Imaging device

Abstract

To obtain a trained model to be used when an image is taken.SOLUTION: An imaging device (20) includes a communication unit (24) that connects to a providing server (10) that provides a trained model used for taking an image via a communication network (30), a connection unit that connects to a storage unit (25) so as to be able to output data, and a control unit (21). The storage unit (25) stores a trained model (251). The control unit (21) transmits, at a predetermined timing, photographing-related information (253) obtained in relation to taking an image to the providing server (10) via the communication unit (24), responds to the photographing-related information (253), and causes the storage unit (25) to store a new trained model transmitted from the providing server (10).SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present disclosure relates to an imaging device that uses a trained model when capturing an image. [Background technology]

[0002] In recent years, the use of machine learning has become widespread in various fields. Cameras also use trained models when taking images (see, for example, Patent Documents 1, 2, and 3). The number of trained models used by a camera when taking an image is not limited to one. A camera may use multiple trained models generated for each process. For example, Patent Document 1 describes that an imaging device uses trained models to perform detection processing, classification processing, area classification processing, image restoration processing, focus processing, exposure processing, and white balance processing. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-072600 [Patent Document 2] Patent Publication No. 2021-093568 [Patent Document 3] Patent Publication No. 2021-033571 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the accuracy of a trained model varies depending on the training data. In addition, even if the same process is targeted, there are multiple trained models generated using different training data. In addition, a new trained model may be generated by re-learning. It is difficult for a user to determine whether the trained model stored in the camera is optimal for the user. It is also difficult for a user to select the optimal trained model from multiple trained models.

[0005] The present disclosure provides an imaging device that utilizes a trained model when capturing an image. [Means for solving the problem]

[0006] The imaging device of the present disclosure includes a communication unit that connects to a provider server that provides a trained model used for capturing an image via a communication network, a connection unit that connects to a storage unit or an external storage unit so as to be able to output data, and a control unit. The storage unit or the external storage unit stores the trained model. The control unit transmits imaging-related information obtained in relation to capturing an image to the provider server via the communication unit at a predetermined timing, and in response to the imaging-related information, stores a new trained model transmitted from the provider server in the storage unit or the external storage unit.

[0007] Such general and specific aspects may be realized by a system, a method, and a computer program, as well as a combination thereof. Effect of the Invention

[0008] The imaging device of the present disclosure can be provided with a trained model to be used in capturing images, or the trained model can be updated. [Brief description of the drawings]

[0009] [Figure 1] 2 is a block diagram showing the configurations of a provision server and a camera according to the first embodiment. FIG. [Diagram 2] 4 is a sequence diagram showing data transmission and reception between a provision server and a camera according to the first embodiment. FIG. [Figure 3A] 1 is an example of an operation screen displayed on a camera for which a trained model needs to be updated. [Figure 3B] 3B is an example of an operation screen displayed on the camera following FIG. 3A. [Figure 3C] FIG. 3C is an example of an operation screen displayed on the camera following FIG. 3B. [Figure 3D] FIG. 3D is an example of an operation screen displayed on the camera following FIG. 3C. [Figure 3E] FIG. 3B is an example of an operation screen displayed on the camera following FIG. 3D. [Figure 4A] This is an example of an operation screen displayed on a camera that has received a new list of trained models. [Figure 4B] 4B is an example of an operation screen displayed on the camera following FIG. 4A. [Figure 4C] FIG. 4C is an example of an operation screen displayed on the camera following FIG. 4B. [Figure 4D] FIG. 4D is an example of an operation screen displayed on the camera following FIG. 4C. [Figure 4E] FIG. 4E is an example of an operation screen displayed on the camera following FIG. 4D. [Figure 4F] FIG. 4B is an example of an operation screen displayed on the camera following FIG. 4E. [Figure 4G] FIG. 4F is an example of an operation screen displayed on the camera. [Diagram 5] FIG. 11 is a block diagram showing a configuration of a provision system according to a second embodiment. [Figure 6A] FIG. 11 is a sequence diagram showing data transmission and reception between a provision server and a camera according to the second embodiment. [Figure 6B] 10 is a flowchart showing an example of processing executed by the camera of the second embodiment. [Figure 6C] 10 is a flowchart showing an example of a captured image storage process executed by the camera of the second embodiment. [Figure 6D] 11 is a flowchart showing an example of a captured image selection process executed by the camera of the second embodiment. [Figure 6E] 13 is a flowchart showing another example of the captured image selection process executed by the camera of the second embodiment. [Figure 6F] 11 is a flowchart showing an example of a captured image evaluation process executed by the camera of the second embodiment. [Figure 7A] 13 is an example of an operation screen displayed on a camera for which re-learning of a trained model is desired. [Figure 7B] 7B is an example of an operation screen displayed on the camera following FIG. 7A. [Figure 7C] FIG. 7C is an example of an operation screen displayed on the camera following FIG. 7B. [Figure 7D] FIG. 7D is an example of an operation screen displayed on the camera following FIG. 7C. [Figure 7E] FIG. 7B is an example of an operation screen displayed on the camera following FIG. 7D. [Figure 8A] This is an example of an operation screen displayed on a camera for which a new trained model is desired to be provided. [Figure 8B] 8B is an example of an operation screen displayed on the camera following FIG. 8A. [Figure 8C] FIG. 8C is an example of an operation screen displayed on the camera following FIG. 8B. [Figure 9] FIG. 13 is a block diagram showing a configuration of a provision system according to Modification 1. [Figure 10] FIG. 11 is a sequence diagram showing data transmission and reception in a provision system according to Modification 1. [Figure 11] FIG. 11 is a schematic diagram showing a configuration of a provision system according to Modification 2. [Figure 12] FIG. 11 is a schematic diagram showing a configuration of a provision system according to Modification 3. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, the embodiments of the present disclosure will be described with reference to the drawings as appropriate. However, in the detailed description, unnecessary parts of the description of the prior art and substantially the same configuration may be omitted. This is for the purpose of simplifying the description. In addition, the following description and the accompanying drawings are disclosed so that a person skilled in the art can fully understand the present disclosure, and are not intended to limit the subject matter of the claims.

[0011] [Embodiment 1] It is known that a trained model is used in various controls in a camera. For example, the trained model is stored in advance in the camera when the camera is manufactured. A user may update the trained model according to his / her tendency of taking images. In the provision system 1 according to the first embodiment, a user who uses a camera can be provided with a trained model according to his / her tendency of taking images in the process of using the camera. In this case, the user who uses the camera can use the trained model proposed by the server without searching for a trained model according to his / her tendency of taking images.

[0012] 1. Configuration 1-1. Provision system FIG. 1 is a block diagram showing a configuration of a providing system 1 according to the first embodiment. In the providing system 1, a providing server 10 and a camera 20 are connected via a communication network 30. The providing system 1 selects, proposes, and provides a trained model to be used when shooting to the camera 20 that is registered in advance in a service that provides trained models. The camera 20 downloads the trained model proposed from the providing server 10 and uses it. In the example shown in FIG. 1, one camera 20 is connected to the providing server 10. However, the number of cameras 20 connected to the providing server 10 may be multiple.

