.
6. 6. A program product comprising a program for causing a computer to execute a process, characterized in that the process has the steps of: continuously capturing images of a person including a monitoring object; performing skeleton estimation of a person in a captured image to detect whether or not the person has a posture abnormality; in the case where the posture abnormality is detected, text information is acquired by inputting an image in which the posture abnormality of the person is detected into a large-scale language model that converts the content of the input image into text information and outputs the text information, the text information being related to the state of the person in the input image, and And notifying the notification address set in advance that the person to be monitored is in abnormal posture together with the acquired text information and the image in which the abnormal posture is detected.
7. 7. An information processing method characterized by comprising the steps of: continuously capturing images of a person including a monitoring object; performing skeleton estimation of a person in a captured image to detect whether or not the person has a posture abnormality; in the case where the posture abnormality is detected, text information is acquired by inputting an image in which the posture abnormality of the person is detected into a large-scale language model that converts the content of the input image into text information and outputs the text information, the text information being related to the state of the person in the input image, and And notifying the notification address set in advance that the person to be monitored is in abnormal posture together with the acquired text information and the image in which the abnormal posture is detected.

Description

Information processing system, information processing method, and program product Technical Field The present invention relates to an information processing system, an information processing method, and a program product. Background International publication No. 2022/249635 discloses an action detection system that notifies a predetermined notification address of the detection of an action when a first notification condition set for each action type specified based on the action of a person included in an image and a second notification condition such as the duration of the action, the degree of congestion around the person, the time period for capturing the image, and the like are all satisfied. Disclosure of Invention In recent years, AI (ARTIFICIAL INTELLIGENCE ) technology has been used in various fields due to the development of the AI technology. For example, a system is proposed in which a person of a monitoring object is photographed with a camera, and AI technology is applied to the photographed image, whereby a fall of the person is detected. The aim is to quickly carry out rescue activities by quickly noticing a fall and giving a warning at the site of medical treatment, nursing welfare. In addition, in a place where dangerous work is performed such as a factory, the machine may be used to stop the machine when dangerous operation such as a fall is detected. However, when a person to be monitored is photographed and the photographed person is detected to fall, even if the person is notified to a preset notification address, the person who receives the notification must confirm the photographed image, and it may not be easy to grasp what state the person to be monitored is. The present invention aims to provide an information processing system, an information processing method, and a program product, which can easily grasp the state of a detected posture abnormality, compared with the case where the posture abnormality of a person to be monitored is detected from an image obtained by capturing only the person and notified. According to a first aspect of the present invention, there is provided an information processing system including a processor that continuously captures an image of a person including a monitoring object, performs skeleton estimation of the person in the captured image to detect presence or absence of a posture abnormality of the person, and when the posture abnormality is detected, acquires text information by inputting the image in which the posture abnormality of the person is detected into a large-scale language model that converts the content of the input image into text information and outputs the text information in association with the state of the person in the input image, and notifies a preset notification address of the fact that the posture abnormality of the person of the monitoring object has occurred together with the acquired text information and the image in which the posture abnormality has been detected. According to a second aspect of the present invention, in the information processing system according to the first aspect, the posture abnormality detected by skeleton estimation is a fall of a person to be monitored. According to a third aspect of the present invention, in the information processing system relating to the second aspect, the processor acquires text information including information on the state of the person of the monitoring object before and after the fall by inputting an image of the person of the monitoring object detected to fall and images before and after the fall into the large-scale language model. According to a fourth aspect of the present invention, in the information processing system according to the first aspect, the processor determines whether the detected posture abnormality is a fall of the person to be monitored when the posture abnormality is detected by skeleton estimation, and acquires text information by inputting an image in which the posture abnormality is detected into the large-scale language model, the text information being related to a state of the person in the input image, and notifies a preset notification address of the fact that the posture abnormality has occurred in the person to be monitored together with the acquired text information and the image in which the posture abnormality has been detected when it is determined that the detected posture abnormality is not a fall of the person to be monitored. According to a fifth aspect of the present invention, in the information processing system according to the fourth aspect, the processor acquires text information including information on the state of the person to be monitored before and after the fall by inputting an image of the person to be monitored before and after the fall and an image of the person to be monitored before and after the fall into the large-scale language model when it is determined that the detected posture abnormality is the fall of