Information identification method and device, medium and intelligent equipment

Abstract

The invention provides an information identification method, an electronic device, a computer readable medium and intelligent equipment, the intelligent equipment carries out picture identification based on a carried identification model, and the method comprises the following steps: obtaining an unread picture; the unread picture is a picture of which the target information cannot be recognized by using the recognition model; according to the no-reading reason of the no-reading picture, performing enhancement processing on the no-reading picture to obtain an enhanced picture; performing identification by using an identification model based on the enhanced picture to obtain truth value information; and optimizing the recognition capability of the recognition model by using a training sample formed by the unread picture and the truth value information. According to the invention, the intelligent device can automatically utilize the unread picture which fails in recognition to carry out optimization training on the recognition model, so that the adaptive capacity of the intelligent device to the production environment is improved in a targeted manner, the code reading recognition rate of the intelligent device is ensured, and the recognition efficiency of the intelligent device is improved. And the time that the user collects and feeds back unread picture samples by the intelligent equipment and controls the intelligent equipment to carry out firmware upgrading is saved, and the production line efficiency is ensured.

Description

Technical Field

[0001] The present invention relates to the field of machine vision technology, and in particular to an information recognition method, an electronic device, a computer-readable medium, and an intelligent device. Background Art

[0002] In barcode reading camera applications, the barcode reading environments vary across various industrial scenarios. The recognition performance of barcode reading products is based on the need to ensure normal barcode reading in most common scenarios. However, in some industries such as lithium batteries, 3C, and medical care, the target installation location, light distribution, dynamic barcode reading, and other necessary production environments will affect the barcode imaging in the captured image. For example, problems such as wrinkles and distortion of stripes and ink reflections will occur in the photo, which in turn affects the camera's barcode reading rate, making it difficult for the camera to meet actual usage needs. Summary of the Invention

[0003] The present invention aims to, to a certain extent, address one of the technical problems in the related art. To this end, the present invention provides an information recognition method, an electronic device, a computer-readable medium, and an intelligent device. The information recognition method can improve the intelligent device's adaptability to different production environments and ensure the intelligent device's code recognition rate.

[0004] To achieve the above objectives, as one aspect of the present invention, there is provided an information recognition method, which is applied to a smart device, wherein the smart device performs image recognition based on a recognition model installed therein; the method comprises: Obtaining a no-read picture; the no-read picture is a picture in which target information cannot be recognized using the recognition model; performing enhancement processing on the no-read picture according to a no-read reason of the no-read picture to obtain an enhanced picture; Performing recognition based on the enhanced image using the recognition model to obtain true value information; The recognition capability of the recognition model is optimized by using the training samples formed by the unread pictures and the true value information.

[0005] As an optional implementation manner of the present invention, the target information is the product number information corresponding to the identification code in the image.

[0006] Optionally, the identification code is a barcode, and the target information is barcode information corresponding to the barcode in the image.

[0007] Optionally, the recognition model is used to read the information parameters of each line in the recognition code, and obtain the target information according to the mapping relationship between the information parameters of the multiple lines based on the internal numerical labels of the recognition model; The internal number mark includes information parameters of a line in the identification code corresponding to each character of the target information, and the information parameters include position information and width information of the line.

[0008] As an optional implementation manner of the present invention, the step of enhancing the no-read picture according to the no-read reason of the no-read picture to obtain the enhanced picture includes: Different types of compensation enhancement are performed on the no-read picture for each no-read reason to obtain the enhanced picture corresponding to the no-read picture for each no-read reason.

[0009] As an optional implementation manner of the present invention, performing different types of compensation enhancement on the no-read picture for each no-read reason includes at least one of the following: For the image without image whose unreading reason is abnormal barcode image size, scaling the barcode image area of ​​the image without image so that the barcode image size is within a preset size range, thereby obtaining the corresponding enhanced image; For the image without image whose unreading reason is abnormal bar width in the barcode image, dilation or corrosion processing is performed on the barcode image area of ​​the image without image to make the bar width within the preset bar width range, thereby obtaining the corresponding enhanced image; For the image without image whose unreading reason is abnormal grayscale of barcode image, grayscale stretching processing is performed on the barcode image area of ​​the image without image so that the grayscale of the barcode image is within a preset grayscale range, thereby obtaining the corresponding enhanced image; For the image without image whose unreading reason is barcode image distortion, performing distortion correction processing on the image without image to obtain the corresponding enhanced image; For the image without image whose unreading reason is the missing of barcode quiet zone, the quiet zone part of the barcode in the image without image is supplemented to obtain the corresponding enhanced image; For the non-pictured picture whose unreading reason is the barcode broken needle, the broken needle part of the barcode in the non-pictured picture is supplemented to obtain the corresponding enhanced picture.

