Image analysis system

Abstract

To analyze a captured image at a work site and provide notifications for multiple purposes.SOLUTION: An image analysis system 100 comprises: a recognition unit that uses a trained model that recognizes objects from images to detect and recognize objects appearing in a captured image 50 of a work site 50; an acquisition unit that acquires additional information associated with the objects and positional information of the objects in a captured image 150; and an execution unit that executes notifications according to the acquired additional information and the positional information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an image analysis system. [Background technology]

[0002] Patent Document 1 discloses technology relating to an intrusion monitoring device and an intrusion monitoring system that monitors intrusions into restricted areas. In this prior art, the intrusion monitoring system includes a monitoring sensor that detects workers entering restricted areas, a camera that takes an image when the monitoring sensor detects a worker and sends the captured image to the intrusion monitoring device, an intrusion monitoring device that determines intrusion into restricted areas using a color code for the worker and a color code for a traffic cone included in the captured image, and a warning light for a worker who has entered a restricted area.

[0003] Patent Document 2 discloses technologies related to a surveillance system, an information processing device, an information processing method, and a program. This prior art includes an imaging device that captures an image including a first endpoint marker that is assigned first endpoint identification information indicating a first endpoint of a boundary line that separates a predetermined no-entry area and is placed at the first endpoint, a second endpoint marker that is assigned second endpoint identification information indicating a second endpoint of the boundary line and is placed at the second endpoint, an image analysis unit that detects the first endpoint marker and the second endpoint marker from the captured image and identifies the positions of the first endpoint marker and the second endpoint marker in the captured image, a setting unit that sets the boundary line in the captured image based on the positions of the first endpoint marker and the second endpoint marker, and a time series analysis unit that detects moving objects passing through the boundary line in the captured image. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2017-04184 [Patent Document 2] Japanese Patent Application Publication No. 2019-16836 Summary of the Invention [Problem to be solved by the invention]

[0005] For example, a system that performs image analysis on captured images to detect pre-set color codes or identification information, sets boundaries, and notifies of worker entry has limited uses because it cannot provide any other notifications other than notifying of worker entry.

[0006] In view of the above, an object of the present invention is to perform image analysis on images taken at a work site and provide notifications for multiple purposes. [Means for solving the problem]

[0007] The first aspect is an image analysis system that includes an identification unit that detects and identifies objects that appear in images taken at a work site using a trained model that identifies objects from images, an acquisition unit that acquires additional information linked to the object and location information of the object in the captured image, and an execution unit that executes a notification based on the acquired additional information and location information.

[0008] In the image analysis system of the first aspect, by detecting and identifying objects in a photographed image using a trained model, the system acquires additional information about the object and the position information of the object in the photographed image, and executes notification according to the additional information and the position information. In this way, the system acquires and notifies the additional information associated with the object, so notifications for multiple purposes can be issued.

[0009] A second aspect is the image analysis system according to the first aspect, wherein the identification unit detects and identifies the shape of the object and at least one of the color and pattern of the object.

[0010] In the second aspect of the image analysis system, the identification unit detects and identifies the shape of the object and at least one of the object's color and pattern, and is therefore less susceptible to the influence of the resolution of the captured image, the shooting angle, etc., compared to when reading an identification code attached to the object.

[0011] A third aspect is an image analysis system according to the first or second aspect, in which the additional information includes any of information on left-behind items, information on occupancy rates, information on the removal of monitored items, count information on the number of workers passing through, work start information, and work end information.

[0012] In the third aspect of the image analysis system, the system can simply detect and recognize objects without the user having to give instructions each time using a keyboard or the like, and can provide information on leftover items, occupancy rates, items being removed from surveillance, a count of the number of workers passing through, and work start and end information. [Effects of the Invention]

[0013] According to the present invention, images taken at a work site can be analyzed and notifications for multiple purposes can be provided. [Brief explanation of the drawings]

[0014] [Figure 1] This is an image taken at a work site. [Figure 2] FIG. [Figure 3] FIG. 2 is a front view of a triangular cone with an identifier placed on the top. [Figure 4] FIG. 10 is a perspective view of a triangular cone covered with an identifier placed in an aisle. [Figure 5] FIG. 10 is a perspective view of a triangular cone covered with an identifier placed in a safety area. [Figure 6] FIG. 10 is a perspective view of a triangular cone covered with an identifier placed in a material area. [Figure 7] This is a perspective view of a triangular cone covered with an identifier placed next to a fire extinguisher. [Figure 8] FIG. 1 is a block diagram of a communication system. [Figure 9] FIG. 1 is a block diagram of a hardware configuration of an image analysis system. [Figure 10] 1 is a block diagram of the functional configuration of an image analysis system. [Figure 11] 10 is a flowchart of the flow of image analysis processing of the image analysis system. DETAILED DESCRIPTION OF THE INVENTION