[0013] 1-2. Provisioning Server The provision server 10 is, for example, an information processing device including a control unit 11, an input unit 12, an output unit 13, a communication unit 14, and a storage unit 15, as shown in FIG.

[0014] The control unit 11 is a controller that controls the entire provision server 10. For example, the control unit 11 reads out and executes a provision program P1, which is a computer program stored in the storage unit 15. In this way, the provision server 10 realizes a process of providing a trained model to the connected camera 20. The control unit 11 may be various processors such as a CPU, an MPU, a GPU, an FPGA, a DSP, an ASIC, or a dedicated hardware circuit.

[0015] The input unit 12 may be an operation unit such as an operation button, a keyboard, a touch panel, etc., used for inputting operations and data, a microphone, etc. The output unit 13 may be a display, a speaker, a lamp, etc., used for outputting processing results and data.

[0016] The communication unit 14 performs data communication between the provision server 10 and an external device such as the camera 20. The data communication is performed wired and / or wirelessly, for example, according to a known communication standard. For example, wired data communication may be performed by using, as the communication unit 14, a communication controller of a semiconductor integrated circuit that operates in compliance with the Ethernet (registered trademark) standard and / or the USB (registered trademark) standard. Wireless data communication may be performed by using, as the communication unit 14, a communication controller of a semiconductor integrated circuit that operates in compliance with the IEEE802.11 standard for wireless LAN (Local Area Network) and / or the fourth / fifth / sixth generation mobile communication system, so-called 4G / 5G / 6G, for mobile communication.

[0017] The storage unit 15 is a recording medium for recording various information. The storage unit 15 is realized by, for example, a RAM, a ROM, a flash memory, a solid state drive (SSD), a hard disk drive, other storage devices, or an appropriate combination of these. The storage unit 15 stores the provision program P1. The storage unit 15 also stores various data and the like obtained by processes such as selection and provision of a trained model in the provision server 10.

[0018] The storage unit 15 stores a plurality of trained models 151 that can be provided to the connected camera 20. For example, the storage unit 15 stores trained models 151 used for a plurality of different processes. Specifically, the storage unit 15 stores trained models 151 for different processes, such as detection processes, identification processes, autofocus processes, and automatic exposure control processes, as processes used for capturing an image. In addition, the storage unit 15 stores a plurality of trained models 151 with different learning data for the trained models used for each process, for example. The trained model 151 receives a signal obtained by a sensor including an imaging sensor of an imaging device when capturing an image, and outputs a predetermined signal.

[0019] For example, each process and the trained model for each process are described below.

[0020] The "detection process" detects the range in which the subject appears in the captured image. For example, a trained model of the detection process can be generated by machine learning using training data including the relationship between multiple sets of images and the range of the subject in the image.

[0021] The "classification process" is a process for identifying the type of subject appearing in a captured image. Examples of the type of subject may include people, dogs, cats, birds, flowers, insects, etc. For example, a trained model for the classification process may be generated by machine learning using training data including relationships between a plurality of sets of images and the types of subjects in the images.

[0022] "Autofocus processing (AF processing)" is a process for adjusting the focus position so that the subject is in focus when photographing. For example, a trained model used for autofocus processing can be generated by machine learning using training data including the relationship between multiple sets of images, ranging information (contrast, phase difference), and a control signal used to adjust the focus position when photographing an image.

[0023] "Automatic exposure control processing (AE processing)" controls the aperture, gain, and / or shutter speed so as to maintain the brightness of an image when shooting. For example, a trained model of the automatic exposure control processing can be generated by machine learning using training data including the relationship between the ISO sensitivity and F-number detected when acquiring a plurality of sets of images and the control signal.

[0024] "Auto white balance processing (AWB processing)" automatically adjusts the white balance of an image when shooting. For example, a trained model used for auto white balance processing can be generated by machine learning using training data including relationships between multiple sets of images, color temperature information in the images, and control signals used to adjust the white balance when the images were shot.

[0025] The storage unit 15 stores registration data 152, which is a list of cameras 20 registered as targets for providing the trained model 151. For example, the registration data 152 includes identification information of a user who uses the camera 20, model information of the camera 20, the type of trained model stored in the camera 20, and the like. For example, when a request signal for use of a service is received from the camera 20, the registration data 152 is generated by adding information included in the request signal.

[0026] 1-3.Camera 1, camera 20 includes control unit 21, input unit 22, output unit 23, communication unit 24, storage unit 25, optical mechanism 26, and drive mechanism 27. Camera 20 can capture images. The images captured by the camera may be still images or moving images.

[0027] The control unit 21, the communication unit 24, and the memory unit 25 are realized by the same specific means as the control unit 11, the communication unit 14, and the memory unit 15 described above. The memory unit 25 in this embodiment is configured with a non-volatile memory such as a flash memory. The input unit 22 is used for inputting operations for shooting and control by the user. The input unit 22 can be, for example, an operation unit such as a shutter button, other operation buttons, or a touch panel. The output unit 23 is used for outputting captured images and outputting information. The output unit 23 can be, for example, a display (display unit) such as an LCD monitor, a speaker, a lamp, etc.

[0028] The optical mechanism 26 includes a series of mechanisms related to capturing an image, and captures an image according to the control from the control unit 21. Specifically, the optical mechanism 26 may include an imaging element, an optical element, an A / D converter, an aperture, a shutter, and the like. The imaging element may be a CMOS image sensor, a CCD image sensor, an NMOS image sensor, and the like. The optical element may be a focus lens, a zoom lens, an optical image stabilization lens (OIS). The focus lens is a lens that changes the focus state of a subject image formed on the optical element. The zoom lens is a lens that changes the magnification of a subject image formed by the optical system. Each lens is composed of one or more lenses.

[0029] The driving mechanism 27 drives the lens included in the optical mechanism 26. For example, the driving mechanism 27 moves the lens along the optical axis of the optical system. In the driving mechanism 27, the configuration for driving the lens can be realized by a DC motor, a stepping motor, a servo motor, an ultrasonic motor, or the like.

[0030] The memory unit 25 stores the photography program P2, a trained model 251, a photographed image 252, and photography-related information 253.

[0031] The photographing program P2 is read out and executed by the control unit 21, and executes various processes related to photographing an image.

[0032] The trained model 251 is used for various processes related to photography, such as detection processing, identification processing, autofocus processing, automatic exposure control processing, auto white balance processing, noise reduction processing, etc. For example, the storage unit 25 stores a general-purpose trained model 251 in advance. The trained model 251 stored in the storage unit 25 can be provided from the provision server 10 and updated according to the user's image tendency.

[0033] The captured image 252 is an image captured by the camera 20. The storage unit 25 accumulates a plurality of captured images 252.

[0034] The shooting-related information 253 is associated with the captured image 252 and is information obtained in relation to the shooting of the image. The shooting-related information 253 may include the camera model, the subject to be shot, the shooting conditions, etc. The shooting-related information may also affect the generation and selection of a trained model.

[0035] The camera model is uniquely determined depending on the camera 20 .

[0036] The photographed subject may include a person, a dog, a cat, a bird, an insect, etc. The photographed subject may be identified by the trained model 251 that performs an identification process when the image is captured.