[0010] As an optional embodiment of the present invention, the information identification method further includes: Acquire imaging information of a plurality of the no-read pictures, and determine a no-read reason for each of the no-read pictures based on the imaging information of the plurality of no-read pictures; Among them, the imaging information includes at least one of the algorithm capability set matching, the contaminated area determination information, and the coding mode compatibility determination information; the algorithm capability set matching includes at least one of the image size of the barcode to be identified in the unread image, whether the barcode to be identified lacks a quiet area, whether a broken needle occurs, and the codeword length of the barcode to be identified.

[0011] As an optional embodiment of the present invention, the obtaining of the no-read image includes obtaining the no-read image from a no-read training library; The information identification method further includes: Storing the no-read images in the no-read training library; Removing the unread image corresponding to the enhanced image for which the true value information is identified from the unread training library; When the number of no-read images in the no-read training library is higher than a preset number threshold, the no-read images in the no-read training library are cleared.

[0012] As an optional embodiment of the present invention, the smart device performs image recognition based on the recognition model in the working mode, and stops image recognition in the sleep mode; The step of optimizing the recognition capability of the recognition model by using the training samples formed by the unread pictures and the true value information is performed when the smart device is in sleep mode.

[0013] Optionally, the method further includes: Based on the camera photo information within a preset time, determine whether the current smart device is in sleep mode.

[0014] Optionally, determining whether the current smart device is in sleep mode based on camera photo information within a preset time includes: When receiving at least one camera photo information within a preset time, it is determined that the current smart device is in working mode; If no camera photo information is received within a preset time, it is determined that the current smart device is in sleep mode.

[0015] As an optional embodiment of the present invention, the information identification method further includes: In response to receiving the picture to be identified, information identification is performed on the received picture to be identified using the current identification model.

[0016] As a second aspect of the present invention, an electronic device is provided, comprising: one or more processors; A memory having one or more computer programs stored thereon, wherein when the one or more computer programs are executed by the one or more processors, the one or more processors implement the aforementioned information identification method.

[0017] As a third aspect of the present invention, a computer-readable medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the information identification method described above is implemented.

[0018] As a fourth aspect of the present invention, a smart device is provided, which includes a photographing device and the aforementioned electronic device, wherein the electronic device is used to obtain a picture taken by the photographing device as a picture to be identified.

[0019] In the information recognition method, electronic device, computer-readable medium and smart device provided by the present invention, the smart device (camera end) can automatically enhance the no-read picture so that the enhanced picture obtained by enhancing at least part of the no-read picture can be successfully recognized by the recognition model as the target information (i.e., the true value information corresponding to the barcode), thereby establishing a mapping relationship between the no-read picture before enhancement and the target information, obtaining a training sample, and then using the training sample to optimize the recognition model for training, so that the optimized recognition model can recognize the target information corresponding to at least part of the no-read picture, thereby realizing real-time optimization of the recognition model for the no-read picture that failed to be recognized, and improving the optimization efficiency of the recognition model on the camera end.

[0020] The information recognition method provided by the present invention can automatically use the unread images that failed to be recognized to optimize the recognition model for training, and obtain a recognition model that can successfully recognize the target information from these unread images, thereby specifically improving the adaptability of the smart device to the production environment in which it is located and ensuring the code reading recognition rate of the smart device.

[0021] Moreover, the optimization training process is automatically completed by the camera end. Compared with the related technology that the design institute integrates the optimization model into the firmware and then upgrades the firmware of the smart device, it can effectively improve the recognition model optimization efficiency of the smart device and enhance the user experience. It also saves the user the process window time for the smart device to collect and feedback unread image samples and control the smart device to upgrade the firmware, thereby ensuring the production line efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The present invention will be further described below in conjunction with the accompanying drawings: Figure 1 is a flow chart of an information identification method provided by an embodiment of the present invention; Figure 2 is a flow chart of an information identification method provided by an embodiment of the present invention; Figure 3 is a flow chart of an information identification method provided by an embodiment of the present invention; Figure 4 Schematic diagram of the principle of the information identification method provided by an embodiment of the present invention; Figure 5 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention; Figure 6 It is a schematic diagram of a computer-readable medium provided by an embodiment of the present invention.