[0015] <Embodiment> An example of an embodiment of the technology of the present disclosure will be described below with reference to the drawings. Note that components and processes that perform the same operations, actions, and functions are given the same reference numerals throughout the drawings, and duplicated descriptions may be omitted or simplified as appropriate. Each drawing is merely a schematic illustration to allow a sufficient understanding of the technology of the present disclosure. Therefore, the technology of the present disclosure is not limited to the illustrated examples. Furthermore, in this embodiment, descriptions of configurations not directly related to the present invention and well-known configurations may be omitted or simplified.

[0016] <Notification system> First, a notification system will be described that monitors a work site 50 (see FIG. 1) such as a construction site using an image analysis system according to one embodiment of the present invention and notifies a manager 20 (see FIG. 1) of various information.

[0017] As shown in Fig. 8, the notification system 10 of this embodiment is configured to include an image analysis system 100 (see also Figs. 9 and 10), a photographing device 12, and a mobile terminal device 14. The image analysis system 100, the photographing device 12, and the mobile terminal device 14 are connected via a network such as Ethernet (registered trademark), FDDI, or Wi-Fi (registered trademark). The image analysis system 100 is installed on a standalone server or a cloud server. Although only one mobile terminal device 14 is shown in Fig. 1, in reality, there are provided as many mobile terminal devices as there are managers 20 and workers 22 (described later) who carry the mobile terminal devices 14. Similarly, although only one imaging device 12 is shown in Fig. 1, multiple imaging devices 12 may be provided.

[0018] The image capturing device 12 is a digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor. The image capturing device 12 can capture video of the entire work site 50 (see FIG. 1), which will be described later.

[0019] The mobile terminal device 14 is a portable information device that can be carried and used while being held in the hand or worn on the body, such as a laptop computer, a smartphone, a tablet terminal, etc. In this embodiment, the mobile terminal device 14 is carried by the manager 20 working at the work site 50.

[0020] <Image analysis system> Next, the image analysis system of this embodiment will be described.

[0021] The image analysis system 100 shown in Fig. 8 detects and recognizes objects that appear in a photographed image 150 (see Fig. 1) of a work site 50 photographed by an image capturing device 12, and acquires additional information associated with the object and location information of the object in the photographed image 150. Then, a notification according to the acquired additional information and location information is sent to the mobile terminal device 14. Note that hereinafter, "detecting and recognizing" may be referred to as "detection and recognition."

[0022] Here, "recognition" means replacing a detected object or group of objects with an identifiable code (hereinafter referred to as an "object identifier").

[0023] The hardware configuration of the image analysis system 100 will be described.

[0024] 9, the image analysis system 100 includes a CPU (Central Processing Unit) 121, a ROM (Read Only Memory) 122, a RAM (Random Access Memory) 123, a storage 124, a communication interface 125, an input unit 126, and a display unit 127. Each component is connected to each other via a bus 129 so as to be able to communicate with each other.

[0025] The CPU 121 is a central processing unit that executes various programs and controls each component. That is, the CPU 121 reads a program from the ROM 122 or the storage 124, and executes the program using the RAM 123 as a work area. The CPU 121 controls the above components and performs various arithmetic processing in accordance with the program recorded in the ROM 122 or the storage 124. In this embodiment, the ROM 122 or the storage 124 stores a captured image 150 (see FIG. 1) captured by the imaging device 12 (see FIG. 8), a program for image analysis of the captured image 150, and the like.

[0026] The ROM 122 stores various programs and various data. The RAM 123 temporarily stores programs or data as a working area. The storage 124 is composed of an HDD (Hard Disk Drive), an SSD (Solid State Drive), etc., and stores various programs including the operating system and various data.

[0027] The communication interface 125 is an interface for communicating with other devices such as the image capturing device 12 and the mobile terminal device 14 (see FIG. 8), and uses standards such as Ethernet (registered trademark), FDDI, and Wi-Fi (registered trademark).