[0037] The shooting conditions are determined according to values ​​detected by sensors including an imaging sensor provided in the camera 20 and values ​​of control signals used during shooting, and are included in the shooting-related information 253. The shooting conditions may include the shooting environment, sensitivity, shutter speed, F-number, etc. The shooting environment is the environment in which an image is shot. For example, the shooting environment may include a forest, a city, the sea, a park, etc. The sensitivity is the ISO sensitivity at the time of shooting, and can be obtained from the ISO value.

[0038] The photography-related information can be obtained by capturing an image only once. However, in the following description, the camera 20 uses the photography-related information obtained by aggregating information obtained by capturing images multiple times. Therefore, the obtained photography-related information is information that represents the tendency of the camera to capture images. By using this photography-related information, the providing server 10 can recommend a trained model according to the tendency of the camera 20 to capture images.

[0039] 2. Work 2-1. Operation of the provided system The overall processing in the provision system 1 will be described with reference to the sequence diagram shown in FIG.

[0040] The camera 20 transmits a request signal to the provision server 10 to register for a service that can provide a trained model (S001).

[0041] When permitting the registration of the service, the providing server 10 transmits a permission notification to the camera 20 (S002). The providing server 10 also registers the user information included in the received request signal in the registration data 152 (S003).

[0042] Upon receiving the permission notification, the camera 20 generates storage information of the trained model to be stored in the memory unit 25 and transmits the storage information to the providing server 10 (S004).

[0043] The provision server 10 adds the received storage information to the registration data 152 (S005).

[0044] When an image is captured, camera 20 stores captured image 252 in storage unit 25 (S006). Camera 20 also stores photography-related information 253 in storage unit 25 in association with captured image 252 (S007).

[0045] The camera 20 transmits the photography related information to the provision server 10 at a predetermined timing (S008).

[0046] The provision server 10 searches for a trained model that matches the photography-related information (S009). In addition, the provision server 10 transmits recommendation data including a list of the trained models that have been searched for to the camera 20 (S010).

[0047] The camera 20 displays the received recommendation data on the output unit 23 (S011). When a trained model included in the recommendation data is selected by the user, the camera 20 transmits a request signal to the provision server 10 to request transmission of the trained model (S012).

[0048] The providing server 10 transmits the requested trained model to the camera 20 (S013). In addition, the providing server 10 updates the information of the trained model used by the camera 20 in the registration data 152 with the transmitted information of the trained model (S014).

[0049] The camera 20 stores the received trained model in the storage unit 25 (S015).

[0050] 2-2. Operation of the provision server The provision server 10 executes processes of registering a service, proposing a trained model, and providing a trained model. Each process executed by the control unit 11 of the provision server 10 will be described below.

[0051] (Service registration) The control unit 11 receives a service registration request signal transmitted from the camera 20 via the communication unit 14 (step S001 in FIG. 2). For example, the service registration request signal may include identification information of the user who uses the camera 20, model information of the camera 20, etc.

[0052] When the information included in the request signal satisfies a predetermined condition, the control unit 11 transmits a permission notification to the camera 20 via the communication unit 14 (step S002 in FIG. 2). For example, when the request signal contains information required for registration and the camera 20 is a model that supports the service, the control unit 11 transmits the permission notification. In addition, the control unit 11 adds the information included in the request signal to the registration data 152 stored in the storage unit 15 (step S003 in FIG. 2).

[0053] In response to the permission notification, the control unit 11 receives the storage information of the trained model transmitted from the camera 20 via the communication unit 24 (step S004 in FIG. 2). The storage information of the trained model is a list of information of all trained models currently stored in the camera 20. Furthermore, the control unit 11 adds the received storage information of the trained model to the registration data 152 in the memory unit 15 (step S005 in FIG. 2). The list of information of the trained model may include the type of the trained model, the version of the trained model, etc.

[0054] (Proposal of a trained model) The control unit 11 acquires the shooting-related information obtained in relation to the shooting of the camera, transmitted from the camera 20, via the communication unit 14 (step S008 in FIG. 2). The control unit 11 selects a trained model 151 that matches the acquired shooting-related information from among a plurality of trained models 151 stored in the storage unit 15 (step S009 in FIG. 2). For example, when information indicating learning data associated with the trained model 151 matches information included in the shooting-related information, the control unit 11 selects the trained model 151. Here, when there are a plurality of trained models 151 associated with the acquired shooting-related information, the control unit 11 may select a plurality of trained models 151.

[0055] The control unit 11 transmits recommendation data including information indicating the selected trained model 151 to the camera 20 via the communication unit 14 (step S010 in FIG. 2). Here, when there are multiple selected trained models 151, the control unit 11 may transmit recommendation data including information indicating the multiple trained models 151 to the camera. For example, the recommendation data includes identification information of the trained model 151 and information indicating the training data used to train the trained model 151.

[0056] The timing of the selection of the trained model 151 based on the imaging-related information and the transmission of the recommendation data by the control unit 11 is not limited. For example, the timing may be the timing when new imaging-related information is received, the timing when a new trained model 151 is stored in the storage unit 15, or a regular timing.

[0057] (Providing trained models) The control unit 11 receives a request signal requesting transmission of the trained model 151 transmitted from the camera 20 in response to the recommended data (step S012 in FIG. 2). The control unit 11 transmits the trained model 151 requested to be transmitted to the camera 20 (step S013 in FIG. 2). After transmitting the trained model 151 to the camera 20, the control unit 11 also updates the registered data 152 in the storage unit 15. Specifically, the control unit 11 updates the information of the trained model 151 stored in the camera 20, which is included in the registered data 152 (step S014 in FIG. 2).

[0058] 2-3. Camera operation The control unit 21 executes processes related to general image capture, as well as processes for registering services, registering information related to capture, transmitting information related to capture, and acquiring a trained model. Each process executed by the control unit 21 will be described below. Each process is executed by displaying an operation screen on the output unit 23, which is a display, and inputting an operation by the user via the input unit 22 in accordance with the operation screen. In the following description, an example of the operation screen displayed on the output unit 23 will be used as necessary.

[0059] (Service registration) The control unit 21 transmits a request signal for service registration to the provision server 10 via the communication unit 24 (step S001 in FIG. 2). The request signal for service registration may include identification information of the user who uses the camera 20, model information of the camera, and the like.

[0060] When the request is permitted by the providing server 10, the control unit 21 receives a permission notification from the providing server 10 via the communication unit 24 (step S002 in FIG. 2). Upon receiving the permission notification, the control unit 21 transmits storage information of the trained models stored in the storage unit 25 (step S003 in FIG. 2). When multiple trained models are stored in the storage unit 25, the control unit 21 transmits an information list of all trained models as the stored information. This enables the providing server 10 to register a list of information on trained models available for use in the camera 20. The list of information on the trained models may include information on the type of trained model, the version of the trained model, the model of the camera that the trained model corresponds to, and the training data.

[0061] (Registering shooting-related information) When an image is captured, the control unit 21 causes the captured image to be stored in the storage unit 25 (step S006 in FIG. 2) and registers the image-capturing related information (step S007 in FIG. 2).