[0023] Description of reference numerals: 101: processor; 102: memory; 103: I / O interface; 104: bus. DETAILED DESCRIPTION

[0024] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described in the embodiments are intended to explain the present invention and are not to be construed as limiting the present invention.

[0025] References in this specification to "one embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment itself can be included in at least one embodiment disclosed herein. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment.

[0026] In related technology, users can collect photos taken with problematic cameras and send them back to the research institute for analysis, including those in which the identification code is not read or is missed. The research institute uses these unread photos as samples and conducts targeted retraining and optimization to obtain an optimized recognition model. Finally, the optimized recognition model is integrated and updated into the camera program. The camera then uses the new model to improve code reading performance through a firmware upgrade.

[0027] However, this method requires multiple interruptions to the camera's working scene when transmitting sample images and performing firmware upgrades. The upgrade operation is complex and inefficient. Moreover, for high-standard scenes, code reading optimization often requires multiple continuous improvements to meet usage requirements, seriously affecting user work efficiency and user experience.

[0028] In order to solve the above technical problems, as one aspect of the present invention, an information recognition method is provided, which is applied to a smart device. The smart device can be a smart camera, or an intelligent processing unit matched with a camera. The intelligent processing unit can be a separate hardware or integrated into the camera. The smart device performs image recognition based on the recognition model installed. Figure 1 As shown, the method includes: Step S1: Obtain a no-read image, where the no-read image is an image in which the target information cannot be recognized using the recognition model; Step S2: performing enhancement processing on the no-read picture according to the no-read reason of the no-read picture to obtain an enhanced picture; Step S3: using the recognition model to perform recognition based on the enhanced image to obtain true value information; Step S4: Optimize the recognition capability of the recognition model using the training samples formed by the unread images and the true value information.

[0029] Alternatively, the method provided by the present invention can be used in a camera barcode reading scenario, where the target information is the product number information corresponding to the identification code in the image. In embodiments of the present invention, the identification code read by the smart device from the image can be a barcode, and the target information is the barcode information corresponding to the barcode. Alternatively, the identification code can be a color barcode, a QR code, or other pattern code.

[0030] It can be understood that enhancement processing refers to adjusting the image of the unreadable picture to make the difficult-to-recognize patterns in the unreadable picture easier to recognize. For example, when the identification code image is too dark or too bright, the grayscale of the picture is adjusted; when the identification code image is blurred, the picture is corroded; when some unimportant features of the barcode in the picture are missing, the barcode image is completed, and so on.

[0031] In an embodiment of the present invention, the camera end can automatically perform enhancement processing on the no-read picture so that the enhanced picture obtained by enhancing at least part of the no-read picture can be successfully recognized by the recognition model as the target information (i.e., the true value information corresponding to the barcode), thereby establishing a mapping relationship between the no-read picture before the enhancement processing and the target information, obtaining a training sample, and then using the training sample to optimize the recognition model for training, so that the optimized recognition model can recognize the target information corresponding to at least part of the no-read picture, thereby realizing real-time optimization of the recognition model for the no-read picture that failed to be recognized, and improving the optimization efficiency of the recognition model on the camera end.

[0032] The information recognition method provided by the present invention can automatically use the unread images that failed to be recognized to optimize the recognition model for training, and obtain an optimized recognition model that can successfully recognize the target information from these unread images, thereby specifically improving the adaptability of the smart device to the production environment in which it is located and ensuring the code reading recognition rate of the smart device.

[0033] Moreover, the optimization training process is automatically completed by the camera end. Compared with the related technology that the design institute integrates the optimization model into the firmware and then upgrades the firmware of the smart device, it can effectively improve the recognition model optimization efficiency of the smart device and enhance the user experience. It also saves the user the process window time for the smart device to collect and feedback unread image samples and control the smart device to upgrade the firmware, thereby ensuring the production line efficiency.

