Control device, imaging apparatus, control method, and program

What is AI technical title?
AI technical title is built by Patsnap AI team. It summarizes the technical point description of the patent document.
The control device optimizes power management in imaging devices by allocating surplus power between driven means and analysis processing units, addressing the challenge of varying power consumption due to environmental conditions and usage status.

JP2025173802APending Publication Date: 2025-11-28CANON KK

View PDF 1 Cites 0 Cited by

Patent Information

Application Number: JP2024079576
Authority / Receiving Office: JP · JP
Patent Type: Applications
Current Assignee / Owner
Filing Date: 2024-05-15
Publication Date: 2025-11-28

AI Technical Summary

Technical Problem

Conventional imaging devices face challenges in managing power consumption between driven means and analysis processing, especially when power supply is limited, as the power consumption varies with environmental conditions and usage status, making it difficult to utilize surplus power effectively.

Method used

A control device and method that includes an analysis processing unit, driven means, and a power management unit to manage power consumption by determining surplus power for analysis processing based on supplied power, ensuring appropriate allocation between driven means and analysis processing units.

Benefits of technology

Enables effective management of power consumption in imaging devices, allowing for stable operation of driven means and analysis processing, even with limited power supplies, by optimizing power usage among different components.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure 2025173802000001_ABST

Patent Text Reader

Abstract

To properly manage power consumed by driving driven means and power consumed by analysis processing by analysis processing means when an imaging apparatus is supplied with power.SOLUTION: The control device includes: analysis processing units (a video analysis unit 105 and an AI processing unit 106) that analyse a video (image) obtained by an imaging unit 101 by using a learned model in a learned model storage unit 107; driven units (a heater 114, an LED 117, and an IRCF drive unit 118) that are driven according to a using situation of the imaging unit 101; and a power management unit 111 that manages first power consumed by driving the driven units and second power usable in analysis processing by the analysis processing units according to supplied power. The power management unit 111 determines the second power usable in analysis processing by the analysis processing unit from excessive power fixed based on the supplied power and the first power.SELECTED DRAWING: Figure 2

Need to check novelty before this filing date? Find Prior Art

Description

[Technical Field]

[0001] The present invention relates to a control device, an imaging device, a control method, and a program. [Background technology]

[0002] In recent years, network cameras have been developed that incorporate dedicated analytical processing means for executing AI (Artificial Intelligence) processing related to deep learning inference processing within the imaging device, and perform video analysis processing using a trained model within the imaging device. In this case, by performing video analysis processing at the edge of the imaging device, it becomes possible to generate object detection events and metadata of analysis results within the imaging device without sending the video to a cloud or server.

[0003] The network camera described above is often supplied with power from a PoE hub, a type of power supply device, and in this case, the amount of power supplied is limited by the PoE standard supported by the hub. In situations where the power supply to the network camera is limited, for example, in order to stably use the heater, infrared illumination light source, or other power-driven devices (driven means) possessed by the network camera, some ingenuity regarding power control is required. Patent Document 1 describes an imaging device that controls the drive of driven means such as a heater or infrared illumination light source so that the amount of voltage drop caused by the use of driven means falls within a certain range. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2023-110528 Summary of the Invention [Problem to be solved by the invention]

[0005] For example, the power consumption of the driven means in the imaging device described in Patent Document 1 is affected by the environment in which the imaging device is placed (the usage status of the imaging device). For example, in an environment where the temperature drops below freezing, a heater is driven to warm the imaging device, and in an outdoor environment with no lighting or in a store after closing time, an infrared illumination light source is driven to light up, so the power consumption of these driven means in the imaging device changes.

[0006] On the other hand, in an imaging device equipped with an analysis processing means that performs AI processing, which is a type of video analysis processing, it is conceivable to operate the analysis processing means so that the power consumption of the analysis processing means does not exceed the surplus power when the above-mentioned driven means is used at maximum. In this case, depending on the usage conditions of the imaging device, even if there is actually surplus power that exceeds the power consumption of the analysis processing means, it may not be possible to use the surplus power for the operation of the analysis processing means.

[0007] In other words, with conventional technology, when an imaging device receives a supply of power, it is difficult to appropriately manage the power consumption consumed in driving the driven means and the power consumption consumed in the analysis processing of the analysis processing means.

[0008] The present invention has been made in consideration of such problems, and aims to enable, when power is supplied to an imaging device, to appropriately manage the power consumption consumed in driving the driven means and the power consumption consumed in the analysis processing of the analysis processing means. [Means for solving the problem]

[0009] The control device of the present invention comprises an analysis processing means that analyzes and processes images obtained by an imaging unit using a trained model, a driven means that is driven according to the usage status of the imaging unit, and a power management means that manages, according to the supplied power, a first power consumed in driving the driven means and a second power that can be used for the analysis processing of the analysis processing means, and the power management means determines the second power that can be used for the analysis processing of the analysis processing means from surplus power determined based on the supplied power and the first power. [Effects of the Invention]

[0010] According to the present invention, when power is supplied to an imaging device, it is possible to appropriately manage the power consumption consumed in driving the driven means and the power consumption consumed in the analysis processing of the analysis processing means. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a diagram illustrating an example of a schematic configuration of an imaging system according to a first embodiment. [Figure 2] 1 is a diagram illustrating an example of the functional configuration of a surveillance camera (imaging device) according to a first embodiment. [Figure 3] 1 is a diagram illustrating an example of a hardware configuration of a surveillance camera (imaging device) according to a first embodiment. [Figure 4] 5 is a flowchart showing an example of a processing procedure in a control method for a surveillance camera (imaging device) according to the first embodiment. [Figure 5] 10 is a flowchart illustrating an example of a detailed processing procedure of a power allocation process by the AI processing unit according to the first embodiment. [Figure 6A] FIG. 2 is a diagram illustrating an example of a relationship table between the standard, power class, and output power compatible with the HUB and the maximum input power received by the surveillance camera according to the first embodiment. [Figure 6B] FIG. 10 illustrates an example of a power consumption information table including power-driven devices of the surveillance camera according to the first embodiment. [Figure 6C]FIG. 10 illustrates an example of a power consumption information table indicating the maximum power consumption for each trained model according to the first embodiment. [Figure 6D] FIG. 10 is a diagram illustrating the first embodiment and is a table illustrating examples 1 to 5 of the supplied power, the base power consumption, the power consumption of the LED (IR LED), the power consumption of the heater and their total, surplus power, and usable trained models. [Figure 7] 5 is a flowchart illustrating a detailed example of a power update process for the power-driven device in step S106 of FIG. 4 according to the first embodiment. [Figure 8] FIG. 10 is a diagram illustrating an example of the functional configuration of a monitoring camera (imaging device) according to a second embodiment. [Figure 9] 10 is a flowchart illustrating an example of a detailed processing procedure of a power allocation process by an AI processing unit according to the second embodiment. [Figure 10] 13 is a flowchart illustrating an example of a detailed processing procedure of a power allocation process by an AI processing unit according to the third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, embodiments of the present invention will be described with reference to the drawings. Note that the embodiments of the present invention described below do not limit the scope of the claims, and not all of the combinations of features described in the embodiments of the present invention are necessarily essential to the means for solving the problems of the present invention.