[0062] (Transmission of shooting-related information) The control unit 21 transmits the photography-related information accumulated in the storage unit 25 to the providing server 10 at a predetermined timing (step S008 in FIG. 2). The timing for transmitting the photography-related information from the camera 20 to the providing server 10 can be, for example, periodically, when requested by a user, when a predetermined amount of captured images 252 has been accumulated in the storage unit 25 of the camera 20, when there has been a change in the tendency of the photography-related information 253 associated with the captured images 252, etc.

[0063] Here, the control unit 21 transmits the photography-related information obtained by tallying up information obtained from multiple image shootings to the provision server 10. For example, the control unit 21 may analyze the tendency of photography from the photography-related information 253 obtained multiple times and transmit the analysis result as the photography-related information. Furthermore, the control unit 21 may edit the photography-related information stored in the storage unit 25 according to an operation from a user via the input unit 22, and transmit the edited photography-related information.

[0064] For example, assume that the user has had many opportunities to "take pictures of birds in the forest" in the past, but is planning to "take pictures of birds at the sea" in the future. In this case, even if the conditions obtained from the photography-related information 253 are transmitted, it is difficult to obtain a learning model optimal for the user. Therefore, the user can use the input unit 22 to edit the photography-related information 253 in consideration of the future plans, and transmit the information to the providing server 10. In addition, there may be cases where the user thinks that the tendency of taking images is not expressed in the photography-related information 253. In this case, the user can also use the input unit 22 to edit the photography-related information 253 to reflect the tendency of taking images of the user, and transmit the information to the providing server 10.

[0065] FIG. 3A shows an example of an operation screen W21 displayed on the output unit 23 and used to operate a request for transmission of shooting-related information. Specifically, FIG. 3A shows an example of an operation screen for operating transmission of shooting-related information in order to update a trained model used in subject identification processing. When “update of subject identification model” 211 is selected on the operation screen W21 of FIG. 3A, the control unit 21 switches the operation screen W21 displayed on the output unit 23 to the operation screen W22 shown in FIG. 3B. From the operation screen W22, the user can select “shooting-related information” 221 for operating confirmation of the contents of the shooting-related information to be transmitted to the provision server 10 in order to update the trained model. For example, when the user wishes to change the “shooting environment” of the “shooting conditions”, the user selects the “shooting environment” item 222 from the operation screen W22. As a result, the control unit 21 switches the operation screen W22 displayed on the output unit 23 to the operation screen W3 shown in FIG. 3C.

[0066] The operation screen W23 presents a plurality of shooting environments to the user, and allows the user to select a shooting environment to be used as a shooting condition. Note that there is a limit to the shooting environments that can be presented in advance on the operation screen. Therefore, as shown in FIG. 3C, the operation screen W23 accepts the user's input of the shooting environment in the other fields. When "sea" 231 is selected as the "shooting environment" from the operation screen W23, the control unit 21 changes the "shooting environment" of the operation screen W24 to "sea" as shown in FIG. 3D. In addition, when the "shooting environment" is changed from "forest" to "sea", the preferred ISO sensitivity may also be changed. Therefore, the user may change the "sensitivity" accordingly. Furthermore, as shown in FIG. 3D, the user may register "flying bird" 241 as an additional condition. The "additional condition" may be freely registered by the user. Note that in the example shown in FIG. 3D, only one additional condition is added, but the number of additional conditions that can be added is not limited.

[0067] Furthermore, when the shooting-related information is changed, the storage of the updated information may be confirmed, as in the example of the operation screen W25 shown in Fig. 3E. When "Yes" 251 is selected on the operation screen W25 shown in Fig. 3E, the control unit 21 may then confirm whether or not it is necessary to transmit the updated shooting-related information to the providing server 10. Furthermore, the control unit 21 may confirm whether or not it is necessary to acquire a new trained model based on the updated shooting-related information from the providing server 10.

[0068] (Receive a trained model) The control unit 21 receives the recommended data provided from the provision server 10 via the communication unit 24 (step S010 in FIG. 2). The control unit 21 causes the output unit 23 to display the received recommended data (step S011 in FIG. 2). The control unit 21 transmits a request signal to the provision server 10 to request transmission of a trained model in accordance with the recommended data (step S012 in FIG. 2). In addition, the control unit 21 receives a new trained model in response to the transmitted request signal (step S012 in FIG. 2). The control unit 21 causes the storage unit 25 to store the received new trained model (step S013 in FIG. 2). This allows the control unit 21 to subsequently execute processing at the time of shooting using the new trained model.

[0069] For example, as shown in FIG. 4A, the control unit 21 displays an operation screen W31 on the output unit 23, which indicates that the recommended data has been received. The operation screen W31 is an operation screen that notifies the reception of the recommended data after transmitting the imaging related information to the providing server 10. When "updated model information" 311 is selected on the operation screen W31, the control unit 21 switches the operation screen W31 displayed on the output unit 23 to an operation screen W32 shown in FIG. 4B. The operation screen W32 is an example in which three trained models, "proposed model 1", "proposed model 2" and "proposed model 3", are proposed. When "proposed model 2" 321 is selected on the operation screen W32, the control unit 21 switches the operation screen W32 displayed on the output unit 23 to an operation screen W33 shown in FIG. 4C.

[0070] The operation screen W33 includes "model information" used for operating the display of detailed information of the model, and "model acquisition" used for operating the acquisition of the model. When "model information" 331 is selected from the operation screen of FIG. 4C, the control unit 21 switches the display of the operation screen W33 displayed on the output unit 23 to an operation screen W34, an example of which is shown in FIG. 4D. The operation screen W34 includes detailed information of the trained model recommended by the recommendation data. For example, in addition to the subject and shooting conditions, it can include the creator of the trained model, version information of the trained model, the creator's comments, an example of an image used as training data, and the like. After checking the detailed information of the trained model on the operation screen W34, the user can switch the display to the operation screen W33 by operating the input unit 22.

[0071] When "Acquire Model" 332 is selected from the operation screen W33, the control unit 21 switches the display of the operation screen W33 displayed on the output unit 23 to an operation screen W35, an example of which is shown in FIG. 4E. When "Yes" 351 is selected from the operation screen W35, the control unit 21 transmits a request signal to the providing server 10 to request transmission of the trained model. If multiple versions of the trained model are stored in the providing server 10 for the selected trained model, the version to be used may be selected. FIG. 4F shows an example of an operation screen W36 for selecting the version of the trained model to be acquired. When reception of the trained model is completed, the control unit 21 may display, for example, an operation screen as shown in FIG. 4G on the output unit 23 to confirm the update.

[0072] 3. Effects etc. As described above, the providing server 10 according to the first embodiment can select and recommend a trained model based on the shooting-related information including the shooting tendency of the camera 20. In addition, the camera 20 can acquire a new trained model based on the shooting-related information. Therefore, the camera 20 can acquire a new trained model when the shooting tendency changes or when a new trained model is added to the storage unit 15. This allows the camera 20 to capture an image using a more suitable trained model.

[0073] [Embodiment 2] 5 is a block diagram showing a provision system 1A according to embodiment 2. The camera 20 of the provision system 1 according to the above-mentioned embodiment 1 uses a trained model provided from the provision server 10. In contrast, the camera 20 of the provision system 1A according to embodiment 2 re-trains the trained model within the camera 20. In addition, the camera 20 provides the re-trained trained model to the provision server 10.

