LED inspection equipment, LED inspection system, imaging camera, LED inspection program
The LED inspection device and system automate LED state determination using AI models, adapting to model changes and reducing human error, ensuring consistent accuracy in LED inspection.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- NEC PLATFROMS LTD
- Filing Date
- 2025-03-21
- Publication Date
- 2026-06-02
Smart Images
Figure 0007868886000001_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to an LED (Light-Emitting Diode) inspection apparatus, an LED inspection system, an imaging camera, and an LED inspection program.
Background Art
[0002] Conventionally, the inspection of the lighting of LEDs in an apparatus equipped with LEDs has been visually confirmed by humans. In this case, since an operator has to inspect the LEDs of several apparatuses manufactured on a factory production line every day, there is a possibility of misjudging the lighting and extinguishing states due to fatigue or familiarity.
Prior Art Documents
Patent Documents
[0003]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0004] The following analysis was made by the present inventor.
[0005] Patent Document 1 describes a method for inspecting LEDs without relying on human vision. In this method, the inspection of the lighting and extinguishing states of a plurality of LEDs included in an inspection target apparatus is performed by guiding the light from each LED through an optical fiber assigned to each LED and imaging these lights.
[0006] However, although the number and arrangement of LEDs in the inspection target apparatus may vary depending on the model, in the method described in Patent Document 1, it is necessary to change the number and arrangement of optical fibers according to the model, and it has been difficult to quickly respond to changes in the model.
[0007] The objective of this disclosure is to provide an LED inspection device, an LED inspection system, an imaging camera, and an LED inspection program that contribute to a rapid response to changes in the model of the equipment being inspected. [Means for solving the problem]
[0008] (1) From the first perspective of this disclosure, an LED inspection device for an inspection device having multiple LEDs is provided. The LED inspection device, • A memory unit that stores the model of the device to be inspected, the number of LEDs, the color of the LEDs, and the inspection items. • Image acquisition unit that acquires an image of the entire device under inspection. - A model identification unit identifies the model of the device to be inspected by inputting an image of the entire device into a model identification model. • A lighting state acquisition unit that acquires the lighting state of a specified inspection target device (hereinafter referred to as "specified inspection target device") by inputting an image of the entire specified inspection target device, in which the LEDs are lit according to the specified inspection items, into a lighting detection model, and A determination unit determines the inspection result for the specified inspection target device by comparing the on / off state obtained for the specified inspection target device with the correct on / off state for the predetermined inspection item. including It is characterized by the following. (2) In a second perspective of the present disclosure, an LED inspection system is provided which includes the LED inspection device of the present disclosure and a camera for imaging the device to be inspected. (3) From a third perspective of this disclosure, a camera is provided which incorporates the LED inspection device of this disclosure and images the device to be inspected. (4) From the fourth perspective of this disclosure, a program is provided for inspecting a device to be inspected that has multiple LEDs. The aforementioned program is installed on the computer. • A process to acquire an image of the entire device to be inspected. • A process of inputting the acquired image into a model identification system to identify the model of the device to be inspected. • A process to obtain the on / off state of the LEDs of the specified device by inputting an image of the entire device, in which the LEDs are lit according to the specified inspection items of the specified device, into a lighting detection model, and - A process to determine the inspection result for the specified inspection target device by comparing the acquired on / off state with the correct on / off state for the predetermined inspection item. Make it run. [Effects of the Invention]
[0009] This disclosure, or any of its perspectives, can contribute to LED inspection that enables rapid response to changes in the model of the equipment being inspected. [Brief explanation of the drawing]