[0013] (First embodiment) First, the first embodiment will be described.

[0014] 1 is a diagram showing an example of a schematic configuration of an imaging system 10 according to the first embodiment. The imaging system 10 includes a plurality of monitoring cameras (imaging devices) 100-1 to 100-3, a PoE HUB (power supply device) 200, a network 300, and a monitoring PC 400.

[0015] Surveillance cameras 100-1 to 100-3 are imaging devices according to this embodiment. Surveillance cameras 100-1 to 100-3 each have an analysis processing unit that uses a trained model to analyze video (interpreted as "images" in this embodiment, including not only moving images but also still images) captured by an imaging unit. Furthermore, surveillance cameras 100-1 to 100-3 each have a driven unit that is driven by power depending on the usage status of the imaging unit, and a power management unit that manages the power consumed by the analysis processing unit and the driven unit depending on the supplied power.

[0016] The PoE hub 200 is a power supply device that supplies power to each of the surveillance cameras 100-1 to 100-3.

[0017] The network 300 is a network in which the PoE hub 200 and the monitoring PC 400 are connected so as to be able to communicate with each other.

[0018] The monitoring PC 400 displays the images obtained by the monitoring cameras 100-1 to 100-3, sets the functions of the monitoring cameras 100-1 to 100-3, and receives metadata, events, and the like from the monitoring cameras 100-1 to 100-3. The monitoring PC 400 is composed of, for example, a PC main unit, a monitor, a keyboard, a mouse, and the like. The monitoring PC 400 is equipped with a web browser for connecting to and communicating with the monitoring cameras 100-1 to 100-3, for example for camera settings, and receives the images obtained by the monitoring cameras 100-1 to 100-3, displays them on the monitor screen, and performs recording. The monitoring PC 400 also has a video management application installed that receives metadata from the monitoring cameras 100-1 to 100-3, which is the result of analysis processing by an analysis processing unit, and saves and displays this metadata on the monitor screen.

[0019] In the following description, when the content common to each of the surveillance cameras 100-1 to 100-3 is described without specifying each of them, the surveillance cameras will simply be referred to as "surveillance camera 100."

[0020] FIG. 2 is a diagram showing an example of the functional configuration of a surveillance camera (imaging device) 100 according to the first embodiment. As shown in FIG. 2, the surveillance camera 100 has the functional configuration of an imaging unit 101, a video compression unit 102, a format conversion unit 103, a communication unit 104, a video analysis unit 105, and an AI (Artificial Intelligence) processing unit 106. Furthermore, as shown in FIG. 2, the surveillance camera 100 has the functional configuration of a trained model storage unit 107, a power consumption information storage unit 108, a metadata creation unit 109, a power control unit 110, a power management unit 111, a temperature control unit 112, a temperature sensor 113, and a heater 114. Furthermore, as shown in FIG. 2, the surveillance camera 100 has the functional configuration of an illuminance control unit 115, an illuminance sensor 116, an LED 117, an infrared cut filter (IRCF) driving unit 118, and a setting unit 119.

[0021] The imaging unit 101 converts light from a subject that is input to the lens into an electrical signal using an image sensor to generate video (in this embodiment, "image" is interpreted as including not only moving images but also still images). The video compression unit 102 performs compression encoding processing on the video obtained by the imaging unit 101. The format conversion unit 103 converts the video data compressed by the video compression unit 102 into a file format suitable for transmission to the network 300. The communication unit 104 transmits video, metadata, etc. to a monitoring PC 400 that is connected to the network 300 for communication, and receives setting information, etc. from the monitoring PC 400. The communication unit 104 also communicates with a PoE HUB (power supply device) 200.

[0022] The video analysis unit 105, in cooperation with the AI processing unit 106, analyzes video (interpreted as "images" in this embodiment, including not only moving images but also still images) obtained by the imaging unit 101 using the trained models stored in the trained model storage unit 107. The AI processing unit 106, in cooperation with the video analysis unit 105, analyzes video obtained by the imaging unit 101 using the trained models stored in the trained model storage unit 107 (performs AI processing related to inference processing of deep learning). The trained model storage unit 107 stores trained models used when the video analysis unit 105 and the AI processing unit 106 analyze video obtained by the imaging unit 101. Here, in this embodiment, it is assumed that the trained model storage unit 107 stores multiple trained models that require different power consumption when the video analysis unit 105 and the AI processing unit 106 analyze video obtained by the imaging unit 101. The power consumption information storage unit 108 stores information on the power consumption required for video analysis processing using each trained model stored in the trained model storage unit 107. The metadata creation unit 109 converts the video analysis processing results output by the video analysis unit 105 into a predetermined format to create metadata.

[0023] The power control unit 110 negotiates the power supply by communicating with a PoE HUB (power supply device) 200 using LLDP (Link Layer Discovery Protocol), for example, via the communication unit 104. The power control unit 110 then acquires the power supply from the PoE HUB 200. The power management unit 111 manages the power consumption of the heater 114, LED 117, and IRCF drive unit 118, as well as the power consumption of the video analysis unit 105 and AI processing unit 106, so that the total power consumption of the surveillance camera 100 does not exceed the power supply acquired from the PoE HUB 200.

[0024] Temperature control unit 112 controls the temperature of surveillance camera 100 by turning on / off the lighting drive of heater 114 based on the temperature measured by temperature sensor 113. Temperature sensor 113 is a sensor that measures the temperature inside surveillance camera 100. Heater 114 is a heater that heats the inside of surveillance camera 100 based on the control of temperature control unit 112.

[0025] Illuminance control unit 115 performs control to appropriately adjust the brightness of the video captured by surveillance camera 100. Specifically, it controls the illuminance of surveillance camera 100 by turning on / off the lighting drive of LED 117 based on the illuminance measured by illuminance sensor 116. Illuminance sensor 116 is a sensor that measures the illuminance of surveillance camera 100. LED 117 is a light source that illuminates the imaging range captured by surveillance camera 100 based on the control of illuminance control unit 115. Here, in this embodiment, LED 117 is an infrared illumination light source (IR LED). IRCF drive unit 118 drives an infrared cut filter to insert or remove it from the optical path of imaging unit 101 based on the control of illuminance control unit 115.

[0026] The setting unit 119 performs processing for setting various functions of the surveillance camera 100. For example, the setting unit 119 provides the monitoring PC 400 with a GUI for setting various functions of the surveillance camera 100.

[0027] Here, regarding the power management of the surveillance camera 100, the relationship between the power management unit 111, the temperature control unit 112, and the illuminance control unit 115 will be described.