[0074] 1. Configuration The provision system 1A according to the second embodiment differs from the provision system 1 described above with reference to Fig. 1 in that the storage unit 25 of the camera 20 stores a learning program P3 used for re-learning the trained model 251. In other respects, the provision system 1A is similar to the provision system 1.

[0075] 2.Operation 2-1. Operation of the provided system The overall processing of the provision system 1A according to the second embodiment will be described with reference to the sequence diagram shown in Fig. 6A. In the sequence diagram shown in Fig. 6A, the same processes as those in Fig. 2 are denoted by the same reference numerals, and the reference numerals of different processes are underlined.

[0076] The camera 20 transmits a request signal to the provision server 10 to register for a service that can provide a trained model (S001).

[0077] When permitting the registration of the service, the providing server 10 transmits a permission notification to the camera 20 (S002). The providing server 10 also registers the user information included in the received request signal in the registration data 152 (S003).

[0078] Upon receiving the permission notification, the camera 20 generates storage information of the trained model to be stored in the memory unit 25 and transmits the storage information to the providing server 10 (S004).

[0079] Furthermore, the camera 20 updates the trained model 251 stored in the storage unit 25 by re-learning (S101).

[0080] The provision server 10 adds the received storage information to the registration data 152 (S005).

[0081] When an image is captured, camera 20 stores captured image 252 in storage unit 25 (S006). Camera 20 also stores photography-related information 253 in storage unit 25 in association with captured image 252 (S007).

[0082] The camera 20 transmits the photography related information to the provision server 10 at a predetermined timing (S008).

[0083] The provision server 10 searches for a trained model that matches the photography-related information (S009). In addition, the provision server 10 transmits recommendation data including a list of the trained models that have been searched for to the camera 20 (S010).

[0084] The camera 20 displays the received recommendation data on the output unit 23 (S011). When a trained model included in the recommendation data is selected, the camera 20 transmits a request signal to the provision server 10 to request transmission of the trained model (S012).

[0085] The providing server 10 transmits the requested trained model to the camera 20 (S013). In addition, the providing server 10 updates the information of the trained model used by the camera 20 in the registration data 152 with the transmitted information of the trained model (S014).

[0086] The camera 20 stores the received trained model in the storage unit 25 (S015).

[0087] The camera 20 receives a decision as to whether or not the trained model 251 updated in step S101 can be uploaded to the providing server 10 (S102).

[0088] When upload is possible (YES in S102), the camera 20 transmits the trained model 251 updated in step S101 to the provision server 10 (S103).

[0089] Furthermore, the providing server 10 registers the trained model 231 transmitted from the camera 20 in step S103 in the storage unit 15 (S104).

[0090] When upload is not possible (NO in S102), or when the updated trained model 251 is sent to the provision server 10 in step S103, the camera 20 returns to step S101 and repeats the processes from step S106 onwards.

[0091] Furthermore, when the updating of the registered data in step S014 or the processing of registering the trained model in step S104 is completed, the provision server 10 returns to step S009 and repeats the processing from step S009 onwards.

[0092] 2-2. Camera operation The control unit 21 executes processes of generating learning data, re-learning, and providing a trained model. Each process executed by the control unit 21 will be described below. Each process is executed by displaying an operation screen on the output unit 23, which is a display, and the user inputting operations via the input unit 22 in accordance with the operation screen. In the following description, an example of the operation screen displayed on the output unit 23 will be used as necessary.

[0093] (Overall processing) The process of providing a trained model from the accumulation of images in the camera will be described using the flowcharts shown in FIGS. 6B to 6F.

[0094] As shown in Fig. 6B, when an image is captured, the control unit 21 stores the captured image 252 in the storage unit 25 (S201). The process of storing the captured image 252 will be described later with reference to the flowchart in Fig. 6C.

[0095] The control unit 21 selects a captured image 252 from the multiple images stored in the storage unit 25 in step S201 (S202). The selected captured image 252 becomes learning data 254. The process of selecting the captured image 252 will be described later with reference to the flowcharts of Figures 6D and 6E. For example, the processes of steps S201 to S202 are processing for generating learning data.

[0096] The control unit 21 uses the captured images obtained by the selection in step S202 to execute re-learning of the trained model 251 (S203).

[0097] The control unit 21 evaluates the new trained model 251 obtained by the re-learning in step S203 (S204). The process of evaluating the new trained model 251 will be described later with reference to the flowchart in FIG. 6F.

[0098] The control unit 21 stores the new trained model 251 evaluated in step S205 in the storage unit 25 (S205). This updates the trained model 251 in the storage unit 25. For example, the processes in steps S203 to S205 are regarded as re-learning processes.

[0099] The control unit 21 transmits the new trained model 251 updated in step S205 to the providing server 10 (S206). This enables the control unit 21 to provide the new trained model 251 to the providing server 10. The new trained model 251 transmitted here is stored in the storage unit 15 of the providing server 10. Thereafter, the providing server 10 becomes able to provide the new trained model 251 to other devices that require it.

[0100] (Generating training data) The control unit 21 accumulates the captured images, and generates learning data using a captured image selected from the plurality of captured images.

[0101] The process of storing the captured image in step S201 will be described with reference to the flowchart in Fig. 6C. When an image is captured in the camera (S2011), while the camera is in the storage mode (YES in S2012) and there is free space in the storage unit 25 (YES in S2013), the control unit 21 stores the captured image 252 in the storage unit 25 (S2014).

[0102] If there is no free space in the storage unit 25 (NO in S2013), the control unit 21 displays a warning on the output unit 23 (S2015).

[0103] The processes in steps S2011 to S2015 are continued until image capturing is completed (YES in S2016).

[0104] If the storage mode is not selected (NO in S2012), there is no need to store images, and the process ends. A case where the storage mode is not selected is, for example, when previewing an image.

[0105] When generating the learning data 254, the control unit 21 may select only the captured images 252 suitable for learning. The control unit 21 may use a selection program stored in advance in the storage unit 25 to select the captured images 252. The control unit 21 may also display each captured image 252 on the output unit 23, which is a display, and accept a user's selection. Note that the selection program and the user's selection may be used in combination. For example, the control unit 21 may associate the selected captured image 252 with the shooting-related information 253 to generate the learning data 254.

[0106] The process of selecting captured images in step S202 will be described with reference to the flowchart in Fig. 6D. Two examples of the image selection process are proposed. The flowchart in Fig. 6D is the first example.

[0107] When the number of captured images stored in the memory unit 25 is equal to or greater than a predetermined number (YES in S2021), the control unit 21 executes a mechanical sorting process (S2012). The mechanical sorting is a process of sorting using a sorting program stored in advance in the memory unit 25.

[0108] When the mechanical sorting is completed, the control unit 21 judges whether visual sorting by the user is necessary (S2023). When visual sorting is necessary, for example, the control unit 11 tests the learning data and judges that visual sorting is necessary when the reliability is low. The control unit 21 may display the reliability on the output unit 23 and allow the user to select visual sorting.