[0010] [Figure 1] A functional configuration diagram of an example of an LED inspection device related to this disclosure. [Figure 2] An example of an LED inspection system including the LED inspection device relating to this disclosure (A), and an example of a camera incorporating the LED inspection device relating to this disclosure (B). [Figure 3] A flowchart illustrating an example of the operation of the LED inspection device related to this disclosure. [Figure 4] A functional configuration diagram of another example of the LED inspection system related to this disclosure. [Figure 5] An example of what the monitor screen displays. [Figure 6] An example of displaying the coordinate information of the LEDs of the device being inspected, as stored by the LED inspection device. [Figure 7] An example of how the keypad area is displayed when the LED illumination status of the device being inspected, as determined by the LED inspection device, is shown. [Figure 8] An example of how the numeric keypad area is displayed when the results of comparing the LED inspection items with the correct answer data are shown. [Figure 9] An example of hardware resource configuration. [Modes for carrying out the invention]
[0011] Preferred embodiments of the present disclosure are shown below, but the present disclosure is not limited thereto. (Embodiment 1) Refer to the first perspective of the present disclosure above. (Embodiment 2) In the LED inspection apparatus according to Embodiment 1, it is preferable that the lighting detection model detects the coordinates of the LED candidates, the lighting / extinguishing state, and the lighting color when lit, from predetermined features in the input image. Preferably. (Embodiment 3) In the LED inspection apparatus according to Embodiment 2, it is preferable that the predetermined features are the shape and color in the image. Preferably. (Embodiment 4) In the LED inspection apparatus according to Embodiment 2, it is preferable that the image is a frame image of the video obtained by imaging the device to be inspected. Preferably. (Embodiment 5) In the LED inspection apparatus according to Embodiment 4, when the lighting state at the coordinates of the detected LED is detected in a certain number of frame images, the determination unit determines that the LED is lit, and when the extinguishing state at the coordinates of the detected LED is detected in a certain number of frame images, the determination unit determines that the LED is extinguished. Preferably. (Embodiment 6) In the LED inspection apparatus according to Embodiment 1, it is preferable that the model identification model is a learning model generated using, as the correct label, the model name and the images obtained by imaging each model as learning data. Preferably. (Embodiment 7) Refer to the second perspective of the present disclosure above. (Embodiment 8) Refer to the third perspective of the present disclosure above. (Embodiment 9) In the LED inspection system according to Embodiment 7, it is preferable that the LED inspection system includes a monitor that at least displays the inspection results for the specific device to be inspected. Preferably. (Embodiment 10) Refer to the fourth perspective of the present disclosure above.
[0012] Furthermore, this disclosure can also be embodied as a computer-executable program, which can be recorded on a computer-readable non-transitive storage medium. In other words, this disclosure can also be embodied as a computer program product. The program can be input to a computer device via an input device or an external communication interface, stored in a storage device, drive a processor according to predetermined steps or processes, and display the results of the processes step by step via a display device, including intermediate states as needed, or communicate with devices (including computers) inside or outside the device via a communication interface, whether wired or wireless. For this purpose, the computer device typically comprises, as an example, a processor, storage devices, input devices, a communication interface, and, if necessary, a display device, all of which are connected to each other by a bus.
[0013] The following is an overview of this disclosure. The reference numerals in the drawings included in this overview are solely for the purpose of aiding understanding this disclosure and are not intended to limit this disclosure to the illustrated configuration. Furthermore, connections between blocks in each drawing can be made by either wired or wireless means.
[0014] Furthermore, in the following descriptions and drawings, elements having the same or common function are denoted by the same reference numeral.
[0015] (Example of LED inspection device configuration) Figure 1 is a functional configuration diagram of an example of an LED inspection device related to this disclosure.
[0016] The LED inspection device 1 is a device for inspecting the lighting status of a device to be inspected that has multiple LEDs, and includes a storage unit 2, an image acquisition unit 3, a model identification unit 4, a lighting status acquisition unit 5, and a determination unit 6.
[0017] (Storage part) The memory unit 2 stores the model of the device under inspection, the number of LEDs, the LED lighting colors, and the inspection items entered by the user or other party. The device under inspection is a device equipped with multiple LEDs, such as a multifunction telephone (see, for example, the applicant's UNIVERGE DT900 / DT500 series), network equipment, or home appliances. The LEDs are used, for example, as backlights for the keypad of a multifunction telephone or illumination lights for function buttons. The number of LEDs and the LED lighting colors are usually determined for each model of the device under inspection.