[0028] The input unit 126 includes a pointing device such as a mouse, a keyboard, etc., and is used to perform various inputs. The display unit 127 is, for example, a liquid crystal display, and displays various types of information. The display unit 127 may employ a touch panel system and function as the input unit 126.

[0029] When executing various programs, the image analysis system 100 uses these hardware resources to realize various functions.

[0030] Next, the functional configuration of the image analysis system 100 will be described.

[0031] 10, the image analysis system 100 has, as functional components, a recognition unit 102, an acquisition unit 104, and an execution unit 106. Each functional component is realized by a CPU 121 reading and executing a program stored in a ROM 122 or a storage 124 (see FIG. 9).

[0032] The recognition unit 102 uses a trained model that recognizes objects from images to detect and recognize objects appearing in a captured image 150 (see FIG. 1) of the work site 50 captured by the image capture device 12 (see FIG. 8). The acquisition unit 104 acquires additional information associated with the object and location information of the object in the captured image. The execution unit 106 sends a notification corresponding to the acquired additional information and location information to the mobile terminal device 14 (see FIG. 8).

[0033] The trained model for detecting and recognizing objects in captured images 150 (see Figure 1) is a model that has been trained on a known machine learning model such as R-CNN (Regions with Convolutional Neural Networks) using separately prepared training data (multiple captured images of known objects and known object identifiers corresponding to those objects).By performing a large number of training sessions, when a newly captured image of a specific object (not included in the training data) is input, it becomes possible to output an object identifier corresponding to that object with a high probability.

[0034] Furthermore, the position information of an object in the captured image 150 (see FIG. 1) is information for estimating the positional relationship of the object in the captured image 150, and may be obtained by any method. For example, a method of converting a distance image obtained by a ToF (Time of Flight) camera or a stereo camera into coordinates, or a method of obtaining a distance image using machine learning based on an image captured by a monocular camera and converting it into coordinates, may be used. Note that an example of a method using machine learning is a method of obtaining a distance image by analyzing shape differences in small image regions caused by lens aberration in the image capture device. Furthermore, the position of a detected object in the captured image may be used as the position information.

[0035] Additionally, the additional information linked to the object identifier is stored in advance in storage 124 (see FIG. 9). Note that it is not necessary for all objects to be linked to object identifiers, and it is not necessary for all object identifiers to be linked to additional information.

[0036] An object is a so-called "physical object," and in this embodiment, refers to an object detected and recognized from the captured image 150. Also, in this embodiment, the additional information refers to information related to an information acquisition method, a notification method, and the like, associated with the detected and recognized object.

[0037] In the embodiment described below, the objects are the manager 20, worker 22, fire extinguisher 34, cart 30, striped box 300, traffic cone 200, and identifiers 211, 212, 213, and 214 (see Figure 1) placed over the traffic cone 200 that appear in the captured image 150.

[0038] Next, an example of the flow of image analysis processing in image analysis system 100 (see FIG. 8) will be described. As described above, the image analysis processing is performed by CPU 121 reading a program from ROM 122 or storage 124 (see FIG. 9), expanding it in RAM 123 (see FIG. 9), and executing it.

[0039] As shown in FIG. 11, in step S102, the CPU 121 (see FIG. 9) detects and recognizes an object from the captured image 150 (see FIG. 1) using the trained model. In step S104, the CPU 121 acquires, from the storage 124 (see FIG. 9), additional information associated with the detected and recognized object, and also acquires position information of the object in the captured image 150. In step S106, the CPU 212 sends a notification according to the acquired additional information and position information to the mobile terminal device 14 (see FIG. 8).

[0040] [Example] Next, a description will be given of a specific example of image analysis by the image analysis system 100. Note that the example is merely an example, and the present invention is not limited to this example.

[0041] In addition, in this embodiment, for ease of explanation, the objects to be detected and recognized in the photographed image 150 of the work site 50 in Figure 1 are the manager 20, the worker 22, the fire extinguisher 34, the cart 30, the traffic cone 200, and the identifiers 211, 212, 213, and 214 placed over the traffic cone 200, but are not limited to these.

[0042] 8 performs image analysis on an image 150 of the work site 50 shown in Fig. 1 taken by the image capturing device 12, and notifies the mobile terminal device 14 (see Fig. 8) carried by the manager 20 of the number of managers 20 and workers 22 passing through the passage 52 of the work site 50, the entry of a cart 30 as an example of an object left behind in the safety area 54, the occupancy rate of materials 32, 33 in the material area 56, and the removal of a fire extinguisher 34 as an example of an object to be monitored. The image analyzing system 100 also notifies the mobile terminal devices 14 (see Fig. 8) carried by the manager 20 of information on the start of work at the work site 50.