[0028] First, regarding the power management of the surveillance camera 100, the relationship between the power management unit 111 and the temperature control unit 112 will be described. The temperature control unit 112 acquires the temperature inside the surveillance camera 100 from the temperature sensor 113, and if the acquired temperature value falls below a predetermined lower limit, it requests the power management unit 111 to use power to heat the inside of the surveillance camera 100 with the heater 114. When the temperature control unit 112 is permitted to use the power by the power management unit 111, it drives the heater 114 to turn on with the permitted power. Furthermore, if the temperature value inside the surveillance camera 100 acquired from the temperature sensor 113 exceeds a predetermined upper limit, the temperature control unit 112 drives the heater 114 to turn off and notifies the power management unit 111 of the return of the power.

[0029] Next, regarding the power management of the surveillance camera 100, the relationship between the power management unit 111 and the illuminance control unit 115 will be described. The illuminance control unit 115 acquires the illuminance of the surveillance camera 100 from the illuminance sensor 116, and if the acquired illuminance value falls below a predetermined lower limit, requests the power management unit 111 to use power. When the illuminance control unit 115 is permitted to use the power by the power management unit 111, it drives the LED (IR LED) 117 to turn on with the permitted power. Furthermore, if the illuminance value of the surveillance camera 100 acquired from the illuminance sensor 116 exceeds a predetermined upper limit, the illuminance control unit 115 drives the LED (IR LED) 117 to turn off and notifies the power management unit 111 of the return of the power. Note that before driving the LED (IR LED) 117 to turn on, the illuminance control unit 115 instructs the IRCF drive unit 118 to remove the infrared cut filter in the surveillance camera 100 from the optical path of the imaging unit 101. Furthermore, the illuminance control unit 115 drives the LED (IR LED) 117 to turn off, and then instructs the IRCF driving unit 118 to insert the infrared cut filter in the surveillance camera 100 into the optical path of the imaging unit 101.

[0030] Next, regarding power management of the surveillance camera 100, the relationship between the power management unit 111 and the video analysis unit 105 will be described. The video analysis unit 105 requests the power management unit 111 to execute power allocation processing. When the power management unit 111 receives a request to execute power allocation processing from the video analysis unit 105, it notifies the video analysis unit 105 of the surplus power that can be allocated. The video analysis unit 105 then requests the power management unit 111 to use power within the range of surplus power notified by the power management unit 111. The power management unit 111 confirms that the value of the power requested to be used by the video analysis unit 105 is within the range of surplus power, and notifies the video analysis unit 105 of permission to use the power. When the video analysis unit 105 receives permission to use the power from the power management unit 111, it loads the set trained model into the AI processing unit 106 and executes the video analysis processing.

[0031] Fig. 3 is a diagram showing an example of the hardware configuration of the surveillance camera (imaging device) 100 according to the first embodiment. In Fig. 3, the same components as those shown in Fig. 2 are denoted by the same reference numerals, and detailed description thereof will be omitted.

[0032] 3, surveillance camera 100 has a hardware configuration including a CPU 131, RAM 132, ROM 133, storage unit 134, operation unit 135, and display unit 136. Furthermore, surveillance camera 100 has a hardware configuration including an imaging unit 101, a communication unit 104, an analysis processing unit 137, a sensor unit 138, an electrically powered device (driven unit) 139, and a bus 140, as shown in FIG.

[0033] The CPU 131 executes programs stored in the ROM 133 and the storage unit 134 using various types of information stored in the ROM 133 and the storage unit 134, thereby comprehensively controlling the operation of the surveillance camera 100 and performing various processes. At this time, the CPU 131 loads the various types of information stored in the ROM 133 and the storage unit 134 and the programs stored in the ROM 133 and the storage unit 134 into the RAM 132, and executes the programs using the various types of information loaded into the RAM 132. The CPU 131 executes the programs loaded into the RAM 132, thereby realizing the video compression unit 102, format conversion unit 103, metadata creation unit 109, power control unit 110, power management unit 111, temperature control unit 112, illuminance control unit 115, and setting unit 119 shown in FIG. 2 .

[0034] The RAM 132 has an area for storing various information and programs loaded from the ROM 133 or the storage unit 134. The RAM 132 also has a work area used when the CPU 131 executes various processes. In this way, the RAM 132 can provide various areas as needed.

[0035] The ROM 133 stores information that does not need to be rewritten, such as the setting data and startup program of the surveillance camera 100.

[0036] The storage unit 134 is a large-capacity storage unit such as a hard disk drive. The storage unit 134 stores an operating system (OS), computer programs used by the CPU 131 to perform various controls and processes on the surveillance camera 100, and various pieces of information. The computer programs and various pieces of information stored in the storage unit 134 are loaded into the RAM 132 as appropriate under the control of the CPU 131, and are then processed by the CPU 131. The storage unit 134 also stores various pieces of information obtained by the CPU 131 performing various controls and processes. In this embodiment, the storage unit 134 includes the trained model storage unit 107 and the power consumption information storage unit 108 shown in FIG. 2.

[0037] The operation unit 135 is configured by a user interface such as a keyboard, a mouse, or a touch panel, and allows the user to input various instructions to the CPU 131 by operating it.

[0038] The display unit 136 is configured with, for example, a liquid crystal screen or a touch panel screen, and can display the processing results of the CPU 131, the status of the surveillance camera 100, and the like, using images and text.

[0039] The analysis processing unit 137 is configured to include, for example, the video analysis unit 105 and the AI processing unit 106 in Fig. 2. The analysis processing unit 137 is an analysis processing means that analyzes the video (interpreted as "image" in this embodiment, including not only moving images but also still images) obtained by the imaging unit 101 using a trained model.

[0040] The sensor unit 138 includes the temperature sensor 113 and the illuminance sensor 116 shown in FIG.

[0041] The electrically driven device (driven unit) 139 includes the heater 114, the LED 117, and the IRCF driving unit 118 shown in Fig. 2. The electrically driven device 139 is a driven means that is driven by electric power depending on the usage status of the surveillance camera (imaging device) 100 including the imaging unit 101.

[0042] The bus 140 is a bus that communicatively connects the CPU 131, RAM 132, ROM 133, memory unit 134, operation unit 135, display unit 136, imaging unit 101, communication unit 104, analysis processing unit 137, sensor unit 138, and power-driven device 139.

[0043] The power management unit 111 in FIG. 2, which is realized by the CPU 131 executing a program, is a power management means that manages the first power consumed by driving the power-driven device 139 and the second power that can be used for the analysis processing of the analysis processing unit 137 according to the supplied power.

[0044] In addition, in this embodiment, the configuration of the surveillance camera (imaging device) 100 shown in Figures 2 and 3 excluding the imaging unit 101 (which may include the video compression unit 102 and format conversion unit 103 in Figure 2) is defined as the "control device" of this embodiment.

[0045] Fig. 4 is a flowchart showing an example of a processing procedure in a control method for the surveillance camera (imaging device) 100 according to the first embodiment. The flowchart shown in Fig. 4 is realized, for example, by the CPU 131 executing a program expanded in the RAM 132 of Fig. 3.