[0018] The inspection items stored in memory unit 2 are one or more inspection items for each model of the device under inspection, for confirming the illumination of LEDs. For example, they consist of an item number (the number of the inspection item) and a combination of LEDs that should be illuminated for each item. By performing all inspection items for a given device under inspection, the illumination of all LEDs on that device is checked (however, in this case, some LEDs may be inspected more than once). If there is only one inspection item, all LEDs will be illuminated (or powered on) in a single inspection.
[0019] (Image acquisition unit) The image acquisition unit 3 acquires an image of the entire device under inspection. The image acquired is usually a video consisting of multiple sequentially consecutive frames, but it may also be one or more still images acquired in sequence.
[0020] Image acquisition can be performed by an external camera that forms a component of the LED inspection system together with the LED inspection device 1, by a camera built into the LED inspection device 1 (in this case, configured as so-called edge AI (Artificial Intelligence)), or by any other arbitrary manner. Figure 2(A) shows an example of an LED inspection system including an external camera 20, and Figure 2(B) shows an example of an imaging camera 21 built into the LED inspection device 1.
[0021] Furthermore, the memory unit 2 may be configured to also store various data such as detection results related to the AI model described later.
[0022] (Model identification section) The model identification unit 4 inputs an image of the entire device under inspection into the model identification model to identify the model of the device under inspection.
[0023] The model identification model is an AI model that uses a CNN (Convolutional Neural Network) for image recognition to identify the model of the device being inspected. More specifically, it is a trained model generated using images (each frame image) of each model as training data, with the model name as the correct label.
[0024] The images of the device under inspection used as training data are images taken of the entire device, and in this case, the device may be photographed from various angles for CNN model training. The device under inspection may also be photographed for several tens of seconds to several minutes. Furthermore, when photographing the device under inspection, the state of each LED is irrelevant. That is, all LEDs may be lit or unlit during imaging, or only one or more LEDs (i.e., not all LEDs) may be lit.
[0025] The device identification unit 4 uses each frame image of the captured device under inspection as training data, and stores the name of the device under inspection as the correct label for the training data, in a folder for the correct label in the storage unit 2 or other storage area. Furthermore, each time a new correct label and training data are added, the device identification unit 4 increases the output layer of the CNN model (device identification model) by one, performs retraining using the training data in the correct label folder, and enables the CNN model (device identification model) to identify newly added devices under inspection. By using such a retrained CNN model (device identification model), the device identification unit 4 can automatically determine the model of the device under inspection from the captured image (frame image) input from the image acquisition unit 3.
[0026] (Lighting status acquisition unit) The lighting status acquisition unit 5 inputs an image of the entire specified inspection target device (hereinafter referred to as "specified inspection target device"), in which the LEDs are lit according to predetermined inspection items of the specified inspection target device, which has been identified by the model identification unit 4, into the lighting detection model to acquire the lighting status of the specified inspection target device.
[0027] The lighting detection model is, for example, an AI model for object detection using bounding boxes (unsupervised machine learning (deep learning) model), which detects the coordinates or coordinate information of an LED, its on / off state, and the color of the LED when it is lit, from predetermined features in an input image (frame image), such as the shape and color in the image.
[0028] The coordinate information obtained from the lighting detection model during this detection process consists of the coordinate information of the bounding box in each frame image of the video, particularly the coordinates of two corners of the bounding box, such as the top-left and bottom-right coordinates.
[0029] Prior to inspecting the on / off state of the device under inspection, the lighting status acquisition unit 5 lights up all LEDs for each model of the device under inspection, and using the lighting detection model, acquires and stores the coordinate information of the LEDs for each color of the LEDs of the device under inspection stored in the storage unit 2, and continues the detection until it has stored the same amount of information as the number of LEDs of the device under inspection stored in the storage unit 2 (good product learning).
[0030] In this process, the LED coordinate information is stored for each frame image. Therefore, slight differences in LED flickering or brightness between frames can cause the same LED to be detected multiple times at slightly different coordinate positions. To prevent this, the lighting state acquisition unit 5 determines that the same LED has been detected if the center position of the bounding box coordinate information detected thereafter falls within a certain range relative to the center position of the bounding box coordinate information detected once.