[0043] As shown in Fig. 2, the identifiers 211, 212, 213, and 214 are made of paper and have a triangular pyramid shape, and are placed on the top 200A of the triangular cone 200 as shown in Fig. 3. Note that Fig. 3 only shows an example in which the identifier 211 is placed on top. The colors of the identifiers 211, 212, 213, and 214 to be trained as shown in Fig. 1 are green, blue, white, and yellow. The color of the triangular cone 200 is red. Note that the horizontally striped identifier 211 shown in Fig. 1 represents green, the checkered identifier 212 represents blue, the vertically striped identifier 213 represents white, and the polka-dot identifier 214 represents yellow.

[0044] As shown in FIG. 1, two triangular cones 200 capped with green identifiers 211 are installed on both sides of a passage 52 in a work area 50 at a construction site (see FIG. 4). Four triangular cones 200 capped with blue identifiers 212 are installed to surround a safety area 54 at the work area 50 (see FIG. 5). Four triangular cones 200 capped with white identifiers 212 are installed to surround a material area 56 at the work area 50 (see FIG. 6). A triangular cone 200 capped with a yellow identifier 212 is installed next to a fire extinguisher 34 at the work area 50 (see FIG. 7). In this embodiment, the manager 20 caps the tops 200A of the triangular cones 200 with identifiers 211, 212, 213, and 214 and installs them in their respective locations.

[0045] As described above, the image analysis system 100 (see FIGS. 8 and 9) detects and recognizes objects from the captured image 150. In this embodiment, the image analysis system 100 detects and recognizes the manager 20, the worker 22, the fire extinguisher 34, the cart 30, the box 300, the traffic cone 200, and the identifiers 211, 212, 213, and 214 placed on the traffic cone 200.

[0046] As described above, detection and recognition of objects in the captured image 150 are performed using a trained model created using a known machine learning method such as R-CNN. In this embodiment, the model is trained to detect and recognize the manager 20, worker 22, fire extinguisher 34, cart 30, traffic cone 200, horizontally striped box 300, and identifiers 211, 212, 213, and 214 placed on the traffic cone 200.

[0047] The captured image 150 is displayed on the display unit 127 (see FIG. 9). The dashed lines in FIGS. 1 and 4 to 7 indicate that each object has been detected and recognized, and are displayed on the display unit 127.

[0048] As described above, the image analysis system 100 (see FIGS. 8 and 9) acquires additional information associated with an object and position information of the object in a captured image. In this embodiment, the image analysis system 100 acquires the position information and additional information of the manager 20, the worker 22, the fire extinguisher 34, the cart 30, the box 300, the traffic cone 200, and the identifiers 211, 212, 213, and 214 placed on the traffic cone 200.

[0049] In this embodiment, the additional information of the cone 200 (see FIG. 4) covered with a green identifier 211 is "information on the number of people passing through," the additional information of the cone 200 (see FIG. 4) covered with a blue identifier 212 is "information on left behind items," the additional information of the cone 200 (see FIG. 5) covered with a white identifier 213 is "information on occupancy rate," and the additional information of the cone 200 (see FIG. 6) covered with a yellow identifier 214 is "information on monitored items." Additionally, the additional information of the horizontally striped box 300 is "information on work start."

[0050] The execution unit 106 executes notification according to the additional information and position information of the manager 20, the worker 22, the fire extinguisher 34, the cart 30, the box 300, and the triangular cone 200 covered with the identifiers 211, 212, 213, and 214.

[0051] In this embodiment, the image analysis system 100 counts the number of people who pass between the two cones 200 covered with green identifiers 211 by the manager 20 and the workers 22 based on the additional information and position information of the two cones 200 covered with green identifiers 211 and the position information of the manager 20 and the workers 22, and notifies the mobile terminal device 14 (see FIG. 8) of the manager 20 of the number of people who have passed (see FIG. 4). Note that, for example, the manager 20 and the workers 22 may be counted separately and notified.