[0046] First, in step S101 of FIG. 4, the power control unit 110 communicates with the PoE hub (power supply device) 200 via the communication unit 104 to negotiate the power supply and receives the amount of power supplied from the PoE hub 200. FIG. 6A illustrates the first embodiment and shows an example of a relationship table 610 showing the standard, power class, and output power corresponding to the hub, and the maximum input power received by the surveillance camera 100. The relationship table 610 shown in FIG. 6A is stored in the storage unit 134. As shown in the relationship table 610 of FIG. 6A, the surveillance camera 100 can receive a maximum input power of 12.95 W from a PoE-standard hub and a maximum input power of 25.5 W from a PoE+-standard hub. Note that, depending on the hub settings, the maximum input power may be limited to 12.95 W even for a PoE+-standard hub. 4, it is assumed that the amount of power received from the PoE HUB 200 is limited to 12.95 W. In this case, the power control unit 110 notifies the power management unit 111 that the power supplied from the PoE HUB 200 is 12.95 W.

[0047] Next, in step S102 of FIG. 4, the power management unit 111 receives a power allocation request from the temperature control unit 112 and the illuminance control unit 115. FIG. 6B illustrates a first embodiment and is a diagram showing an example of a power consumption information table 620 including the power-driven device 139 of the surveillance camera 100. The power consumption information table 620 shown in FIG. 6B is stored in the storage unit 134 (e.g., the power consumption information storage unit 108). In FIG. 6B, "Base" represents only operations related to the basic functions of the surveillance camera 100, such as capturing and encoding video, and video distribution, and indicates a state in which the power-driven device 139 installed in the surveillance camera 100 is not driven and its power consumption is 5 W. FIG. 6B also illustrates that the power consumption when the LED (IR LED) 117 included in the power-driven device 139 is turned on is 4 W. FIG. 6B also illustrates that the power consumption when the heater 114 included in the power-driven device 139 is turned on is 3 W. 6B further shows that the power consumption for driving the IRCF driving unit 118 included in the power-driven device 139 is 1.5 W. Note that, because the driving of the infrared cut filter by the IRCF driving unit 118 is instantaneous, in this embodiment, it is excluded from the calculation of total power consumption by controlling it exclusively with the lighting driving of the LED (IR LED) 117. Here, assuming that there is a request from the temperature control unit 112 to allocate power to the heater 114 in step S102 of FIG. 4, the power management unit 111 permits the temperature control unit 112 to use 3 W of power.

[0048] Next, in step S103 of FIG. 4, the power management unit 111 determines whether or not there is a power allocation request from the video analysis unit 105 to the AI processing unit 106.

[0049] In step S103 of FIG. 4, if the power management unit 111 determines that there is a power allocation request from the AI processing unit 106 (S103 / Yes), the process proceeds to step S104. 4, the power management unit 111 executes power allocation processing for the AI processing unit 106. Here, the detailed processing of step S104 in FIG. 4 will be described with reference to FIG.

[0050] Fig. 5 is a flowchart illustrating a detailed example of the power allocation process of the AI processing unit 106 according to the first embodiment. The flowchart illustrated in Fig. 5 is implemented, for example, by the CPU 131 executing a program loaded in the RAM 132 in Fig. 3.

[0051] 5 starts, first, in step S201 in Fig. 5, the power management unit 111 notifies the video analysis unit 105 of the surplus power that can be allocated. In the example of this embodiment, the power supplied from the PoE HUB 200 is 12.95 W, the base power consumption of the surveillance camera 100 is 5 W, and the power consumption of the heater 114 is 3 W, so the surplus power is 4.95 W.

[0052] Next, in step S202 of Fig. 5, the video analysis unit 105 refers to the power consumption information stored in the power consumption information storage unit 108 to predict the amount of power consumption for each trained model. Fig. 6C shows the first embodiment and is a diagram illustrating an example of a power consumption information table 630 indicating the maximum power consumption for each trained model. The power consumption information table 630 shown in Fig. 6C is stored in the power consumption information storage unit 108. Here, as an example, consider a trained model that can detect three types of objects, namely, people, cars, and motorcycles, from video captured by the imaging unit 101. In the power consumption information table 630 shown in Fig. 6C, a trained model with a model ID of 3 is is a trained model with high detection accuracy but high power consumption, and the trained model with model ID 1 In the power consumption information table 630 shown in FIG. 6C, the trained model with model ID 2 has medium detection accuracy and low power consumption. The difference in processing power among these three types of trained models is due to differences in the structure of the trained model, the depth of the CNN (Convolutional Neural Network) layers, the number of nodes in each layer, etc. Specifically, in the example shown in Figure 6C, the trained model with model ID 1 has the lowest processing power and a maximum power consumption of 2W. ,The trained model with model ID 2 has medium processing power and a maximum power consumption of 4W, The trained model of 3 has the highest processing power and a maximum power consumption of 6W.

[0053] Next, in step S203 of Figure 5, the video analysis unit 105 determines whether there is a trained model that can be used based on the relationship between the surplus power obtained in step S201 and the power consumption for each trained model predicted in step S202.

[0054] In step S203 of FIG. 5, if the video analysis unit 105 determines that there is a usable trained model (S203 / Yes), the process proceeds to step S204. FIG. 6D shows the first embodiment and is a diagram illustrating a table 640 illustrating examples 1 to 5 of the supplied power, the base power consumption, the power consumption of the LED (IR LED) 117, the power consumption of the heater 114, and their total, surplus power, and usable trained models. The example of this embodiment corresponds to example 2 of FIG. 6D, since the supplied power is 12.95 W, the power consumption of the LED (IR LED) 117 is 0 W because it is turned off, and the power consumption of the heater 114 is 3 W because it is turned on. In the case of example 2 of FIG. 6D, the surplus power is 4.95 W, so the trained model with model ID 1 and the trained model with model ID 2 shown in FIG. 6C can be used, and therefore, step S204 is performed. Step S203 / Yes is returned, and the process proceeds to step S204.

[0055] When the process proceeds to step S204 in FIG. 5, the video analysis unit 105 notifies the power management unit 111 of the model with the highest performance (highest power consumption) and model ID of 2 from among the available trained models. Requires a power budget of 4W to run the trained model.

[0056] 5, the power management unit 111 refers to the surplus power obtained in step S201, and since the power allocation requested in step S204 is feasible, the power management unit 111 permits the power allocation. As a result, the surplus power managed by the power management unit 111 is reduced to 0.95 W.

[0057] Next, in step S206 of FIG. 5, the video analysis unit 105 performs power allocation on the trained model for which power allocation was requested in step S204 (in this embodiment, the trained model with model ID 2). ) is read into the AI processing unit 106 to perform video analysis processing.

[0058] Also, in step S203 of FIG. 5, if the video analysis unit 105 determines that there is no usable trained model (S203 / No), the process proceeds to step S207. In step S207 of FIG. 5, the video analysis unit 105 notifies the power management unit 111 of the end of the power allocation process.