[0034] It is understood that the smart device in the present invention may be a smart camera. That is, the information recognition method provided by the present invention is executed by the internal system of the barcode reading camera itself. The photo captured by the camera's camera assembly is transmitted to the internal system of the barcode reading camera as the image to be recognized. The internal system then performs image recognition based on the onboard recognition model to determine the target information (e.g., product number information) corresponding to the identification code image in the image to be recognized. Alternatively, the smart device may be an intelligent processing unit used in conjunction with a camera. The intelligent processing unit may be a separate piece of hardware. For example, the intelligent processing unit may be located in a different location from the camera in the factory. The camera transmits the image to be recognized to the intelligent processing unit via a cable or wireless transmission, and the intelligent processing unit performs image recognition based on the recognition model it carries out. Alternatively, the intelligent processing unit may also be integrated into the camera, that is, the intelligent processing unit is fixed inside the camera or detachably installed in a corresponding installation slot of the camera, and is directly connected to the camera's photographing component for local communication.

[0035] As an optional embodiment of the present invention, Figure 2 As shown, the information identification method also includes: Step S5: In response to receiving the image to be identified, the current recognition model is used to perform information recognition on the received image to be identified. That is, after the optimized recognition model is trained, the optimized recognition model replaces the original recognition model as the current recognition model to identify subsequent received images to be identified.

[0036] Optionally, the recognition model is specifically used to read information parameters corresponding to multiple lines in the recognition code, and obtain target information based on the mapping relationship between the information parameters of the multiple lines and the internal numerical labels of the recognition model; Each character of the internal numerical mark includes target information corresponding to information parameters of a line in the identification code, and the information parameters include position information and width information of the line.

[0037] To further improve the optimization efficiency of the recognition model, the method provided in the embodiment of the present invention can perform different modes of compensation enhancement on the no-read images with different no-read reasons according to different types of no-read reasons. That is, step S2 can specifically include: Different types of compensation enhancement are performed on the no-read images with each no-read reason to obtain enhanced images corresponding to the no-read images with each no-read reason.

[0038] In an embodiment of the present invention, a smart device (such as a smart camera) can automatically determine the no-read reason of a no-read image and adopt different compensation enhancement methods for no-read images with different no-read reasons, so that the enhanced images obtained by compensating and enhancing the no-read images with various no-read reasons can be more easily identified, thereby ensuring the sample number of training samples and further ensuring the success rate of code reading of the smart device (camera).

[0039] As an optional embodiment of the present invention, the no-read reason may include at least one of abnormal barcode image size (i.e., the size of the barcode portion is too large or too small), abnormal bar width (i.e., the width of each barcode line is too wide or too narrow), abnormal barcode image grayscale (i.e., the barcode image is too bright or too dark, resulting in recognition failure), barcode image distortion (i.e., the barcode image is distorted due to distortion), missing barcode quiet area (i.e., the barcode quiet area is missing the corresponding pattern due to occlusion or reflection), and broken barcode needle (i.e., part of the line pattern is missing). Accordingly, the steps of performing different types of compensation enhancement on the no-read image for each no-read reason may specifically include at least one of the following: For a non-image image whose unreadable reason is an abnormal barcode image size, the barcode image area of ​​the non-image image is scaled (i.e., enlarged or reduced) so that the barcode image size is within a preset size range, thereby obtaining a corresponding enhanced image; For a non-image image whose unreading reason is an abnormal bar width in the barcode image, the barcode image area of ​​the non-image image is expanded or eroded to make the bar width within the preset bar width range, thereby obtaining a corresponding enhanced image; For the image without image that cannot be read due to abnormal grayscale of barcode image, grayscale stretching processing is performed on the barcode image area of ​​the image without image (i.e., the grayscale value of the barcode image area is adjusted) so that the grayscale of the barcode image is within the preset grayscale range, and the corresponding enhanced image is obtained; For the image without image whose unreading reason is barcode image distortion, the image without image is subjected to distortion correction processing to obtain the corresponding enhanced image; For a non-imaged image whose unread reason is the missing quiet zone of the barcode, the quiet zone of the barcode in the non-imaged image is supplemented (i.e., the quiet zone of the image is supplemented using an image with the same quiet zone pattern as other images), and the corresponding enhanced image is obtained; For a non-image image whose unreading reason is a barcode pin break, the broken pin part of the barcode in the non-image image is completed to obtain the corresponding enhanced image.

[0040] As an optional embodiment of the present invention, the reason for no reading may also include abnormal codeword length and incompatible coding mode; the information identification method further includes: Acquire imaging information of a plurality of no-read images, and determine a no-read reason for each no-read image based on the imaging information of the plurality of no-read images; Among them, the imaging information includes at least one of the algorithm capability set matching, the contaminated area determination information, and the coding mode compatibility determination information. The algorithm capability set matching includes at least one of the image size of the barcode to be identified in the unread image (for example, the length of the barcode to be identified or the number of pixels corresponding to the barcode to be identified in the image), whether the barcode to be identified lacks a quiet area, whether a broken needle occurs, and the codeword length of the barcode to be identified.