[0109] When visual selection is required (YES in S2023), the target photographed images 252 are displayed on the output unit 23, and the user performs visual selection processing, and then the series of processing for selecting photographed images is terminated (S2024). On the other hand, when visual selection is not required (NO in S2023), the series of processing for selecting photographed images is terminated.

[0110] A second example of the captured image selection process in step S202 will be described with reference to the flowchart in FIG. 6E.

[0111] It is determined whether the captured images stored in the memory unit 25 are subject to visual sorting (S20254). Cases where the captured images are not subject to visual sorting are when the "mechanical sorting mode" or the "combined sorting mode" that uses both mechanical and visual sorting is set.

[0112] If visual sorting is not required (NO in S2050), the control unit 21 executes a mechanical sorting process (S2026).

[0113] When the mechanical sorting is completed, the control unit 21 judges whether the stored captured images are the target of mechanical sorting (S2027). If the captured images are the target of mechanical sorting (YES in S2027), the sorting is completed, and the series of processes for sorting the captured images is terminated.

[0114] On the other hand, if the product is not subject to mechanical sorting (NO in S2027), it is subject to combined sorting in which both mechanical and visual sorting are performed. If the product is subject to visual sorting (YES in S2025) or is subject to combined sorting (NO in S2027), visual sorting is performed (S2028).

[0115] (Relearn) Re-learning is possible by operating the operation screen shown in Fig. 7A. When "Set pre-train model" 811 is selected on the operation screen W81, the control unit 21 displays the operation screen W82, an example of which is shown in Fig. 7B, on the output unit 23. The operation screen W82 is an operation screen used to specify the pre-train model. For example, when "Latest (in use) model" 821 is selected on the operation screen W82, the control unit 21 executes re-learning by using the currently available trained model 251 as the pre-train model.

[0116] When "Designate learning data" 812 is selected on the operation screen W81, the control unit 21 displays an operation screen as shown in Fig. 7C or Fig. 7D. For example, when "Memory in camera" 831 is selected as "Designate learning data" on the operation screen W83 of Fig. 7C and "Use" 832 is selected, the control unit 21 performs re-learning using the data stored in the memory in camera. Also, when "Designate from SD card" 841 is selected on the operation screen W84 of Fig. 7D and "slot1" 842 is selected, the control unit 21 performs re-learning using the data stored in the SD card inserted in slot1.

[0117] When "Relearn" 813 is selected from the operation screen W81 after the pre-train model and learning data are specified, the control unit 21 executes re-learning. While learning is being executed, the control unit 21 may display a message such as "Learning in progress." Furthermore, when learning is completed, the control unit 21 may display a message such as "Learning completed."

[0118] When "Model Evaluation" 814 is selected from the operation screen 8W after the learning is completed, the control unit 21 executes the process of evaluating the model. The control unit 21 may display a message such as "Test in progress" while the evaluation is being performed. Furthermore, when the process of evaluating the model is completed, the control unit 21 displays, for example, the operation screen W85 including the evaluation result 851 shown in FIG. 7E. When "Yes" 852 is selected from the operation screen, the trained model is updated to a new trained model.

[0119] The process of evaluating the trained model in step S204 will be described with reference to the flowchart in Fig. 6F. The control unit 21 executes a test of the new trained model obtained in step S203 (S2411).

[0120] The control unit 21 determines whether to execute a numerical evaluation (S2042).

[0121] When performing the numerical evaluation (YES in S2042), the control unit 21 calculates an index value for the evaluation (S2043). For example, the control unit 21 may calculate a matching rate, a correct answer rate, reproducibility, etc. as the index value. The control unit 21 also displays the index value calculated in step S2044 on the output unit 23 (S2044). This allows the user to quantitatively check the test results using numerical values.

[0122] On the other hand, when the numerical evaluation is not performed (NO in S2042), the control unit 21 displays the test result on the output unit 23 (S2015), so that the user can visually check the test result.

[0123] (Providing trained models) By operating the operation screen W91 shown in FIG. 8A, a new trained model can be provided to the provision server 10. When "Update subject identification model" 911 is selected on the operation screen W91, the control unit 21 can display, on the output unit 23, an operation screen W92 in which information about the trained model can be input, as shown in FIG. 8B, for example. For example, the user may change any of the items shown in FIG. 8B to more easily understandable content. The user may also register a comment about the new trained model.

[0124] Thereafter, when "Upload" 931 is selected on the operation screen W93 as shown in FIG. 8C and "Yes" 932 is selected in response to the question "Do you want to upload the model to the cloud?", the control unit 21 transmits the new trained model to the provision server 10. While the transmission of the trained model is being executed, the control unit 21 may display a message such as "Upload in progress." Furthermore, when the transmission of the trained model is completed, the control unit 21 may display a message such as "Upload completed."

[0125] In the above example, the control unit 21 executes re-learning by using the learning data 254 including the captured image 252. However, there may be a trained model 251 that does not require the captured image 252. In re-learning of such a trained model 251, only the shooting-related information 253 can be used as the learning data 254.

[0126] 3. Effects etc. As described above, camera 20 according to the second embodiment executes re-learning by using captured images and / or imaging-related information 253 in storage unit 25. This allows camera 20 to generate a new trained model 251 that is optimal for its own imaging. Furthermore, camera 20 can provide the generated new trained model 251 to providing server 10. Therefore, providing server 10 can provide the new trained model 251 received from camera 20 to other cameras, etc.

[0127] If a system were to be created in which a wide variety of trained models were distributed, the provision system 1A according to the above-described second embodiment would be useful. For example, a trained model generated by a user through re-learning on his / her own camera 20 may be useful to other users who use other cameras. The provision system 1A provides the trained model from the camera 20 to the provision server 10. This allows the provision server 10 to provide the trained model provided from the camera 20 to other cameras that wish to use it.

[0128] Furthermore, for example, when a user replaces a camera, the new camera does not store the trained model generated by re-learning in the old camera 20. Therefore, a user who uses the new camera cannot use the trained model generated by re-learning in the old camera 20. If the trained model generated in the old camera 20 is registered in the providing server 10, the new camera can be provided with the trained model generated in the old camera from the providing server 10. In this way, the provision system 1A can be used to distribute trained models.

[0129] (Other embodiments) As described above, the first and second embodiments have been described as examples of the technology disclosed in the present application. However, the technology in the present disclosure is not limited to these, and can be applied to embodiments in which modifications, substitutions, additions, omissions, etc. are appropriately performed. In addition, it is also possible to combine the components described in each of the above embodiments to create a new embodiment. Therefore, other embodiments will be exemplified below.

[0130] In each of the above embodiments, the camera 20 including the output unit (display unit) 23 has been described, but the camera 20 of this embodiment does not necessarily have to include a display unit. In this embodiment, the camera 20 may include a connection unit such as an interface circuit that displays various information on an external display monitor, and may be, for example, a box camera. For example, the camera 20 of the embodiment may include a connection terminal (connection unit) such as an HDMI (registered trademark) terminal or an SDI terminal, and may be connected to an external device such as an external monitor. Moreover, the camera 20 of this embodiment may include both the above-mentioned interface circuit and a display monitor.