[0059] When the process of step S206 in FIG. 5 is completed, or when the process of step S207 in FIG. 5 is completed, the process of the flowchart in FIG. 5 is completed, and the process proceeds to the next processing step in FIG.

[0060] Here, we return to the explanation of FIG. 4 is completed (the processing of the flowchart in FIG. 5 is completed), the process proceeds to step S105. Also, in step S103 in FIG. 4, if the power management unit 111 determines that there is no power allocation request from the AI processing unit 106 (S103 / No), the process proceeds to step S105. In step S105 of FIG. 4, the power management unit 111 determines whether or not there is a new power allocation request from the temperature control unit 112 or the illuminance control unit 115 that controls the power-driven device 139.

[0061] In step S105 of FIG. 4, if the power management unit 111 determines that there is a new power allocation request from the temperature control unit 112 or the illuminance control unit 115 that controls the power-driven device 139 (S105 / Yes), the process proceeds to step S106. 4, the power management unit 111 executes a power update process for the power-driven device 139. Here, the detailed process of step S106 in FIG. 4 will be described with reference to FIG.

[0062] Fig. 7 is a flowchart showing a first embodiment, illustrating an example of a detailed processing procedure of the power update processing of the power-driven device 139 in step S106 in Fig. 4. The flowchart shown in Fig. 7 is realized, for example, by the CPU 131 executing a program expanded in the RAM 132 in Fig. 3.

[0063] 7 starts, first, in step S301 of FIG. 7, the power management unit 111 estimates the surplus power that can be newly allocated to the power-driven device 139. Here, it is assumed that the illuminance control unit 115 has made a new power allocation request for driving the LED (IR LED) 117. In this case, referring to FIG. 6B, because 4 W of power is required to drive the LED (IR LED) 117, the estimated surplus power results in a power shortage of 3.05 W.

[0064] Next, in step S302 of FIG. 7, the power management unit 111 determines whether or not there is sufficient power without changing the video analysis process by the AI processing unit 106.

[0065] In step S302 of FIG. 7, if the power management unit 111 determines that there is sufficient power without changing the video analysis process by the AI processing unit 106 (S302 / Yes), the process proceeds to step S303. In step S303 of FIG. 7, the power management unit 111 permits the temperature control unit 112 and the illuminance control unit 115, which control the power-driven device 139, to use power.

[0066] 7, if the power management unit 111 determines that there will be insufficient power unless the video analysis process by the AI processing unit 106 is changed (S302 / No), the process proceeds to step S304. In the example of the present embodiment described above, a power shortage of 3.05 W occurs, so the result is step S302 / No, and the process proceeds to step S304. 7, the power management unit 111 instructs the video analysis unit 105 to stop using power. Then, upon receiving this instruction, the video analysis unit 105 stops the AI processing by the AI processing unit 106. After that, when the AI processing unit 106 has completely stopped the AI processing, the video analysis unit 105 notifies the power management unit 111 that the AI processing has been stopped and that power has been returned, and requests the start of new power allocation processing. As a result, the surplus power managed by the power management unit 111 increases from 0.95W to 4.95W.

[0067] As described above, stopping the AI processing by the AI processing unit 106 restores the surplus power to 4.95 W, making it possible to drive the power-driven device 139 (specifically, in this embodiment, to light the LED (IR LED) 117). Therefore, in this embodiment, in step S305 of FIG. 7, the power management unit 111 permits the illuminance control unit 115 to use power. As a result, the illuminance control unit 115 lights the LED (IR LED) 117. This example corresponds to example 4 in FIG. 6D, and the surplus power is reduced to 0.95 W.

[0068] Next, in step S306 of FIG. 7, the power management unit 111 executes the power allocation process of the AI processing unit 106 again. The detailed process of this step S306 is the same as the process of the flowchart shown in FIG. 5 described above. In the example of this embodiment, the surplus power is 0.95 W, while even the trained model with the model ID of 1, which has the smallest maximum power consumption, consumes 2 W (see FIG. 6C), there is no trained model that can be used. In this case, the result is step S203 / No in Fig. 5, and in the following step S207, the video analysis unit 105 notifies the power management unit 111 of the end of the power allocation process.

[0069] When the process of step S303 in FIG. 7 is completed, or when the process of step S306 in FIG. 7 is completed, the process of the flowchart in FIG. 7 is completed, and the process proceeds to the next processing step in FIG.

[0070] Here, we return to the explanation of FIG. 4 is completed (the processing of the flowchart in FIG. 7 is completed), the process proceeds to step S107. Also, in step S105 in FIG. 4, if the power management unit 111 determines that there is no new power allocation request from the temperature control unit 112 or the illuminance control unit 115 that controls the power-driven device 139 (S105 / No), the process proceeds to step S107. In step S107 of FIG. 4, the power management unit 111 determines whether or not a power return notification has been received from the temperature control unit 112 or the illuminance control unit 115 that controls the power-driven device 139.

[0071] In step S107 of FIG. 4, if the power management unit 111 determines that it has received a power return notification from the temperature control unit 112 or the illuminance control unit 115 that controls the power-driven device 139 (S107 / Yes), the process proceeds to step S108. 4, the power management unit 111 increases the surplus power by the amount of power returned in the power return notification in step S107. Here, it is assumed that the temperature control unit 112 turns off the heater 114 and notifies the power management unit 111 of the return of power. In this case, the power consumed by the heater 114, 3 W (see FIG. 6B), is returned, and the surplus power managed by the power management unit 111 is restored to 3.95 W.

[0072] Next, in step S109 of FIG. 4, the power management unit 111 executes the power allocation process of the AI processing unit 106 again. The detailed process of this step S109 is the same as the process of the flowchart shown in FIG. 5 described above. In the example of this embodiment, the power consumption allocated to the heater 114 has been returned, so the surplus power has increased to 3.95 W, and the trained model with a model ID of 1 and a maximum power consumption of 2 W (see FIG. 6C) is now available for use. This case corresponds to Example 3 in FIG. 6D, and in FIG. 5, step S203 / Yes is selected. In the following step S204 in FIG. 5, the video analysis unit 105 requests 2 W of power from the power management unit 111. In this case, in the following step S205 in FIG. 5, the power management unit 111 allows the allocation of this power. Then, in the following step S206 in FIG. 5, the video analysis unit 105 loads the trained model with model ID 1 into the AI processing unit 106. As a result, the surplus power managed by the power management unit 111 becomes 1.95W.

[0073] 4 is completed (the processing of the flowchart in FIG. 5 is completed), the process proceeds to step S110. Also, in step S107 in FIG. 4, if the power management unit 111 determines that it has not received a power return notification from the temperature control unit 112 or the illuminance control unit 115 that controls the power-driven device 139 (S107 / No), the process proceeds to step S110. When the process proceeds to step S110 in FIG. 4, the power control unit 110 communicates with the PoE HUB (power supply device) 200 via the communication unit 104 to determine whether or not the supply power has been changed.