[0041] It can be understood that the damaged area judgment information refers to the judgment result of the camera end on whether the lost area can be repaired when the identification code is blocked by dirt or damaged and missing. For example, when the damaged area is located on the positioning point of the QR code or in the common part of the barcode (such as the quiet areas on the left and right ends), a compensation enhancement method can be used to complete the damaged area.

[0042] Coding mode compatibility determination information refers to whether the camera can identify at least some of the true value information corresponding to the lines in the barcode after recognizing them. If any part of the lines in the barcode in the unread image cannot be successfully read, the coding mode of the barcode in the image is incompatible with this camera. The unread image can be discarded, or maintenance personnel can manually supplement the coding mode of the recognition model.

[0043] Optionally, the information recognition method further includes expanding the recognition model in step S4 when there is a no-read picture whose no-read reason is abnormal codeword length.

[0044] Optionally, the information identification method further includes: When there are unreadable images due to incompatible coding modes, a coding mode supplement prompt is generated so that maintenance personnel can add a new coding mode to the recognition model in a timely manner based on this prompt.

[0045] As an optional embodiment of the present invention, the smart device performs image recognition based on the onboard recognition model in the working mode, and stops image recognition in the sleep mode; Step S4 is performed when the smart device is in sleep mode. That is, model optimization training is performed only when the smart device is not operating (for example, when the production line is paused or undergoing maintenance, and the smart device is not taking photos of goods). This allows for the optimal allocation of the smart device's processor resources to ensure optimal performance.

[0046] As an optional embodiment of the present invention, obtaining a no-read image in step S1 specifically includes obtaining a no-read image from a no-read training library; The information identification method further includes: Store the no-read images in the no-read training library; Remove the unreadable images corresponding to the enhanced images with true value information from the unreadable training database That is, as the recognition model is gradually optimized, the no-read images that can be recognized in the no-read training library are gradually eliminated accordingly. Ideally, the recognition model can eventually recognize the target information of all no-read images in the original no-read training library, that is, the no-read training library is cleared.

[0047] As an optional embodiment of the present invention, the information identification method further includes: When the number of read-free images in the read-free training library exceeds a preset threshold, the read-free images in the read-free training library are cleared to prevent excessive read-free images that cannot be recognized from affecting the normal use of the camera.

[0048] For example, when the maximum number of images that can be stored in the no-read training library is 10,000, the preset number threshold can be set to 500, that is, when the number of no-read images in the no-read training library is higher than 500, the camera automatically clears the no-read training library.

[0049] It can be understood that steps S1 to S4 are repeatable steps. Whenever a picture that cannot be recognized by the recognition model of the smart device appears, the smart device can automatically add the picture as a no-read picture to the no-read training library, enhance the no-read pictures collected in the no-read training library, and identify the true value information. Each time the smart device is in a non-working mode (such as a sleep mode) or in a periodic maintenance time, the recognition model is optimized and trained using a training sample consisting of no-read pictures and true value information. The optimized recognition model obtained after each execution of step S4 is automatically switched as the recognition model for subsequent recognition pictures, so that Figure 4 As shown, each execution of step S4 is a further optimization training based on the optimization result generated in the previous cycle. After the recognition model obtained by each optimization training takes effect, the code reading success rate of the smart device will increase and the no-read rate will decrease.

[0050] Optionally, the method further includes: Based on the camera photo information within a preset time, it is determined whether the current smart device is in sleep mode. When the camera device in the smart device performs a photo taking action, it generates camera photo information. The embodiment of the present invention uses the camera photo information as a trigger signal to determine in real time whether the smart device has entered sleep mode.

[0051] Optionally, determining whether the current smart device is in sleep mode based on camera photo information within a preset time includes: When receiving at least one camera photo information within a preset time, it is determined that the current smart device is in working mode; If no camera photo information is received within a preset time, it is determined that the current smart device is in sleep mode.

[0052] That is, in the embodiment of the present invention, when the smart device does not take a photo within a preset time, it is determined that the smart device is dormant, and then the step of executing model optimization training is started.