[0031] Furthermore, when inspecting the on / off state of a specific device under inspection, the lighting state acquisition unit 5, in order to prevent false detection of the on / off state of an LED, determines that the LED at a given coordinate (information) is lit if it detects the LED corresponding to a given coordinate information stored in the memory lit up for a certain number of consecutive frames in the input video, and determines that the LED at that coordinate is off if it fails to detect the LED corresponding to the given coordinate information stored in the memory lit up for a certain number of consecutive frames.
[0032] Furthermore, the lighting status acquisition unit 5 determines whether the LED is lit or unlit, and by combining this with the color information of the detected lighting color, it determines whether the LED corresponding to the stored coordinate information (LED position) is lit or unlit in which color.
[0033] (Judgment Department) The determination unit 6 determines the inspection result for the specified inspection target device by comparing the acquired on / off state with the correct on / off state for a predetermined inspection item (for the specified inspection target device).
[0034] (Example of LED inspection device operation) Figure 3 shows a flowchart illustrating an example of the operation of the LED inspection device of this disclosure.
[0035] As shown in Figure 3, first, the LED inspection device 1 acquires an image of the entire device to be inspected (S1; where "S" stands for step).
[0036] Next, the LED inspection device 1 inputs an image of the entire device to be inspected into the model identification model to identify the model (S2).
[0037] Next, the LED inspection device 1 lights up the LEDs of the device(s) whose model has been identified ("specified device to be inspected") according to the inspection items of that device(s) (S3), and inputs the image of the LEDs ("inspection image") into the lighting detection model to obtain the actual lighting state ("actual state") of the device(s) (S4).
[0038] Next, the LED inspection device 1 compares the actual state of the specific inspection target device with the correct on / off state ("correct state") for the inspection item that was performed (S5).
[0039] If the actual state does not match the correct state (N in S6), the LED inspection device 1 determines that the specific inspection target device is unsuccessful (S7) and terminates.
[0040] If the actual state matches the correct state (Y in S6), the LED inspection device 1 determines that the specified inspection target device has passed (S9) and terminates if all inspection items for the specified inspection target device have been completed (Y in S8). If there are still inspection items remaining (N in S8), the process returns to S3.
[0041] In this way, the LED inspection device according to this disclosure can automatically detect the inspection target and perform LED inspection even if the inspection target device is changed midway through the process, and it can contribute to eliminating the need for human visual confirmation, suppressing human errors caused by fatigue and familiarity resulting from prolonged inspection work, and ensuring that a consistent level of inspection accuracy is always achieved.
[0042] Furthermore, the LED inspection device disclosed herein can also be used in systems that remotely monitor devices that indicate their status through color changes, such as LED illumination, including network equipment and home appliances, using camera footage, as well as in systems that perform inventory management and stocktaking by determining the presence or absence of items from color changes in camera footage.
[0043] (Another example of the LED inspection system related to this disclosure) Figure 4 is a functional configuration diagram of another example of the LED inspection system related to this disclosure.
[0044] The LED inspection system 100 includes a monitor 30 in addition to the LED inspection device 1 and external camera 20 described above.
[0045] The monitor 30 displays information such as how the LED inspection device 1 stores the coordinate information of the LEDs of the device under inspection, how the LED inspection device 1 determines the lighting status of the LEDs of the device under inspection, and how the LED inspection device 1 determines the result of the comparison with the correct data for the LED inspection item.
[0046] Figure 5 shows an example of the display on the monitor 30 screen. The screen 31 of the monitor 30 in the figure shows an example of a multifunction telephone 300, which includes 12 keypads 310 each with LEDs, 6 function buttons 320, and 1 illumination light 330. In this example of the multifunction telephone 300, the keypads 310 and function buttons 320 light up in different colors; in this example, the keypads 310 light up in orange and the function buttons 320 light up in green. The illumination light 330 lights up in a color that is pre-set according to the caller when a call is received. Although only an image of the multifunction telephone 300 is displayed on screen 31, other images may also be displayed.