[0074] In step S110 of FIG. 4, if the power control unit 110 determines that the supplied power has been changed (S110 / Yes), the process proceeds to step S111. 4, the power management unit 111 performs processing to revoke all power usage permissions for the temperature control unit 112 and illuminance control unit 115 that control the power-driven device 139, and the video analysis unit 105 that controls the AI processing unit 106. Specifically, in step S111, the temperature control unit 112 turns off the heater 114, the illuminance control unit 115 turns off the LED (IR LED) 117, and the video analysis unit 105 stops the video analysis processing by the AI processing unit 106. In this case, the temperature control unit 112, the illuminance control unit 115, and the video analysis unit 105 return a stop response to the power management unit 111 and then request a new power allocation. When the processing of step S111 in FIG. 4 is completed, the processing returns to step S102 in FIG. 4, and power allocation processing is performed based on the new value of supplied power.

[0075] Furthermore, in step S110 of FIG. 4, if the power control unit 110 determines that the supplied power has not been changed (S110 / No), the process returns to step S103 and the processes from step S103 onwards are performed again.

[0076] By performing the processing of the flowchart in FIG. 4 , the power management unit 111 performs management to reduce the second power available for the analysis processing of the analysis processing unit 137 when the first power consumed in driving the power-driven device 139 increases based on the supplied power. Furthermore, the power management unit 111 performs management to increase the second power available for the analysis processing of the analysis processing unit 137 when the first power consumed in driving the power-driven device 139 decreases based on the supplied power. Then, when the power management unit 111 reduces the second power, the analysis processing unit 137 may set a trained model that requires less power consumption than when the second power is increased as the trained model that actually performs the analysis processing of the video. Furthermore, when the power management unit 111 increases the second power, the analysis processing unit 137 may set a trained model that requires more power consumption than when the second power is decreased as the trained model that actually performs the analysis processing of the video.

[0077] The surveillance camera (imaging device) 100 of the first embodiment described above includes an analysis processing unit 137, which corresponds to analysis processing means that analyzes video (interpreted as "images" in this embodiment, including not only moving images but also still images) captured by the imaging unit 101 using a trained model. The surveillance camera (imaging device) 100 of the first embodiment also includes a power-driven device 139, which corresponds to driven means that is driven by power depending on the usage status of the imaging unit 101. The surveillance camera (imaging device) 100 of the first embodiment also includes a power management unit 111, which corresponds to power management means that manages, depending on the supplied power, a first power consumed to drive the power-driven device 139 and a second power available for analysis processing by the analysis processing unit 137. The power management unit 111 then determines the second power available for analysis processing by the analysis processing unit 137 from surplus power determined based on the supplied power and the first power. In the first embodiment, the "control device" obtained by removing the imaging unit 101 (which may include the video compression unit 102 and the format conversion unit 103) from the surveillance camera 100 also has the above-mentioned analysis processing unit 137, power-driven device 139, and power management unit 111. With this configuration, when the surveillance camera (imaging device) 100 receives a supply of power, the power consumption consumed by driving the power-driven device 139 and the power consumption consumed by the analysis processing of the analysis processing unit 137 can be appropriately managed.

[0078] Furthermore, in the first embodiment, the analysis processing unit 137 selects a trained model that will actually perform video analysis processing from among multiple trained models requiring different power consumption so that the power consumption falls within the second power range. In this case, if there are multiple trained models that fall within the second power range, the analysis processing unit 137 selects the trained model requiring the greatest power consumption as the trained model that will actually perform video analysis processing. According to this configuration, a trained model that actually performs the video analysis process is selected from among multiple trained models that require different power consumption depending on the surplus power, making it possible to perform the video analysis process according to the supplied power and the usage status of the power-driven device 139.

[0079] (Second embodiment) Next, a second embodiment will be described. In the following description of the second embodiment, matters common to the first embodiment will be omitted, and only matters different from the first embodiment will be described.

[0080] The schematic configuration of the imaging system according to the second embodiment is similar to the schematic configuration of the imaging system 10 according to the first embodiment shown in FIG.

[0081] Fig. 8 is a diagram showing an example of the functional configuration of a surveillance camera (imaging device) 100 according to the second embodiment. In Fig. 8, the same components as those shown in Fig. 2 are denoted by the same reference numerals, and detailed description thereof will be omitted.

[0082] The surveillance camera (imaging device) 100 according to the second embodiment shown in FIG. 8 has a configuration in which a detection application unit 120 is added to the functional configuration of the surveillance camera (imaging device) 100 according to the first embodiment shown in FIG. 2.

[0083] 8 is a component that is installed or pre-installed and operable in, for example, the surveillance camera 100 (or the above-mentioned "control device"). The video analysis unit 105 performs video analysis processing in accordance with instructions from the detection application unit 120, and executes object detection processing.

[0084] The detection application unit 120 receives parameters related to object detection, such as the type and area of object detection and the action to be taken when an event occurs, from the monitoring PC 400 connected to the network 300 via the communication unit 104. For example, the detection application unit 120 generates an event when a person enters the set area, and sets the event to display an alert message on the monitoring screen displayed on the monitoring PC 400.

[0085] Furthermore, the detection application unit 120 requests a list of trained models for person detection that can be executed from the video analysis unit 105 in accordance with user settings. The detection application unit 120 then selects a trained model from the list received from the video analysis unit 105 and instructs the video analysis unit 105 to execute the selected trained model. In accordance with the instruction from the detection application unit 120, the video analysis unit 105 acquires a trained model for person detection from the trained model storage unit 107, and loads the acquired trained model into the AI processing unit 106 to execute person detection processing.

[0086] The control method for the surveillance camera (imaging device) 100 according to the second embodiment is similar to the processing procedure in the control method for the surveillance camera (imaging device) 100 according to the first embodiment shown in Fig. 4. In the second embodiment, the detailed processing content of the AI power allocation processing (power allocation processing of the AI processing unit 106) in steps S104 and S109 in Fig. 4 and step S306 in Fig. 7 differs from that in the first embodiment.

[0087] 9 is a flowchart illustrating a second embodiment, showing an example of a detailed processing procedure of power allocation processing by the AI processing unit 106. The flowchart shown in Fig. 9 is implemented, for example, by the CPU 131 executing a program loaded in the RAM 132 in Fig. 3.

[0088] The processing in steps S401 to S403 in FIG. 9 is the same as the processing in steps S201 to S203 in FIG. 5, and therefore a description thereof will be omitted.

[0089] In step S403 of FIG. 9, if the video analysis unit 105 determines that there is a trained model that can be used (S403 / Yes), the process proceeds to step S404.

[0090] Proceeding to step S404 in FIG. 9, the video analysis unit 105 notifies the detection application unit 120 of usable trained models. Here, if the power supplied from the PoE HUB 200 is 12.95 W and the heater 114 and the LED (IR LED) 117 are off, the surplus power is 7.95 W (corresponding to example 1 in FIG. 6D). In this case, as shown in example 1 in FIG. 6D, usable trained models are all trained models with model IDs 1 to 3, based on the relationship of the maximum power consumption shown in FIG. 6C. Therefore, the video analysis unit 105 notifies the detection application unit 120 of the trained models with model IDs 1 to 3 as usable trained models.