[0053] Optionally, the method further includes the step of outputting target information after the recognition model successfully recognizes the image to be recognized. Specifically, the method further includes: Step S51: formatting the target information identified by the recognition model to generate an output signal corresponding to the target information; Step S52: output the output signal to a host computer (eg, an industrial computer).

[0054] That is, the smart device performs different actions when the code reading recognition result is successful or failed in the working mode, such as Figure 3 As shown in the figure, after the recognition model successfully reads the code, the target information it reads is sent to the upper-level application for formatting, assembled according to the protocol required by the customer, and then transmitted as an output signal, thereby realizing barcode information recognition for each product on the production line. If the code reading fails, the image is stored in the no-read training library to be used as a training sample for optimizing the recognition model in sleep mode.

[0055] As a second aspect of the present invention, there is provided an electronic device, such as Figure 5 As shown, the electronic device includes: One or more processors 101; The memory 102 stores one or more computer programs. When the one or more computer programs are executed by the one or more processors 101, the one or more processors 101 implement the information identification method provided by the embodiment of the present invention.

[0056] The electronic device provided by the present invention can automatically collect and store no-read images that fail to be recognized into a no-read training library, and use the no-read training library to optimize and train the recognition model to obtain a recognition model that can successfully recognize target information from these no-read images, thereby specifically improving the adaptability of smart devices to the production environment in which they are located and ensuring the code reading recognition rate of smart devices.

[0057] Moreover, the optimization training process is automatically completed by the electronic device installed on the camera end, so that the smart device can optimize the recognition model in real time for the no-read images that fail to be recognized. Compared with the solution in related technologies that the design institute integrates the optimization model into the firmware and then upgrades the firmware of the smart device, it can effectively improve the recognition model optimization efficiency of the smart device, enhance the user experience, and save the user's process window time for the smart device to collect and feedback no-read image samples and control the smart device to upgrade the firmware, thereby ensuring production line efficiency.

[0058] It is understood that the electronic device can be an integral part of the barcode reader camera, for example, as a system module or a detachable intelligent processing unit within the barcode reader camera, directly communicating with the camera's image capture component. Alternatively, the electronic device can be used as a separate piece of hardware, for example, located in a separate location within the factory from the camera, and receiving images to be recognized from the camera via cable or wireless transmission.

[0059] Alternatively, as Figure 5 As shown, the electronic device may further include one or more I / O interfaces 103 connected between the processor 101 and the memory 102 and configured to implement information interaction between the processor 101 and the memory 102 .

[0060] Among them, the processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 102 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically such as SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read-write interface) is connected between the processor and the memory, and can realize information exchange between the processor and the memory, including but not limited to a data bus (Bus), etc.

[0061] In some embodiments, the processor 101 , the memory 102 , and the I / O interface 103 are connected to each other via a bus 104 , and further connected to other components of the computing device.

[0062] As a third aspect of the present invention, a computer readable medium is provided. Figure 6 As shown, a computer program is stored thereon, and when the computer program is executed by the processor, the above information identification method is implemented.

[0063] The computer-readable medium provided by the present invention stores a computer program for executing the information recognition method provided by the present invention. The information recognition method can automatically collect and store no-read images that fail to be recognized in a no-read training library, and use the no-read training library to optimize and train the recognition model to obtain a recognition model that can successfully recognize the target information from these no-read images, thereby specifically improving the adaptability of smart devices to the production environment in which they are located and ensuring the code reading recognition rate of smart devices.

[0064] Moreover, the optimization training process is automatically completed by the camera end, so that the smart device can optimize the recognition model in real time for the no-read images that fail to be recognized. Compared with the solution in related technologies where the design institute integrates the optimization model into the firmware and then upgrades the firmware of the smart device, it can effectively improve the recognition model optimization efficiency of the smart device, enhance the user experience, and save the user the process window time for the smart device to collect and feedback no-read image samples and control the smart device to upgrade the firmware, thereby ensuring production line efficiency.

[0065] As a fourth aspect of the present invention, a smart device (smart camera) is provided. The smart device includes a photographing device and a front electronic device, and the electronic device is used to obtain a picture taken by the photographing device as a picture to be identified.

[0066] The intelligent device provided by the present invention can automatically collect and store no-read images that fail to be recognized into a no-read training library, and use the no-read training library to optimize and train the recognition model to obtain a recognition model that can successfully recognize target information from these no-read images, thereby specifically improving the adaptability of the intelligent device to the production environment in which it is located and ensuring the code reading recognition rate.