[0052] Furthermore, the image analysis system 100 determines the area surrounded by the four triangular cones 200 covered with blue identifiers 212 based on the additional information and location information of the four triangular cones 200 covered with blue identifiers 212, and the location information of the cart 30 as an example of left-behind property, as a safety area 54, and if the cart 30 remains within the safety area 54 for a predetermined time (for example, 30 seconds or more), it determines that the cart 30 has been placed and notifies the mobile terminal device 14 of the manager 20 (see Figure 5).

[0053] Furthermore, the image analysis system 100 notifies the mobile terminal device 14 of the manager 20 of the occupancy rate of the materials 32, 33 in the material area 56 surrounded by the four triangular cones 200 covered with the white identifiers 213, based on the additional information and position information of the four triangular cones 200 covered with the white identifiers 213 (see FIG. 6). Note that the occupancy rate may be calculated by any method, but in this embodiment it is calculated as the percentage of the area of materials 32, 33 that are a color different from the floor color to the entire area of the material area 56. Note that the floor color may be input and stored, for example, when the identifiers 213 are installed.

[0054] Furthermore, two triangular cones 200 covered with green identifying objects 211, four triangular cones 200 covered with blue identifying objects 212, and four triangular cones 200 covered with white identifying objects 213 are each an example of a group of objects.

[0055] Furthermore, the image analysis system 100 uses the additional information and location information of the triangular cone 200 covered with the yellow identifier 214, and the location information of the fire extinguisher 34 as an example of a monitored object, and notifies the mobile terminal device 14 of the manager 20 when the fire extinguisher 34 is removed from a predetermined range (for example, within a radius of 0.5 m) of the triangular cone 200 covered with the yellow identifier 214, that is, when the fire extinguisher 34 can no longer be detected within a radius of 0.5 m (see FIG. 7). This notification can also suggest the possibility that a fire has broken out somewhere on the construction site.

[0056] Furthermore, the image analysis system 100 notifies the mobile terminal devices 14 of the manager 20 and the worker 22 of a signal to start work based on the incidental information of the vertically long box 300 with a horizontal stripe pattern. Note that when the box 300 is laid down on the floor and becomes a horizontally long box 300 with a vertical stripe pattern, a signal to end work may be notified to the mobile terminal devices 14 of the manager 20 and the worker 22.

[0057] Here, the above-mentioned information acquisition method and notification method can be realized by the administrator 20 or the like changing the position and type of the object.

[0058] For example, if the location where notification of the number of people passing through needs to be made changes, the triangular cone 200 covered with the green identifier 211 can be moved to the new location.

[0059] Furthermore, for example, if the location desired as the safety area 54 changes, the triangular cone 200 covered with the blue identifier 212 can be moved to surround the new location.

[0060] Furthermore, for example, if it is desired to expand the range of the material area 56, the triangular cone 200 covered with the white identifier 213 can be moved to expand the area.

[0061] Furthermore, for example, when the execution of various notifications is started, that is, when a cone 200 covered with an identifying object 211, 212, 213, 214 is detected and recognized, the notification may be sent to the mobile terminal device 14 of the manager 20, or may be displayed on a monitor installed at the work site 50. <effect> Next, the operation of this embodiment will be described.

[0062] The image analysis system 100 detects and recognizes objects in captured images using a trained model, acquires additional information about the objects and location information about the objects in the captured images, and issues notifications based on the additional information and location information. Because the system acquires and notifies additional information linked to objects in this way, it is possible to issue notifications for multiple purposes and various information required for the work site 50.

[0063] In a specific embodiment, the image analysis system 100 uses a trained model to detect and recognize the manager 20, worker 22, fire extinguisher 34, cart 30, box 300, and triangular cone 200 covered with identifiers 211, 212, 213, and 214 from the captured image 150, thereby obtaining additional information and location information within the captured image 150 and executing various notifications.

[0064] From another perspective, the function and range can be defined by placing the identifiers 211, 212, 213, and 214 over the cone 200 in a preset arrangement without using a keyboard, mouse, or the like for each notification content, which allows for easy setting changes and enables easy response to changes in the work site 50 with little effort. Also, the start of work can be notified simply by placing the box 300 at the work site 50 so that it appears in the captured image 150.

[0065] Furthermore, since the image analysis system 100 detects and recognizes the shape, color, and pattern of an object, it is less susceptible to the effects of the resolution and shooting angle of the captured image compared to reading an identification code attached to the object.

[0066] In a specific embodiment, the shape of the cone 200 and the color of the identifiers 211, 212, 213, and 214 placed on the cone 200 are detected and recognized, as well as the shape and striped pattern of the box 300. Therefore, compared to reading a two-dimensional code or the like, the system is less susceptible to the effects of the resolution of the captured image, the angle of capture, etc.

[0067] In addition, by using a trained model, detection and recognition is possible even without high-resolution images. Furthermore, by creating a trained model that has learned a large number of objects using images taken at construction sites, it is possible to improve tracking of environmental changes and reduce false detections.

[0068] <Other> The present invention is not limited to the above embodiment.

[0069] For example, in the above embodiment, the shape of the cone 200 and the colors of the identifiers 211, 212, 213, and 214 placed on the cone 200 are detected and recognized, but the present invention is not limited to this. For example, a trained model may be created by machine learning to detect and recognize a state in which two cones of different colors are stacked or side-by-side. Alternatively, a trained model may be created by machine learning to detect and recognize a state in which a cone is knocked over.

[0070] Furthermore, the way in which different colored triangular cones are stacked may be changed. For example, the incidental information may be changed depending on the way red, yellow, and blue triangular cones are stacked. For example, a trained model may be created by machine learning the following stacking order from top to bottom: red, yellow, blue; red, blue, yellow; yellow, blue, red; and yellow, red, blue, and different incidental information may be associated with each of these patterns.

[0071] In the above embodiment, the shape and color of the object are detected and recognized (the shape and color of the triangular cone 200) or the shape and pattern of the object (the pattern of the box 300), but this is not limiting. The shape, color, and pattern of the object may be detected and recognized. Only the shape of the object may be detected and recognized.

[0072] Also, for example, in the above embodiment, the notification is sent to the mobile terminal device 14, but this is not limited to this. It may also be displayed on a monitor installed at the work site 50. Notifications of the start and end of work may also be sent to an industrial robot or the like. In this case, when the industrial robot or the like receives a notification of the start of work, it may automatically enter a state of preparation for the start of work. Notification may also be sent to an information terminal that is not carried around, such as a desktop computer.

[0073] Furthermore, for example, the notification system 10 described in the above embodiment is composed of a photographing device 12, a mobile terminal device 14, and an image analysis system 100, but this is just one example, and changes may be made or new components may be added depending on the situation within the scope of the main idea.

[0074] Furthermore, the hardware configuration and functional configuration of the image analysis system 100 are merely examples, and may be changed or new components added depending on the situation without departing from the spirit of the invention.

[0075] Furthermore, the processing flow of the program described in the above embodiment is an example, and unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged within the scope of the main idea.

[0076] In the above embodiment, the processing according to the embodiment is implemented by a software configuration using a computer by executing a program, but this is not limiting. For example, the processing may be implemented by a hardware configuration, or a combination of a hardware configuration and a software configuration.

[0077] Furthermore, the operations of the processors in the above embodiments may not only be performed by a single processor, but may also be performed by multiple processors located at physically separate locations working together. Furthermore, the order of the operations of the processors is not limited to the order described in the above embodiments, and may be changed as appropriate.

[0078] Furthermore, the present invention can be embodied in various forms without departing from the spirit and scope of the present invention. A plurality of embodiments and modifications can be implemented in combination as appropriate. [Explanation of symbols]

[0079] 10. Notification System 12 Imaging equipment 14 Mobile terminal devices 20 Administrator (an example of an object) 22 Worker (an example of an object) 30 Dolly (an example of an object) 34 Fire extinguisher (example of an object) 50 Work Site 100 Image Analysis System 102 Recognition part 104 Acquisition Department 106 Executive Department 150 images 200 Triangular cone (example of an object) 211 Identifier (an example of an object) 212 Identifier (an example of an object) 213 Identifier (an example of an object) 214 Identifier (an example of an object) 300 Box (Example of an object)

Claims

1. a recognition unit that detects and recognizes an object appearing in an image taken at a work site using a trained model that recognizes the object from an image; an acquisition unit that acquires supplementary information associated with the object and position information of the object within the captured image; an execution unit that executes a notification according to the acquired supplementary information and the acquired location information; Image analysis system equipped with

2. the recognition unit detects and recognizes the shape of the object and at least one of the color and pattern of the object; The image analysis system according to claim 1 .

3. The additional information is Information about left behind items, Occupancy information, Information on the removal of monitored items, Count information on the number of workers passing through, Work start information, Work completion information, Contains any of the following:

3. The image analysis system according to claim 1 or 2.

Images