[0047] Figure 6 shows an example of the display of LED coordinate information of the device under inspection (multifunction telephone) stored in the LED inspection device 1. However, in this figure, for the sake of simplicity of explanation, only the area of the keypad 310 (keypad area 32) of the multifunction telephone 300 is shown (the same applies to Figures 7 and 8).
[0048] Figure 6(A) shows an example of the display of the numeric keypad area 32 before the LED coordinate information is stored. In this case, only the numeric keypad is displayed. The white circles indicate numeric keypads 311 whose LEDs are not lit (the same applies to Figures 7 and 8).
[0049] Figure 6(B) shows an example of the display of the keypad area 32 when the LED inspection device 1 detects LEDs in a device under inspection (multifunction telephone) where all LEDs are lit, and bounding boxes are assigned to the coordinates of the detected LEDs. The black circles indicate the keypad 312 with lit LEDs, and the black rectangular frames surrounding each black circle are bounding boxes 400 (the same applies to Figures 7 and 8). The color of the bounding box 400 is displayed according to the color information of the detected LED.
[0050] Figure 6(C) shows an example of the display of the numeric keypad area 32 after the LED inspection device 1 has stored the coordinate information of all LEDs of the device under inspection, all LEDs have been turned off, and the bounding boxes 410 of the detected LED coordinates are displayed. In this case, the frame of the bounding box 410 is displayed in gray (the same applies to Figures 7 and 8).
[0051] Figure 7 shows an example of the display in the keypad area 32 when the LED lighting status of the device under inspection (multifunction telephone) as determined by the LED inspection device 1 is displayed.
[0052] In the example shown in Figure 7, a bounding box 400 matching the color of the illuminated key is displayed at the position of keypad 312, which is determined to be lit, and a bounding box 410 of a different color (gray frame) is displayed at the position of keypad 311, which is determined to be unlit.
[0053] Figure 8 shows an example of the display in the numeric keypad area 32 when the result of matching with the correct answer data for one of the multiple LED inspection items is displayed.
[0054] Figure 8(A) shows the case where the data matches the correct answer, and the test result for the test item is a pass. For example, the entire screen is surrounded by a blue frame (not shown), and the word "OK" is displayed in the upper left corner of the screen.
[0055] Figure 8(B) shows a case where the data does not match the correct answer. In this case, the test result for the inspection item will be a failure. For example, the entire screen will be surrounded by a red frame (not shown), and the word "NG" will be displayed in the upper left corner of the screen. In this case, it is also possible to display a dotted bounding box 420 (for example, red, matching the frame color) outside the bounding box of the mismatched LED so that the mismatched LED can be easily identified.
[0056] (Hardware resources) The above-mentioned LED inspection device can be configured using so-called hardware resources, and one with the configuration illustrated in Figure 9 can be used. For example, the hardware resources 1000 may include a processor 1001, memory 1002, network interface 1003, etc., which are interconnected by an internal bus 1004.
[0057] However, the configuration shown in Figure 9 is not intended to limit the hardware configuration of hardware resource 1000. Hardware resource 1000 may include hardware not shown (e.g., input / output interfaces). For example, the processor 1001 can be a CPU (Central Processing Unit), an MPU (Micro Processor Unit), a GPU (Graphics Processing Unit), etc.
[0058] Furthermore, the memory 1002 can be, for example, RAM (Random Access Memory), ROM (Read Only Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), etc. Here, the memory 1002 can store predetermined threshold values related to the predetermined physical quantities mentioned above, and can also store a control program for performing the above control.
[0059] Furthermore, the network interface 1003 can utilize, for example, a LAN (Local Area Network) card, a network adapter, a network interface card, or the like.
[0060] Furthermore, the functions of the hardware resource 1000 are realized by a processing module. This processing module is realized, for example, by the processor 1001 executing a program stored in memory 1002. This program can be downloaded via a network or updated using a storage medium containing the program. Moreover, the processing module may be realized by a semiconductor chip. In other words, the functions performed by the processing module can be realized by the execution of software on some hardware.