[0067] Moreover, the optimization training process is automatically completed by the smart device, so that it can optimize the recognition model in real time for no-read images that fail to be recognized, effectively improving the recognition model optimization efficiency of the smart device, enhancing the user experience, and saving the user the process window time for the smart device to collect and feedback no-read image samples and control the smart device to upgrade the firmware, thereby ensuring production line efficiency.

[0068] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. Accordingly, the computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can implement the method of any of the above-described embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0069] The above are only specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Those skilled in the art should understand that the present invention includes but is not limited to the contents described in the drawings and the above specific embodiments. Any modifications that do not deviate from the functional and structural principles of the present invention are intended to be included within the scope of the claims.

Claims

1. An information identification method, characterized in that: Applied to a smart device, the smart device performs image recognition based on a recognition model installed on it; the method comprises: Acquire a no-read picture; the no-read picture is a picture in which target information cannot be recognized using the recognition model; According to the no-read reason of the no-read picture, the no-read picture is enhanced to obtain an enhanced picture; Based on the enhanced image, recognition is performed using the recognition model to obtain true value information; The recognition capability of the recognition model is optimized by using the training samples formed by the unread pictures and the true value information.

2. The information identification method according to claim 1, characterized in that: The step of performing enhancement processing on the no-read picture according to the no-read reason of the no-read picture to obtain an enhanced picture includes: Different types of compensation enhancement are performed on the no-read picture for each no-read reason to obtain the enhanced picture corresponding to the no-read picture for each no-read reason.

3. The information identification method according to claim 2, characterized in that: The performing different types of compensation enhancement on the no-read picture for each no-read reason includes at least one of the following: For the image without image whose unreading reason is the abnormal barcode image size, scaling the barcode image area of ​​the image without image so that the barcode image size is within a preset size range, and obtaining the corresponding enhanced image; For the image without image whose unreading reason is abnormal bar width in the barcode image, dilation or corrosion processing is performed on the barcode image area of ​​the image without image to make the bar width within a preset bar width range, thereby obtaining the corresponding enhanced image; For the image without image whose unreading reason is abnormal grayscale of barcode image, grayscale stretching processing is performed on the barcode image area of ​​the image without image, so that the grayscale of the barcode image is within a preset grayscale range, and the corresponding enhanced image is obtained; For the image without image whose unreading reason is barcode image distortion, the image without image is subjected to distortion correction processing to obtain the corresponding enhanced image; For the image without image whose no-reading reason is the missing of the barcode quiet zone, the quiet zone part of the barcode in the image without image is completed to obtain the corresponding enhanced image; For the picture without image whose no-reading reason is the broken needle of the barcode, the broken needle part of the barcode in the picture without image is completed to obtain the corresponding enhanced picture.

4. The information identification method according to any one of claims 1 to 3, characterized in that: The smart device performs image recognition based on the recognition model in the working mode, and stops image recognition in the sleeping mode; The step of optimizing the recognition capability of the recognition model by using the training samples formed by the unread pictures and the true value information is performed when the smart device is in sleep mode.

5. The information identification method according to any one of claims 1 to 3, characterized in that: The smart device includes a camera, and the information recognition method further includes: The photographing device obtains a picture taken by the photographing device as a picture to be identified, and performs picture recognition on the picture to be identified based on the recognition model.

6. The information identification method according to any one of claims 1 to 3, characterized in that: The obtaining of the no-read picture comprises obtaining the no-read picture from a no-read training library; The information identification method further includes: Storing the no-read pictures in the no-read training library; The no-read picture corresponding to the enhanced picture from which the true value information is identified is removed from the no-read training library; When the number of no-read images in the no-read training library is higher than a preset number threshold, the no-read images in the no-read training library are cleared.

7. The information identification method according to any one of claims 1 to 3, characterized in that: The target information includes barcode information.

8. An electronic device, characterized in that: The electronic device comprises: one or more processors; A memory having one or more computer programs stored thereon, wherein when the one or more computer programs are executed by the one or more processors, the one or more processors implement the information identification method according to any one of claims 1 to 7.

9. A computer readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the information identification method according to any one of claims 1 to 7 is implemented.

10. A smart device, characterized in that: The smart device includes a photographing device and the electronic device according to claim 8, and the electronic device is used to obtain a picture taken by the photographing device as a picture to be identified.

Images