[0131] In each of the above embodiments, the camera 20 is described as having a storage unit 25 such as a flash memory. In this embodiment, the camera 20 may not have a storage unit 25 inside. In this embodiment, when the camera 20 is configured with a memory card as an external storage unit, the camera 20 has a card slot into which the memory card can be attached and detached. The card slot accesses the memory card based on the control from the control unit 21. The card slot is an example of a connection unit in the camera 20 of this embodiment. The camera 20 may not have a card slot. The camera 20 of this embodiment may have a connection unit such as a slot for recording image data to various recording media such as SSD, not limited to recording to a memory card. The camera 20 of this embodiment may have a connection terminal (connection unit) such as a USB terminal and be connected to an external storage device. The camera 20 of this embodiment may also have both the connection unit and the storage unit 25.

[0132] In each of the above embodiments, the camera 20 equipped with the optical mechanism 26 including optical elements such as a focus lens, a zoom lens, and an optical image stabilization lens (OIS), and the drive mechanism 27 have been exemplified.

[0133] The imaging device of this embodiment does not need to include the optical mechanism 26 including the optical elements and the drive mechanism 27, and may be, for example, a camera with an interchangeable lens. If the camera 20 is a camera with an interchangeable lens, it can be mechanically and electrically connected to a lens mount of an interchangeable lens via a body mount (not shown).

[0134] [Variation 1] 1. Configuration 9 is a block diagram showing a provision system 1B according to a modified example. The provision system 1B differs from the provision system 1 according to the first embodiment described above with reference to FIG. 1 in that it includes a user terminal 40. In the provision system 1, data is transmitted and received between the provision server 10 and the camera 20. In contrast, in the provision system 1B according to the first modified example, the camera 20 transmits and receives data to and from the provision server 10 via the user terminal 40.

[0135] 9, the user terminal 40 includes a control unit 41, an input unit 42, an output unit 43, a communication unit 44, a storage unit 45, etc. The user terminal 40 is an information processing terminal used by a user, such as a smartphone, a tablet terminal, a personal computer, etc. The control unit 41, the input unit 42, the output unit 43, the communication unit 44, and the storage unit 45 are realized by the same specific means as the control unit 11, the input unit 12, the output unit 13, the communication unit 14, and the storage unit 15 of the provision server 10 described above.

[0136] 2. Operation of the provided system In the provision system 1B according to the first modification, the user inputs each operation signal via the input unit 42 of the user terminal 40. Moreover, the display screens described using FIGS. 3A to 4G are displayed on the output unit 43, which is the display of the user terminal 40.

[0137] The overall processing in the provision system 1B will be described with reference to the sequence diagram shown in Fig. 10. In each step, the same processing as in Fig. 2 will be described with the same processing numbers.

[0138] First, the user terminal 40 used by the user of the camera 20 transmits a request signal for service registration to the providing server 10 (S101).

[0139] When permitting the registration of the service, the provision server 10 transmits a permission notification to the user terminal 40 (S102). In addition, the provision server 10 registers the user information included in the received request signal in the registration data 152 (S003).

[0140] The user terminal 40 that has received the permission notification acquires the storage information of the trained model stored in the storage unit 25 from the camera 20. In addition, the user terminal 40 transmits the storage information of the trained model to the provision server 10 (S104).

[0141] The provision server 10 adds the stored information to the registration data 152 (S005).

[0142] When an image is captured, camera 20 stores the captured image in storage unit 25 (S006). Camera 20 also stores image-capturing related information in storage unit 25 in association with the captured image (S007).

[0143] The user terminal 40 acquires the photography related information at a predetermined timing from the camera 20. In addition, the user terminal 40 transmits the photography related information to the provision server 10 (S108).

[0144] The providing server 10 searches for a trained model that matches the photography related information (S009). In addition, the providing server 10 transmits recommendation data including a list of the trained models that have been searched for to the user terminal 40 (S010).

[0145] The user terminal 40 displays the received recommendation data on the output unit 43 (S111). When a trained model included in the recommendation data is selected, the user terminal 40 transmits a request signal to the provision server 10 to request transmission of the trained model (S112).

[0146] The providing server 10 transmits the trained model requested by the user terminal 40 to the camera 20 via the user terminal 40 (S113).

[0147] The provision server 10 updates the information on the trained model used by the camera 20 in the registration data 152 with the transmitted information on the trained model (S014).

[0148] The camera 20 stores the trained model received via the user terminal 40 in the storage unit 25 (S015).

[0149] 3. Effects etc. The provision system 1B according to the first modification is useful when the user is more accustomed to inputting and checking information on the user terminal 40 than on the camera 20, or when the data communication of the user terminal 40 is stable.

[0150] [Variation 2] 11 is a schematic diagram showing a provision system 1C according to Modification 2. In the provision system 1A according to the above-described embodiment 2, re-learning of a trained model is performed in the camera 20. In contrast, in the provision system 1C according to Modification 2, re-learning is performed in the user terminal 40 connected to the camera 20.

[0151] Therefore, each process such as generation of learning data, re-learning, provision of a learned model, etc., executed by the control unit 21 described above in the second embodiment may be executed by the control unit 41 of the user terminal 40. In this case, the control unit 41 generates learning data 254 using the captured image 252 and / or the shooting related information 253 stored in the storage unit 25 of the camera 20. The new trained model 251 may be transmitted from the user terminal 40 to the providing server 10. Also, the new trained model 251 may be transmitted from the camera 20 to the providing server 10.

[0152] The provision system 1C according to the second modification is useful when the processing capability and operability of the camera 20 are not sufficient for image selection and relearning.

[0153] [Variation 3] FIG. 12 is a schematic diagram showing a providing system 1D according to Modification 3. The providing system 1D according to Modification 3 can generate a new trained model in the providing server 10 by using the captured image and shooting-related information acquired by the target camera 20. At this time, the camera 20 can provide only the selected image and shooting-related information to the providing server 10 as learning data, as described above in the second embodiment. In addition, the image and shooting-related information may be stored in an image server 50 as shown in the figure, instead of being transmitted from the camera 20 to the providing server 10. The image server 50 may be a server that stores data for backup purposes.

[0154] The provision system 1D according to the third modification is useful when the processing capabilities of the camera 20 and the user terminal 40 are not sufficient for relearning.

[0155] Overview of the embodiment (1) An imaging device according to the present disclosure includes a communication unit that connects to a server that provides a trained model to be used for capturing an image via a communication network, a connection unit that connects to a storage unit or an external storage unit so as to be able to output data, and a control unit; The storage unit or the external storage unit stores the trained model, The control unit is transmitting image-capturing-related information obtained in relation to image capturing to the providing server via the communication unit at a predetermined timing; In response to the photography-related information, a new trained model transmitted from the provision server can be stored in the memory unit or the external memory unit.

[0156] This makes it possible to use a trained model that is suitable for shooting with an imaging device.

[0157] (2) The control unit of (1) Before receiving the new trained model, receiving recommendation data including information on training data used to generate the new trained model; When an operation to transmit the new trained model is received by a user using the recommendation data, a request signal requesting transmission of the new trained model may be transmitted to the provision server.

[0158] This allows the imaging device to obtain only the more preferable trained model.