[0091] 9, the detection application unit 120 selects the trained model with the highest detection accuracy and the highest power consumption, which has a model ID of 3, from among the available trained models notified in step S404. Then, the detection application unit 120 instructs the video analysis unit 105 to execute the selected trained model with a model ID of 3.

[0092] Next, in step S406 of FIG. 9, the video analysis unit 105 requests the power management unit 111 to allocate 6 W of power to execute the trained model with model ID 3. .

[0093] Next, in step S407 of FIG. 9, the power management unit 111 refers to the surplus power obtained in step S401, and since the power allocation requested in step S406 is feasible, the power allocation is permitted.

[0094] Next, in step S408 of FIG. 5, the video analysis unit 105 performs power allocation on the trained model for which power allocation was requested in step S406 (in this embodiment, the trained model with model ID 3). ) is read into the AI processing unit 106 to perform video analysis processing.

[0095] 9, if the video analysis unit 105 determines that there is no available trained model (S403 / No), the process proceeds to step S409. For example, if the heater 114 and the LED (IR LED) 117 are on, the surplus power is 0.95 W (see example 4 in FIG. 6D), so the result is step S403 / No and the process proceeds to step S409. When the process proceeds to step S409 in FIG. 9, the video analysis unit 105 notifies the detection application unit 120 that the AI processing unit 106 has stopped performing AI processing (for example, object detection processing).

[0096] Next, in step S410 of FIG. 9, the detection application unit 120 displays an alert message on the setting screen or monitoring screen of the monitoring PC 400 via the communication unit 104 indicating that AI processing (e.g., object detection processing) cannot be performed.

[0097] When the process of step S408 in FIG. 9 is completed, or when the process of step S410 in FIG. 9 is completed, the process of the flowchart in FIG. 9 is completed, and the process proceeds to the next process step in FIG. 4, etc.

[0098] The surveillance camera 100 of the second embodiment described above includes a detection application unit 120, which is a means for performing processing to detect an object from a video (image) and corresponds to a detection means for selecting a trained model from a plurality of trained models that actually performs video analysis processing. Note that in the second embodiment, the above-described detection application unit 120 is also included in a "control device" obtained by excluding the imaging unit 101 (which may include the video compression unit 102 and the format conversion unit 103) from the surveillance camera 100. Then, in the second embodiment, the analysis processing unit 137, which includes the video analysis unit 105 and the AI processing unit 106, sets the trained model selected by the detection application unit 120 as the trained model that actually performs video analysis processing. With this configuration, it is possible to execute video analysis processing according to the power supply and the usage status of the power-driven device 139 in accordance with instructions from the detection application unit 120.

[0099] (Third embodiment) Next, a third embodiment will be described. In the following description of the third embodiment, matters common to the first and second embodiments will be omitted, and only matters different from the first and second embodiments will be described.

[0100] The schematic configuration of the imaging system according to the third embodiment is similar to the schematic configuration of the imaging system 10 according to the first embodiment shown in Fig. 1. The functional configuration of the surveillance camera (imaging device) 100 according to the third embodiment is similar to the functional configuration of the surveillance camera (imaging device) 100 according to the second embodiment shown in Fig. 8.

[0101] In the second embodiment, the detection application unit 120 selects the trained model, but in the third embodiment, the video analysis unit 105, rather than the detection application unit 120, selects the trained model, which is different from the second embodiment.

[0102] In the third embodiment, the detection application unit 120 requests the video analysis unit 105 to perform analysis processing on the video (image) required for the detection content set by the monitoring PC 400 via the communication unit 104. Then, the video analysis unit 105 selects a trained model in response to the request from the detection application unit 120 and notifies the detection application unit 120 of the selected trained model. In addition, the detection application unit 120 receives the analysis processing result of the video (image) from the video analysis unit 105 and determines the detection event in accordance with the selected detection content.

[0103] Furthermore, in the third embodiment, the detailed processing content of the AI power allocation processing (power allocation processing of the AI processing unit 106) described above differs from that in the second embodiment.

[0104] Fig. 10 is a flowchart illustrating a third embodiment, showing an example of a detailed processing procedure of power allocation processing by the AI processing unit 106. The flowchart shown in Fig. 10 is implemented, for example, by the CPU 131 executing a program loaded in the RAM 132 in Fig. 3.

[0105] The processing in steps S501 to S506 in FIG. 10 is the same as the processing in steps S201 to S205 in FIG. 5, and therefore a description thereof will be omitted.

[0106] When the process of step S506 in FIG. 10 is completed, the process proceeds to step S507. When the process proceeds to step S507 in FIG. 10, the video analysis unit 105 notifies the detection application unit 120 of the trained model that will actually perform the video analysis process.

[0107] Next, in step S508 of Figure 10, the detection application unit 120 displays the currently running trained model and the contents of the video analysis process (AI processing) on the setting screen or monitoring screen of the monitoring PC 400 via the communication unit 104.

[0108] Also, in step S503 of FIG. 10, if the video analysis unit 105 determines that there is no usable trained model (S503 / No), the process proceeds to step S509. When the process proceeds to step S509 in FIG. 10, the video analysis unit 105 notifies the detection application unit 120 that the AI processing unit 106 has stopped performing AI processing (for example, object detection processing).

[0109] Next, in step S510 of FIG. 10, the detection application unit 120 displays an alert message on the setting screen or monitoring screen of the monitoring PC 400 via the communication unit 104 indicating that AI processing (e.g., object detection processing) cannot be performed.

[0110] When the process of step S508 in FIG. 10 is completed, or when the process of step S510 in FIG. 10 is completed, the process of the flowchart in FIG. 10 is completed, and the process proceeds to the next process step in FIG. 4, etc.

[0111] In the third embodiment, by notifying the detection application unit 120 of a trained model that performs video (image) analysis processing, the detection application unit 120 can execute the detection processing with power appropriate for the situation, while still allowing the user to understand the status of the detection processing. Furthermore, in the third embodiment, the video analysis unit 105 selects a trained model, making it possible to support multiple detection applications. For example, even in a case where a parking detection application and a person tracking application are run simultaneously, the video analysis unit 105 can select a trained model that detects people and cars that supports both detections.

[0112] (Other embodiments) In the above-described embodiment, a trained model is selected based on the power available for AI processing by the AI processing unit 106. However, the resolution of the video (image) analyzed by the AI processing may also be changed. In this case, information on the power consumption for each processing resolution is prepared for each trained model, and the detection application unit 120 selects the trained model and the resolution of the input video (image) according to the detection settings. For example, when the detection application unit 120 wants to detect a person that appears small on the screen, it prioritizes the resolution of the video (image) and switches to a trained model with lower power consumption. Also, for example, when the size of the subject on the screen is sufficient, the detection application unit 120 prioritizes the accuracy of the trained model and reduces the resolution of the input video (image). In this way, by considering the detection priority of the detection application unit 120, it is possible to perform video (image) analysis processing that is more suited to the situation.