[0061] Some or all of the above embodiments may also be described as follows, but are not limited to the following: [Note 1] An LED inspection device for an inspection target device that has multiple LEDs. The LED inspection device, • A memory unit that stores the model of the device to be inspected, the number of LEDs, the color of the LEDs, and the inspection items. • Image acquisition unit that acquires an image of the entire device under inspection. - A model identification unit identifies the model of the device to be inspected by inputting an image of the entire device into a model identification model. • A lighting state acquisition unit that acquires the lighting state of a specified inspection target device (hereinafter referred to as "specified inspection target device") by inputting an image of the entire specified inspection target device, in which the LEDs are lit according to the specified inspection items, into a lighting detection model, and A determination unit determines the inspection result for the specified inspection target device by comparing the on / off state obtained for the specified inspection target device with the correct on / off state for the predetermined inspection item. Includes. [Note 2] In the LED inspection device described above, The aforementioned lighting detection model detects the coordinates of candidate LEDs, their on / off state, and the color of the LEDs when they are lit, based on predetermined features in the input image. [Note 3] In the LED inspection device described above, The aforementioned predetermined features are the shape and color in the image. [Note 4] In the LED inspection device described above, The image above is a frame image of video footage of the device being inspected. [Note 5] In the LED inspection device described above, The determination unit determines that an LED (candidate) is lit if the lit state at the detected coordinates (position) is detected in a certain number of frame images, and determines that an LED (candidate) is off if the off state at the detected coordinates (position) is detected in a certain number of frame images. [Note 6] In the LED inspection device described above, The aforementioned model identification model uses the model name as the correct label and is a learning model generated using images captured of each model as training data. [Note 7] An LED inspection system including the above-mentioned LED inspection device and a camera for imaging the device to be inspected. [Note 8] A camera that incorporates the above-mentioned LED inspection device and captures images of the device to be inspected. [Note 9] In the LED inspection system described above, The LED inspection system includes at least a monitor that displays the inspection results for the specified inspection target device. [Note 10] A program for inspecting a device that has multiple LEDs. The aforementioned program is installed on the computer. • A process to acquire an image of the entire device to be inspected. • A process of inputting the acquired image into a model identification system to identify the model of the device to be inspected. • A process to obtain the on / off state of the LEDs of the specified device by inputting an image of the entire device, in which the LEDs are lit according to the specified inspection items of the specified device, into a lighting detection model, and - A process to determine the inspection result for the specified inspection target device by comparing the acquired on / off state with the correct on / off state for the predetermined inspection item. Make it run. [Note 11] A method for inspecting a device that has multiple LEDs. In the above method, - A step of acquiring an image of the entire device to be inspected. The process of inputting the acquired image into a model identification system to identify the model of the device to be inspected, The process of obtaining the on / off state of the LEDs of the specified device by inputting an image of the entire device, in which the LEDs are lit according to the predetermined inspection items of the specified device, into a lighting detection model, and - A step of determining the inspection result for the specified inspection target device by comparing the acquired on / off state with the correct on / off state for the predetermined inspection item. This is executed (by a computer). [Note 12] In the LED inspection device described above, Prior to inspecting the on / off state of the device under inspection, the aforementioned lighting state acquisition unit inputs an image of all LEDs lit for each model of the device under inspection into the lighting detection model. For each LED color stored in the storage unit, it acquires and stores the coordinate information of the LEDs, and continues the detection until it has stored the same number of LEDs as the number of LEDs in that device under inspection stored in the storage unit. [Note 13] In the LED inspection device described above, The coordinate information obtained from the aforementioned lighting detection model is the coordinate information of the bounding box in each frame image of the video, in particular the coordinates of two corners of the bounding box, for example, the top-left and bottom-right coordinates. [Note 14] In the LED inspection device described above, The lighting status acquisition unit determines that the same LED has been detected if the center position of the coordinate information of a bounding box detected thereafter falls within a certain range relative to the center position of the coordinate information of the bounding box detected once. [Note 15] In the LED inspection device described above, The lighting status acquisition unit determines whether the LED is lit or extinguished, and by combining this with the color information of the detected lighting color, it determines whether the LED corresponding to the stored coordinate information (LED position) is lit or extinguished in a specific color.