[0159] (3) The control unit according to (1) or (2) Analyzing the imaging-related information; before transmitting the imaging related information to the providing server, displaying an analysis result of the imaging related information on a display unit; When an operation for transmitting the photographing related information is accepted from a user, the photographing related information may be transmitted to the providing server.

[0160] This allows the imaging device to obtain a more preferable trained model by not transmitting unnecessary imaging-related information.

[0161] (4) In the imaging device according to (1) to (3), receiving an operation signal for changing the analysis result by the user; When an operation signal for change is received, information changed in accordance with the operation signal may be transmitted to the providing server as the photography-related information.

[0162] This allows the imaging device to obtain a more preferable trained model by transmitting the necessary shooting-related information.

[0163] (5) The control unit of (1) to (4) The shooting-related information stored in the storage unit or the external storage unit is used as new learning data, and machine learning is performed to generate a new trained model; The generated new trained model may be stored in the memory unit or the external memory unit.

[0164] This makes it possible to generate a more preferable trained model within the imaging device using imaging-related information based on imaging by the device itself.

[0165] (6) The control unit of (5) The newly generated trained model may be transmitted to the provision server.

[0166] This makes it possible to widely provide the trained model generated by the imaging device to other imaging devices.

[0167] (7) The control unit according to (5) or (6) displaying the photography-related information stored in the storage unit or the external storage unit; A new trained model may be generated by using the shooting-related information selected by the user from the displayed shooting-related information as new training data.

[0168] This allows the imaging device to generate a trained model using the selected shooting-related information.

[0169] (8) The control unit of (6) Obtaining from the provision server a new trained model previously generated by another imaging device used by a user of the imaging device; The acquired new trained model may be stored in the memory unit or the external memory unit.

[0170] This allows a user of an imaging device to use the trained model that was used before the imaging device was replaced, in the new imaging device.

[0171] (9) A provision server according to the present disclosure includes a communication unit that is connected to an imaging device via a communication network, a storage unit that stores a plurality of trained models used to capture an image and information on training data used to generate each of the trained models in association with each other, and a control unit; The control unit is acquiring, via the communication unit, imaging-related information obtained in relation to imaging and transmitted from the imaging device; Select a trained model from the plurality of trained models stored in the storage unit, the trained model including the acquired shooting-related information, in which information on associated training data is included; The selected trained model can be transmitted to the imaging device via the communication unit.

[0172] This makes it possible to provide a trained model suitable for photography with an imaging device.

[0173] (10) The control unit of (9) Before transmitting the trained model to the imaging device, recommendation data including information associated with the selected trained model is transmitted to the imaging device via the communication unit; When a request signal requesting transmission of a trained model is received in response to the recommendation data transmitted from the imaging device via the communication unit, the trained model may be transmitted to the imaging device.

[0174] This makes it possible to provide the imaging device with a more preferable trained model.

[0175] (11) The learning data of (9) or (10) includes an image; The control unit is Acquire, from an external terminal connected via the communication unit, a new trained model generated in the external terminal and information obtained in association with capturing an image included in training data used in the new trained model; The newly acquired trained model may be associated with information obtained in connection with capturing the image and stored in the memory unit.

[0176] This allows a new trained model to be stored in the storage device.

[0177] (12) The control unit according to (9) to (11) Acquire new learning data from an external terminal connected via the communication unit; Generate a new trained model using the training data; The generated new trained model may be stored in the memory unit in association with information obtained in connection with capturing the images included in the new training data.

[0178] This makes it possible to generate a trained model using the training data generated by the imaging device. [Industrial Applicability]

[0179] The imaging device of the present disclosure is useful for obtaining or providing a trained model to be used when capturing an image. [Explanation of symbols]

[0180] 1. Provision system 10. Provisioning Server 11 Control section 12 Input section 13 Output section 14 Communications Department 15 Storage section 20 Camera 21 Control section 22 Input section 23 Output section 24 Communications Department 25 Memory section 26 Optical mechanism

Claims

1. A communication unit that connects to a server that provides a trained model to be used for capturing images via a communication network; A connection unit for connecting to the storage unit or the external storage unit so as to be able to output data; A control unit; Equipped with The storage unit or the external storage unit stores the trained model, The control unit is transmitting image-capturing-related information obtained in relation to image capturing to the providing server via the communication unit at a predetermined timing; In response to the imaging-related information, a new trained model transmitted from the providing server is stored in the storage unit or the external storage unit. Imaging device.

2. The control unit is Before receiving the new trained model, receiving recommendation data including information on training data used to generate the new trained model; When an operation of transmitting the new trained model is received by a user using the recommendation data, a request signal is transmitted to the providing server to request transmission of the new trained model. The imaging device according to claim 1 .

3. The control unit is Analyzing the imaging-related information; before transmitting the imaging related information to the providing server, displaying an analysis result of the imaging related information on a display unit; When an operation for transmitting the photographing related information is accepted by a user, the photographing related information is transmitted to the providing server. The imaging device according to claim 1 .

4. receiving an operation signal for changing the analysis result by the user; When an operation signal for change is received, the information changed in accordance with the operation signal is transmitted to the providing server as the photography-related information. The imaging device according to claim 3 .

5. The control unit is The shooting-related information stored in the storage unit or the external storage unit is used as new learning data, and machine learning is performed to generate a new trained model; The generated new trained model is stored in the storage unit or the external storage unit. The imaging device according to claim 1 .

6. The control unit is The generated new trained model is transmitted to the provision server. The imaging device according to claim 5 .

7. The control unit is displaying the photography-related information stored in the storage unit or the external storage unit; From the displayed shooting-related information, the shooting-related information selected by the user is used as new learning data to generate a new trained model. The imaging device according to claim 5 .

8. The control unit is Obtaining from the provision server a new trained model previously generated by another imaging device used by a user of the imaging device; The acquired new trained model is stored in the storage unit or the external storage unit. The imaging device according to claim 6.

9. a communication unit that is connected to the imaging device via a communication network; A storage unit that stores a plurality of trained models used for capturing images and information on training data used for generating each trained model in association with each other; Control unit and Equipped with The control unit is acquiring, via the communication unit, imaging-related information obtained in relation to imaging and transmitted from the imaging device; Select a trained model from the plurality of trained models stored in the storage unit, the trained model including the acquired shooting-related information, in which information on associated training data is included; Transmitting the selected trained model to the imaging device via the communication unit; Provisioning server.

10. The control unit is Before transmitting the trained model to the imaging device, recommendation data including information associated with the selected trained model is transmitted to the imaging device via the communication unit; When a request signal requesting transmission of a trained model is received in response to the recommendation data transmitted from the imaging device via the communication unit, the trained model is transmitted to the imaging device. The provision server according to claim 9.

11. the training data includes images; The control unit is Acquire, from an external terminal connected via the communication unit, a new trained model generated in the external terminal and information obtained in association with capturing an image included in training data used in the new trained model; The acquired new trained model is associated with information obtained in relation to the shooting of the image, and stored in the storage unit. The provision server according to claim 9.

12. The control unit is Acquire new learning data from an external terminal connected via the communication unit; Generate a new trained model using the training data; The generated new trained model is associated with information obtained in relation to capturing the image included in the new training data and stored in the storage unit. The provision server according to claim 9.

Images