[0113] Furthermore, in the above-described embodiment, the trained model loaded into the AI processing unit 106 is pre-stored in the trained model storage unit 107 inside the surveillance camera (imaging device) 100. However, the present invention is not limited to this configuration. For example, the trained model may be stored in a server, which is an external device connected to the network 300, and the surveillance camera (imaging device) 100 may download and acquire the trained model from the server when actually executing the surveillance camera (imaging device) 100 and execute it. This embodiment reduces the amount of storage space required inside the surveillance camera (imaging device) 100. Preferably, the trained model and information on power consumption when the analysis processing unit 137 performs video analysis processing using the trained model are also stored in the external server. In this case, the surveillance camera (imaging device) 100 preferably acquires the above-described power consumption information when downloading and acquiring the trained model from the server when actually executing the surveillance camera (imaging device) 100. In this manner, by acquiring the trained model and power consumption information from the external server as needed and performing an update process, it is possible to execute the latest trained model on the server while managing power consumption.

[0114] In addition to selecting a trained model, it may be possible to select to have the CPU 131 execute a video (image) analysis process using the trained model. When there is not enough surplus power to execute the video analysis process in the analysis processing unit 137 including the video analysis unit 105 and the AI processing unit 106, a configuration may be adopted in which the CPU 131 executes the process using the trained model. In this configuration, although the execution speed and accuracy of the process using the trained model decrease, it becomes possible to execute the detection process without stopping the video analysis process.

[0115] The present invention can also be realized by supplying a program that realizes one or more functions of the above-described embodiments to a system or device via a network or a storage medium, and having one or more processors in the computer of the system or device read and execute the program.The present invention can also be realized by a circuit (e.g., ASIC) that realizes one or more functions. This program and a computer-readable storage medium storing the program are included in the present invention.

[0116] It should be noted that the above-described embodiments of the present invention are merely illustrative examples of the implementation of the present invention, and the technical scope of the present invention should not be construed as being limited by these. In other words, the present invention can be implemented in various forms without departing from its technical concept or main features.

[0117] The disclosure of this embodiment includes the following configuration, method, and program. [Configuration 1] an analysis processing means for analyzing an image obtained by the imaging unit using the trained model; a driven means that is driven in accordance with the usage status of the imaging unit; a power management means for managing a first power consumed in the driving of the driven means and a second power available for the analysis processing of the analysis processing means according to the supplied power; and The power management means determines the second power available for the analysis processing of the analysis processing means from surplus power determined based on the supplied power and the first power. A control device characterized by: [Configuration 2] The trained model includes a plurality of trained models that require different power consumption when the analysis processing means performs the analysis processing, The analysis processing means selects a trained model that actually performs the analysis processing from among the plurality of trained models so that the trained model falls within the second power range. 2. The control device according to configuration 1, [Configuration 3] When there are a plurality of trained models that fall within the second power range, the trained model that requires the greatest power consumption is set as the trained model that actually performs the analysis processing. 3. The control device according to configuration 2. [Configuration 4] The trained model includes a plurality of trained models that require different power consumption when the analysis processing means performs the analysis processing, the power management means performs management such that when the first power increases, the second power is decreased, and when the first power decreases, the second power is increased; The analysis processing means When the power management means reduces the second power, a trained model that requires less power consumption than when the power management means increases the second power is set as a trained model that actually performs the analysis processing; When the power management means increases the second power, a trained model that requires greater power consumption than when the power management means decreases the second power is set as a trained model that actually performs the analysis processing. 4. The control device according to any one of configurations 1 to 3. [Configuration 5] The trained model includes a plurality of trained models that require different power consumption when the analysis processing means performs the analysis processing, A means for performing processing to detect an object from the image, further comprising: a detection means for selecting a trained model that actually performs the analysis processing from among the plurality of trained models so that the trained model falls within the second power range; The analysis processing means sets the trained model selected by the detection means as the trained model that actually performs the analysis processing. 5. The control device according to any one of configurations 1 to 4. [Configuration 6] The image processing device further includes a detecting means for performing a process to detect an object from the image, the detecting means displaying the details of the analysis process performed by the analysis processing means on a screen. 5. The control device according to any one of configurations 1 to 4. [Configuration 7] Acquire information on the trained model and power consumption when the analysis processing means performs the analysis processing using the trained model from an external device. 7. The control device according to any one of configurations 1 to 6. [Configuration 8] Obtaining the supply power from a power supply device 8. The control device according to any one of configurations 1 to 7. [Configuration 9] A control device according to any one of configurations 1 to 8; the imaging unit; An imaging device comprising: [Method 1] an analysis processing step of analyzing an image obtained by the imaging unit using the trained model by an analysis processing means; a driven step of driving a driven means in accordance with a usage state of the imaging unit; a power management step of managing a first power consumed in the driving of the driven means and a second power available for the analysis processing of the analysis processing means according to the supplied power; and The power management step determines the second power available for the analysis processing of the analysis processing means from surplus power determined based on the supplied power and the first power. A control method comprising: [Program 1] an analysis processing step of analyzing an image obtained by the imaging unit using the trained model by an analysis processing means; a driven step of driving a driven means in accordance with a usage state of the imaging unit; a power management step of managing a first power consumed in the driving of the driven means and a second power available for the analysis processing of the analysis processing means according to the supplied power; on the computer, The power management step determines the second power available for the analysis processing of the analysis processing means from surplus power determined based on the supplied power and the first power. A program characterized by: [Explanation of symbols]

[0118] 100-1 to 100-3: surveillance cameras (imaging devices), 101: imaging unit, 102: video compression unit, 103: format conversion unit, 104: communication unit, 105: video analysis unit, 106: AI processing unit, 107: learned model storage unit, 108: power consumption information storage unit, 109: metadata creation unit, 110: power control unit, 111: power management unit, 112: temperature control unit, 113: temperature sensor, 114: Heater, 115: Illuminance control unit, 116: Illuminance sensor, 117: LED, 118: IRCF driving unit, 119: Setting unit, 120: Detection application unit, 131: CPU, 132: RAM, 133: ROM, 134: Memory unit, 135: Operation unit, 136: Display unit, 137: Analysis processing unit, 138: Sensor unit, 139: Power-driven device (driven unit), 140: Bus, 200: PoE HUB (power supply device), 300: Network, 400: Monitoring PC

Claims

1. an analysis processing means for analyzing an image obtained by the imaging unit using the trained model; a driven means that is driven in accordance with the usage status of the imaging unit; a power management means for managing a first power consumed in the driving of the driven means and a second power available for the analysis processing of the analysis processing means in accordance with the supplied power; and The power management means determines the second power available for the analysis processing of the analysis processing means from surplus power determined based on the supplied power and the first power. A control device characterized by:

2. The trained model includes a plurality of trained models that require different power consumption when the analysis processing means performs the analysis processing, The analysis processing means selects a trained model that actually performs the analysis processing from among the plurality of trained models so that the trained model falls within the second power range.