[0062] Within the framework of the full disclosure of the present invention (including the claims), further modifications and adjustments to embodiments or examples are possible based on the fundamental technical concept. Furthermore, within the framework of the full disclosure of the present invention, various combinations or selections (including partial deletions) of various disclosed elements (including each element of each claim, each element of each embodiment or example, each element of each drawing, etc.) are possible. In other words, the present invention naturally includes various modifications and alterations that a person skilled in the art could make in accordance with the full disclosure, including the claims, and the technical concept. [Explanation of symbols]
[0063] 1 LED inspection device 2 Storage section 3 Image acquisition unit 4 Model identification section 5. Lighting status acquisition unit 6 Judgment section 10 LED Inspection Systems 20 External Cameras 21 Imaging camera 100 LED Inspection System 30 monitors 31 Monitor screen 300 multifunction phone 310 Numeric Keypad 320 Function Buttons 330 Illumination Lights 32 Numeric keypad area 311 Keypad (lights off) 312 Numeric keypad (illuminated) 400 bounding boxes (for lighting) 410 Bounding Box (for turning off lights) 420 Bounding Box (for malfunctions) 1000 hardware resources 1001 Processor 1002 memory 1003 Network Interface 1004 Internal Bus
Claims
1. An LED inspection device for an inspection target device having multiple LEDs, The aforementioned LED inspection device is - A memory unit that stores the model of the device to be inspected, the number of LEDs, the color of the LEDs, and the inspection items. - Image acquisition unit that acquires an image of the entire device to be inspected. - A model identification unit that identifies the model of the device to be inspected by inputting an image of the entire device into a model identification model. - A lighting state acquisition unit that acquires the lighting state of a specified inspection target device (hereinafter referred to as "specified inspection target device") by inputting an image of the entire specified inspection target device, in which the LEDs are lit according to the specified inspection items, into a lighting detection model, and A determination unit determines the inspection result for the specified inspection target device by comparing the on / off state obtained for the specified inspection target device with the correct on / off state for the predetermined inspection item. including An LED inspection device featuring the following characteristics.
2. In the LED inspection device according to claim 1, The aforementioned lighting detection model detects the coordinates of candidate LEDs, their on / off state, and the color of the LEDs when they are lit, from predetermined features in the input image. An LED inspection device featuring the following characteristics.
3. In the LED inspection device according to claim 2, The aforementioned predetermined features are the shape and color in the image. An LED inspection device featuring the following characteristics.
4. In the LED inspection device according to claim 2, The aforementioned image is a frame image of video footage of the device being inspected. An LED inspection device featuring the following characteristics.
5. In the LED inspection device according to claim 4, The determination unit determines that an LED is lit if the lit state at the detected LED coordinate is detected in a certain number of frame images, and determines that an LED is off if the off state at the detected LED coordinate is detected in a certain number of frame images. An LED inspection device featuring the following characteristics.
6. In the LED inspection device according to claim 1, The aforementioned model identification model uses the model name as the correct label and is a learning model generated using images captured of each model as training data. An LED inspection device featuring the following characteristics.
7. An LED inspection system comprising an LED inspection device according to any one of claims 1 to 6, and a camera for imaging the device to be inspected.
8. A camera that incorporates an LED inspection device according to any one of claims 1 to 6 and captures an image of the device to be inspected.
9. In the LED inspection system according to claim 7, The LED inspection system includes at least a monitor that displays the inspection results for the specified inspection target device. An LED inspection system featuring the following characteristics.
10. A program for inspecting a device having multiple LEDs, The aforementioned program is installed on the computer. - A process to acquire an image of the entire device to be inspected. - A process to identify the model of the device to be inspected by inputting the acquired image into the model identification model. - A process to obtain the on / off state of the specified device by inputting an image of the entire device, in which the LEDs are lit according to the specified inspection items of the specified device, into a lighting detection model, and, - A process to determine the inspection result for the specified inspection target device by comparing the acquired on / off status with the correct on / off status for the predetermined inspection item. Make it happen A program characterized by the following: