Information processing device, information processing method, and program

Abstract

[Problem] To implement practical application of a code that is more adapted to an environment. [Solution] This information processing device comprises a recognition unit that recognizes a code image from a composite image that is generated using the code image and a cover image, wherein the recognition unit recognizes the code image from the composite image by using a recognizer that has learned to match a recognition result of the code image and a recognition result of the composite image.

Description

[Technical field]

[0001] The present disclosure relates to an information processing device, an information processing method, and a program. [Background technology]

[0002] In recent years, various two-dimensional codes have been developed and are being widely used. For example, Patent Document 1 discloses a logistics management system that uses two-dimensional codes. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2002-87532 A Summary of the Invention [Problem to be solved by the invention]

[0004] Two-dimensional codes can store and read various information, and can also provide services that utilize that information. However, two-dimensional codes generally have unique shapes that conform to their standards, and so may not fit in with the environment in which they are used. [Means for solving the problem]

[0005] According to one aspect of the present disclosure, there is provided an information processing device including a recognition unit that recognizes a code image from a composite image generated using a code image and a cover image, wherein the recognition unit recognizes the code image from the composite image using a recognizer that has been trained so that a recognition result for the code image and a recognition result for the composite image are identical.

[0006] According to another aspect of the present disclosure, there is provided an information processing method including a processor recognizing the code image from a composite image generated using a code image and a cover image, and the recognition further includes recognizing the code image from the composite image using a recognizer that has been trained so that a recognition result of the code image and a recognition result of the composite image are identical.

[0007] According to another aspect of the present disclosure, there is provided a program for causing a computer to function as an information processing device, comprising: a recognition unit that recognizes the code image from a composite image generated using a code image and a cover image, the recognition unit using a recognizer that has been trained to make the recognition result of the code image and the recognition result of the composite image identical to each other. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram for explaining an overview of a composite image according to an embodiment of the present disclosure. [Diagram 2] 13 is a diagram showing an example of utilization of a composite image I3 according to the embodiment. FIG. [Diagram 3] 2 is a block diagram showing an example of a functional configuration of a learning device 10 according to the embodiment. FIG. [Figure 4] 2 is a block diagram showing an example of a functional configuration of a recognition device 20 according to the embodiment. FIG. [Diagram 5] 13 is a flowchart simply illustrating an example of a learning flow for recognizing a synthetic image I3 according to the embodiment. [Figure 6] 11 is a diagram for explaining learning related to generation of a synthetic image I3 according to the embodiment. FIG. [Figure 7] 13 is a diagram for explaining the improvement in design achieved by the composite image I3 according to the embodiment. FIG. [Figure 8] 13 is a diagram showing an example of application of a composite image I3 according to the embodiment to identity verification. FIG. [Figure 9] 13 is a diagram showing an example of applying a synthetic image I3 according to the embodiment to a moving image. FIG. [Figure 10] FIG. 2 is a block diagram showing an example of a hardware configuration of an information processing device 90 according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In this specification and the drawings, components having substantially the same functional configurations are denoted by the same reference numerals, and redundant description will be omitted.

[0010] The explanation will be given in the following order. 1. Embodiment 1.1. Overview 1.2. Configuration Example 1.3. Learning Details 1.4.Application Examples 2. Hardware configuration example 3. Summary

[0011] <1. Embodiment> <<1.1. Overview>> First, an overview of one embodiment of the present disclosure will be described.

[0012] As described above, in recent years, various types of two-dimensional codes have been developed and are widely used. The applications of two-dimensional codes are diverse, including tracking in logistics, advertising in marketing, payment in sales, and various types of control in the robotics field.

[0013] Two-dimensional codes, which are now widely used, allow information to be stored and read easily, and make it possible to build the various systems described above.

[0014] However, since two-dimensional codes generally have unique shapes according to standards, they may stand out more than necessary in the environment in which they are used.

[0015] For example, consider a case where a two-dimensional code is used in a historical building or area, etc. In this case, the two-dimensional code may appear too modern compared to the atmosphere of the building or area, giving the impression that it does not fit in with the atmosphere of the place.

[0016] Also, for example, a case is considered in which a two-dimensional code is used for mechanical control of an autonomous moving body (for example, route control of an autonomous moving body). In this case, the two-dimensional code is a medium that is unrelated to a person. For this reason, some people may feel uncomfortable in an environment where many two-dimensional codes like those described above are placed.

[0017] The technical concept of one embodiment of the present disclosure was conceived with a focus on the above points, and realizes the utilization of code that is more familiar with the environment.

[0018] To this end, in one embodiment of the present disclosure, a machine learning technique called Adversarial Example is applied. Adversarial Example generally refers to an image generated to cause a classification model to misclassify.

[0019] For example, suppose there is an image that is normally classified as a "dog" in a classification model, and of course, that image is also recognized as a "dog" by humans.

[0020] On the other hand, an adversarial example generated by adding noise to the above image that is indistinguishable to the human eye can cause the original class "dog" to be classified as another class (e.g., car).

[0021] In other words, Adversarial Examples make it possible to differentiate between recognition by the human eye and recognition in computer vision.

[0022] Therefore, the recognition device 20 according to an embodiment of the present disclosure includes a recognition unit 210 that recognizes the code image from a composite image generated using the code image and the cover image. Fruit One of the features of this method is that it recognizes a code image from a composite image using a recognizer that has been trained so that the recognition results for the composite image are the same as those for the original image.

[0023] FIG. 1 is a diagram for explaining an overview of a composite image according to an embodiment of the present disclosure.

[0024] FIG. 1 shows an example of an arbitrary code image I1, an arbitrary cover image I2, and a composite image I3 generated from the code image I1 and the cover image I2.

[0025] Here, the code image I1 according to this embodiment may be, for example, an image corresponding to various codes according to any standard.

[0026] For example, the code image I1 according to this embodiment may use a commonly available two-dimensional code, or may use a unique code designed according to the application.

[0027] The cover image I2 according to the present embodiment is an image used to conceal the code image I1 from human eyes. That is, an image that is desired to be recognized by humans is adopted as the cover image I2 according to the present embodiment.

[0028] For example, the cover image I2 according to this embodiment may be selected depending on the environment in which the composite image I3 is attached or displayed.

[0029] As an example, the cover image I2 according to this embodiment may be an image that reproduces the texture of an object to which the composite image I3 is attached or displayed.

[0030] For example, when the composite image I3 according to this embodiment is attached to a wood-grain wall, the cover image I2 may be an image that reproduces the wood-grain texture of the wall.

[0031] In this case, the composite image I3 generated using the code image I1 and the cover image I2 may be an image that has higher pixel-based similarity to the cover image I2 than to the code image I1.

[0032] In addition, the composite image I3 of this embodiment may be defined as an image in which another recognition device capable of recognizing the code image I1 based on certain image features (for example, in the case where a QR code (registered trademark) is used for the code image I1, a general QR code reader) cannot recognize the code image I1 from the composite image I3 based on the image features.

[0033] According to the above-described composite image I3, while it appears to the human eye to be equivalent to the cover image I2, it is possible for computer vision to recognize the code image I1.

[0034] FIG. 2 is a diagram showing an example of the use of the composite image I3 according to this embodiment. 2 shows an object O1, which is a building. A composite image I3 generated using a cover image I2 that reproduces (takes a photograph of) the texture of the exterior wall of the object O1 is attached to the exterior wall.

[0035] In this case, the synthetic image I3 has an appearance that is almost the same as the outer wall of the object O1, and therefore is not conspicuous to the human eye and blends in well with the environment.

[0036] On the other hand, the composite image I3 is an image in which the code image I1 can be recognized by the recognition device 20 realized as, for example, a smartphone.

[0037] Therefore, the recognition device 20 can read various pieces of information stored in the code image I1 by photographing the composite image I3. The recognition device 20 can also execute various operations according to the read information. The operations include, for example, accessing an Internet site and displaying AR. When the attachment position of the composite image I3 changes or the composite image I3 is dynamically displayed, the recognition device 20 can also be applied to gesture recognition and motion recognition.

[0038] The recognition device 20 according to the present embodiment may be mounted on various autonomous moving bodies such as an autonomous vehicle V1. In this case, the autonomous vehicle V1 can read various pieces of information stored in the code image I1 by capturing the composite image I3 and perform autonomous operations according to the read information.

[0039] As described above, according to the technical concept of this embodiment, it is possible to realize a code that is not noticeable to humans as a code, but is recognizable by computer vision.

[0040] This also makes it possible to use the code in historic buildings and areas such as those mentioned above without spoiling the atmosphere of the place.

[0041] Furthermore, even when the code is used for mechanical control of an autonomous moving body or the like (for example, route control of an autonomous moving body), operation can be performed without people being aware of the code.

[0042] <<1.2. Configuration example>> Next, a detailed description will be given of an example of the configuration of a system that realizes the above-mentioned effects.

[0043] (Learning device 10) First, the functional configuration of the learning device 10 according to this embodiment will be described.

[0044] The learning device 10 according to this embodiment is an information processing device that performs one or both of learning related to generation of a composite image I3 and learning related to recognition of a code image I1 from the composite image I3.

[0045] 3 is a block diagram showing an example of a functional configuration of the learning device 10 according to this embodiment. As shown in FIG.

[0046] (Learning section 110) The learning unit 110 according to this embodiment performs one or both of learning related to generation of the composite image I3 and learning related to recognition of the code image I1 from the composite image I3.

[0047] The learning section 110 according to this embodiment performs the above-mentioned learning using a neural network.

[0048] The functions of the learning unit 110 according to this embodiment are realized by a processor such as a GPU. Details of the learning by the learning unit 110 will be described separately.

[0049] (Storage unit 120) The storage unit 120 according to this embodiment stores various information related to the learning performed by the learning unit 110. For example, the storage unit 120 stores the structure of a network used for learning by the learning unit 110, various parameters related to the network, learning data, and the like.

[0050] An example of the functional configuration of the learning device 10 according to this embodiment has been described above. Note that the functional configuration described above using Fig. 3 is merely an example, and the functional configuration of the learning device 10 according to this embodiment is not limited to this example.

[0051] For example, the study device 10 according to this embodiment may further include an operation unit that accepts operations by a user, a display unit that displays various types of information, and the like.

[0052] The functional configuration of the learning device 10 according to this embodiment can be flexibly modified according to the specifications and operation.

[0053] (Recognition device 20) Next, an example of the functional configuration of the recognition device 20 according to this embodiment will be described.

[0054] The recognition device 20 according to this embodiment is an information processing device that recognizes a code image I1 from a composite image I3.

[0055] The recognition device 20 according to the present embodiment may be realized as, for example, a smartphone, a tablet, a wearable device, or a dedicated reader. The recognition device 20 according to the present embodiment may be mounted on an autonomous moving object such as a self-driving vehicle, a drone, or a robot.

[0056] 4 is a block diagram showing an example of a functional configuration of the recognition device 20 according to the present embodiment. As shown in FIG. 4, the recognition device 20 according to the present embodiment may include a recognition unit 210, an imaging unit 220, an operation unit 230, and a storage unit 240.

[0057] (Recognition unit 210) The recognition unit 210 according to this embodiment recognizes the code image I1 from the composite image I3 generated using the code image I1 and the cover image I2.

[0058] In addition, one of the features of the recognition unit 210 according to this embodiment is that it recognizes the code image I1 from the composite image I3 using a recognizer that has been trained so that the recognition result of the code image I1 and the recognition result of the composite image I3 are identical.

[0059] According to the recognition unit 210 of this embodiment, it becomes possible to read information from the synthetic image I3 that is not noticeable as a code to the human eye, and to utilize the information for various purposes.

[0060] The functions of the recognition unit 210 according to this embodiment are realized by various processors such as a GPU, etc. Details of the functions of the recognition unit 210 according to this embodiment will be described separately.

[0061] (Camera Division 220) The image capturing section 220 according to this embodiment captures an image of the surrounding environment. In particular, the image capturing section 220 according to this embodiment may capture an image of a composite image I3 that exists in the surrounding environment as a subject.

[0062] For this reason, the imaging section 220 according to this embodiment includes various types of imaging elements.

[0063] (Operation unit 230) The operation unit 230 according to this embodiment executes various operations based on the code image I1 recognized by the recognition unit 210. Note that the reading of information from the code image I1 may be performed by the recognition unit 210 or the operation unit 230.

[0064] Furthermore, the operation performed by the operation unit 230 according to this embodiment is designed appropriately according to the characteristics of the recognition device 20, the device in which the recognition device 20 is installed, and the application or service that utilizes the synthetic image I3.

[0065] (Storage unit 240) The storage unit 240 according to this embodiment stores information used by each component included in the recognition device 20. The storage unit 240 according to this embodiment stores, for example, information about a recognizer used by the recognition unit 210 for recognition, an image captured by the imaging unit 220, a program used for operation by the operation unit 230, and the like.

[0066] The above describes an example of the functional configuration of the recognition device 20 according to the present embodiment. Note that the functional configuration described with reference to Fig. 4 is merely an example, and the functional configuration of the recognition device 20 according to the present embodiment is not limited to this example.

[0067] For example, the recognition device 20 according to the present embodiment may further include an operation unit that accepts operations by a user, a display unit that displays various information, and the like.

[0068] The functional configuration of the recognition device 20 according to this embodiment can be flexibly modified according to the specifications and operation.

[0069] <<1.3. Learning details>> Next, learning by the learning device 10 according to this embodiment will be described in detail. First, learning related to recognition of the synthetic image I3 by the learning device 10 will be described.

[0070] FIG. 5 is a flowchart showing a simplified example of the flow of learning related to recognition of the synthetic image I3.

[0071] In the example shown in FIG. 5, the learning unit 110 first performs a first learning to recognize the code image I1 (S102), and then performs a second learning to make the recognition result of the composite image I3 the same as the recognition result of the code image I1 (S104).

[0072] First, the first learning in step S102 will be described. Here, learning is performed in which the network D serving as a recognizer receives an input image a (which may not include the code image I1) and generates an output b. The output b may include information about the position of the code image I1 and information encoded in the code image I1.

[0073] The relationship between the input image a, the network D, and the output b can be defined as shown in the following equation (1).

[0074]

number

[0075] In the first learning, a dataset including photographs of a code image I1 conforming to an arbitrary standard taken under various conditions (e.g., lighting, etc.) with various sizes, positions, and orientations may be used, which allows for more robust recognition of the code image I1.

[0076] In the second learning in the following step S104, a synthetic image I3 that is as close as possible to the cover image I2 to the human eye is repeatedly generated, while learning is performed to elicit from the network D a response similar to that to the code image I1.

[0077] This optimization problem can be written as a minimization problem of a loss function as shown in the following formula (2).

[0078]

number

[0079] In the above formula (2), x denotes the input image, i.e., the synthetic image I3, which is iteratively updated to minimize the loss function. T is the distribution of the differentiable image transformation function t(x), which is selected to simulate the variations in the angle and distance range, camera specifications, lighting conditions, etc., at which the code image I1 is captured in real-life scenarios.

[0080] In addition, in the above formula (2), λ k is the loss term L k are empirically set weights for

[0081] In addition, L k may be defined, for example, as follows:

[0082] L0(t(x),t(x cover ):t(x) and t(x cover ), where t(x) and t(x cover ) are the images x (composite image I3) and x cover This represents the result of applying image transformation function t to (cover image I2).

[0083] L1(y,y code ):y=F(t(x)) and y=F(t(x code )), where y and y code are t(x) and t(x code ) represents the response of network D to the code corresponds to the code image I1. Also, the function F may be the same as the network D.

[0084] Other losses may be appropriately designed according to the code standard to be adopted, the characteristics of the application, etc. For example, when it is assumed that the composite image I3 will be mainly printed using a standard color printer, a non-printability score or the like may be set as a loss.

[0085] The above describes the learning related to the recognition of the composite image I3. According to the learning described above, a classifier capable of recognizing the code image I1 from the composite image I3 can be generated. The generated classifier is used by the recognition device 20.

[0086] Next, a description will be given of the learning involved in generating the composite image I3. The composite image I3 according to this embodiment is generated using a generator that is generated by learning based on the code image I1 and the cover image I2.

[0087] The generator may be generated by learning to make the output when the code image I1 is input closer to the output when the cover image I2 is input.

[0088] FIG. 6 is a diagram for explaining the learning related to the generation of the synthetic image I3.

[0089] A general method for generating Adversarial Examples is shown in the upper part of Figure 6. In Figure 6, a represents the input image, and b represents the final output of Network D. Each rectangle represents an intermediate layer of Network D.

[0090] In a general method for generating Adversarial Examples, the loss L is, for example, b and bb goal It is defined as the distance between

[0091] where b is the final output of network D for input image a (code image I1), and b goal is the input image a goal This may be the final output of network D for (cover image I2).

[0092] By learning using the loss defined above, it is possible to generate a synthetic image I3 in which the code image I1 is modified so as to be closer to the cover image I2.

[0093] On the other hand, the output of the intermediate layer in a neural network may contain useful information depending on the application.

[0094] For example, given two images (a photo of a person's face and Van Gogh's "Starry Night"), it is possible to perform computations using the intermediate responses of an image classifier network such as VGG-19 as input to create a new image that incorporates the content of one image and the style of the other (e.g., redrawing the person's face in the style of "Starry Night").

[0095] For this reason, the generator of this embodiment may be generated by learning to bring the output of a predetermined intermediate layer when the code image I1 is input closer to the output of a predetermined intermediate layer when the cover image I2 is input.

[0096] In this case, the loss L is calculated by the output b k and output b k goal It may be defined as the distance between

[0097] Or, more specifically, it may be defined as L1 shown in the following formula (3).

[0098]

number

[0099] In the above formula (3), x code represents the code image I1. i l Let denote the i-th element in layer l of the intermediate output (i.e., feature map) of network D. Also, L denotes the set of intermediate layers whose outputs are considered in the computation of L1.

[0100] According to L1 defined as above, it is possible to adjust the similarity between the generated synthetic image I3 and the cover image I2 by defining set L. For example, if a lower layer (a layer close to the input) of network D is included in set L, it is expected that the recognition accuracy of the generated synthetic image I3 will improve while the possibility that it will look like a code to the human eye increases.

[0101] Other loss terms include, for example, simply X and X cover The mean squared error (L0) between X and X cover Examples include the color divergence score (L2), which is obtained by adding the RGB variance of all pixels in the difference between the two.

[0102] <<1.4. Application Examples>> Next, a specific example of application of the composite image I3 according to this embodiment will be described. In Fig. 2, the composite image I3 according to this embodiment is applied as a marker that blends in with the environment, but the composite image I3 according to this embodiment is not limited to this example and can be applied to various situations.

[0103] FIG. 7 is a diagram for explaining an improvement in design by the composite image I3 according to the present embodiment.

[0104] A typical pizza package O2a is shown on the left side of Figure 7. A typical pizza package O2a requires the printing of a code image I1 (here, a barcode) for product management, which reduces the design quality.

[0105] On the other hand, a package O2b to which a composite image I3 according to this embodiment is applied is shown on the right side of Fig. 7. A composite image I3 generated using a code image I1 and a cover image I2 that resembles a chef's hat is printed on the package O2b, making it possible to manage products without printing a barcode itself.

[0106] In this way, the composite image I3 according to this embodiment can eliminate design restrictions such as barcode printing, and improve the design. Note that, although Fig. 7 shows an example in which the composite image I3 is applied to product packaging, similar improvements in design are expected for posters, pamphlets, flyers, etc.

[0107] Next, a case where the composite image I3 according to this embodiment is used for user identification will be described. Fig. 8 is a diagram showing an example in which the composite image I3 according to this embodiment is applied to identification.

[0108] In recent years, tickets are sometimes issued with a two-dimensional code printed with purchase information to prevent the counterfeiting and resale of tickets for concerts and other events. A system has also been proposed in which a user's face image is uploaded to a server when purchasing a ticket, and the face image is used for identification purposes in addition to confirmation by a two-dimensional code when entering a venue. The left side of Fig. 8 shows a ticket O3a with a code image I1a (here, a QR code) printed on it for the above-mentioned purpose.

[0109] On the other hand, the right side of Fig. 8 shows a ticket O3b on which a composite image I3 generated using a code image I1b adopting a user's face image and a cover image I2 is printed. In this way, the code image I1 according to this embodiment may be a face image.

[0110] In this case, the composite image I3 is generated using the code image I1b uploaded by the user when purchasing the ticket, and the code image I1b may be discarded without being stored on the server after the composite image I3 is generated.

[0111] Such a ticket O3b can simultaneously prevent counterfeiting and resale and provide identity verification based on a facial image. Furthermore, since there is no need to store the facial image on a server for a long period of time, it is also useful from the perspective of protecting personal information.

[0112] The composite image I3 using the code image I1 that employs such a facial image can also be applied to, for example, ID cards, etc. This can meet the needs of users who are reluctant to have their own faces printed on cards.

[0113] In addition, when the above-described composite image I3 is applied to a credit card, even if a third party illegally obtains a credit card and attempts to use it, the facial image (code image I1) can be confirmed at the register, so it is expected that fraudulent use will be prevented.

[0114] Next, a case where the composite image I3 according to the present embodiment is applied to a moving image will be described. Although the above mainly illustrates a case where the composite image I3 according to the present embodiment is statically printed, the composite image I3 according to the present embodiment may be dynamically displayed.

[0115] Fig. 9 is a diagram showing an example of applying a composite image I3 according to this embodiment to a moving image. In the example shown in Fig. 9, the composite image I3 is generated as a cover image I2 of a broadcast program displayed on a television O4. At this time, a code image I1 storing arbitrary information is embedded in a frame of the video.

[0116] According to the above-described technique, a user can use the recognition device 20 to obtain information distributed in association with a broadcast program while viewing the broadcast program.

[0117] Furthermore, distributors of broadcast programs can provide more information to viewers without having to split the screen.

[0118] The application of the composite image I3 to moving images is not limited to broadcast programs, but can also be widely applied to digital signage and the like.

[0119] The application of the composite image I3 according to this embodiment has been described above with specific examples. Although the above describes an example in which the composite image I3 according to this embodiment is pasted or displayed on a plane (two-dimensional), the composite image I3 according to this embodiment can also be generated as a code (2.5-dimensional) that includes some depth, such as a manhole or tile. The composite image I3 according to this embodiment can also be realized as a three-dimensional code, such as a sculpture or artificial flower with an embedded code.

[0120] As described above, the synthetic image I3 according to this embodiment can be generated by adjusting the robustness against changes in distance, angle, lighting conditions, etc. For this reason, the synthetic image I3 according to this embodiment can be generated according to the purpose, for example, so that it can be recognized only up close or only from a specific angle.

[0121] <2. Hardware configuration example> Next, a hardware configuration example common to the learning device 10 and the recognition device 20 according to an embodiment of the present disclosure will be described. FIG. 10 is a block diagram showing a hardware configuration example of an information processing device 90 according to an embodiment of the present disclosure. The information processing device 90 may be a device having a hardware configuration equivalent to each of the above devices. As shown in FIG. 10, the information processing device 90 has, for example, a processor 871, a ROM 872, a RAM 873, a host bus 874, a bridge 875, an external bus 876, an interface 877, an input device 878, an output device 879, a storage 880, a drive 881, a connection port 882, and a communication device 883. Note that the hardware configuration shown here is an example, and some of the components may be omitted. In addition, the information processing device 90 may further include components other than the components shown here.

[0122] (Processor 871) The processor 871 functions, for example, as an arithmetic processing device or control device, and controls all or part of the operation of each component based on various programs recorded in the ROM 872, the RAM 873, the storage 880, or the removable recording medium 901.

[0123] (ROM 872, RAM 873) The ROM 872 is a means for storing the programs read into the processor 871, data used in calculations, etc. The RAM 873 temporarily or permanently stores, for example, the programs read into the processor 871 and various parameters that change appropriately when the programs are executed.

[0124] (host bus 874, bridge 875, external bus 876, interface 877) The processor 871, ROM 872, and RAM 873 are connected to one another via, for example, a host bus 874 capable of high-speed data transmission. On the other hand, the host bus 874 is connected to an external bus 876 having a relatively low data transmission speed via, for example, a bridge 875. In addition, the external bus 876 is connected to various components via an interface 877.

[0125] (Input Device 878) The input device 878 may be, for example, a mouse, a keyboard, a touch panel, a button, a switch, a lever, etc. Furthermore, a remote controller capable of transmitting a control signal using infrared rays or other radio waves may be used as the input device 878. The input device 878 may also include an audio input device such as a microphone.

[0126] (Output Device 879) The output device 879 is a device capable of visually or audibly notifying the user of acquired information, such as a display device such as a CRT (Cathode Ray Tube), LCD, or organic EL, an audio output device such as a speaker or a headphone, a printer, a mobile phone, a facsimile, etc. The output device 879 according to the present disclosure also includes various vibration devices capable of outputting tactile stimuli.

[0127] (Storage 880) The storage 880 is a device for storing various types of data. For example, a magnetic storage device such as a hard disk drive (HDD), a semiconductor storage device, an optical storage device, or a magneto-optical storage device may be used as the storage 880.

[0128] (Drive 881) The drive 881 is a device that reads information recorded on a removable recording medium 901 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, or writes information to the removable recording medium 901.

[0129] (Removable recording media 901) The removable recording medium 901 is, for example, a DVD medium, a Blu-ray (registered trademark) medium, an HD DVD medium, various semiconductor storage media, etc. Of course, the removable recording medium 901 may be, for example, an IC card equipped with a non-contact type IC chip, an electronic device, etc.

[0130] (Connection port 882) The connection port 882 is a port for connecting an external device 902, such as a Universal Serial Bus (USB) port, an IEEE 1394 port, a Small Computer System Interface (SCSI), an RS-232C port, or an optical audio terminal.

[0131] (External connection device 902) The external connection device 902 is, for example, a printer, a portable music player, a digital camera, a digital video camera, or an IC recorder.

[0132] (Communication Device 883) The communication device 883 is a communication device for connecting to a network, such as a communication card for wired or wireless LAN, Bluetooth (registered trademark), or WUSB (Wireless USB), a router for optical communications, a router for ADSL (Asymmetric Digital Subscriber Line), or a modem for various types of communications.

[0133] <3. Summary> As described above, the recognition device 20 according to an embodiment of the present disclosure includes the recognition unit 210 that recognizes a code image from a composite image generated using a code image and a cover image. Also, one of the features of the recognition unit 210 according to an embodiment of the present disclosure is that it recognizes a code image from a composite image using a recognizer that has been trained so that the recognition result of the code image and the recognition result of the composite image are the same.

[0134] According to the above configuration, it is possible to realize the use of code that is more compatible with the environment.

[0135] Although the preferred embodiment of the present disclosure has been described in detail above with reference to the attached drawings, the technical scope of the present disclosure is not limited to such examples. It is clear that a person having ordinary knowledge in the technical field of the present disclosure can conceive of various modified or amended examples within the scope of the technical ideas described in the claims, and it is understood that these also naturally belong to the technical scope of the present disclosure.

[0136] Furthermore, the steps of the processes described in this specification do not necessarily have to be processed in chronological order according to the order described in the flowchart or sequence diagram. For example, the steps of the processes of each device may be processed in an order different from the order described, or may be processed in parallel.

[0137] Furthermore, the series of processes performed by each device described in this specification may be realized using any of software, hardware, and a combination of software and hardware. The programs constituting the software are stored in advance, for example, in a recording medium (non-transitory medium) provided inside or outside each device. Then, each program is loaded into a RAM when executed by a computer, for example, and executed by various processors. The recording medium is, for example, a magnetic disk, an optical disk, a magneto-optical disk, a flash memory, etc. Furthermore, the computer programs may be distributed, for example, via a network, without using a recording medium.

[0138] In addition, the effects described in this specification are merely descriptive or exemplary and are not limiting. In other words, the technology according to the present disclosure may achieve other effects that are apparent to a person skilled in the art from the description of this specification, in addition to or in place of the above effects.

[0139] Note that the following configurations also fall within the technical scope of the present disclosure. (1) a recognition unit that recognizes the code image from a composite image generated using the code image and a cover image; Equipped with the recognition unit recognizes the code image from the composite image using a recognizer that has been trained so that a recognition result of the code image and a recognition result of the composite image are identical; Information processing device. (2) the composite image is an image that has a higher pixel-based similarity to the cover image than to the code image; The information processing device according to (1). (3) the recognizer is generated by a first learning for recognizing the code image and a second learning for making a recognition result of the composite image the same as a recognition result of the code image; The information processing device according to (1) or (2). (4) The cover image is selected according to an environment in which the composite image is to be pasted or displayed. The information processing device according to any one of (1) to (3). (5) the cover image includes an image reproducing a texture of an object to which the composite image is attached or displayed; The information processing device according to (4). (6) The code image includes a two-dimensional code. The information processing device according to any one of (1) to (5). (7) The code image includes a face image. The information processing device according to any one of (1) to (5). (8) The composite image includes an image in which another recognition device capable of recognizing the code image based on a certain image feature cannot recognize the code image from the composite image based on the image feature. The information processing device according to any one of (1) to (7). (9) The composite image is generated using a generator that is generated by learning based on the code image and the cover image. The information processing device according to any one of (1) to (8). (10) The generator is generated by learning to make an output when the code image is input closer to an output when the cover image is input. The information processing device according to (9). (11) The generator is generated by learning to make an output of a predetermined intermediate layer when the code image is input closer to an output of a predetermined intermediate layer when the cover image is input. The information processing device according to (10). (12) an imaging unit that captures the composite image; Further comprising: The information processing device according to any one of (1) to (11). (13) an operating unit that operates based on the code image recognized by the recognition unit; Further comprising: The information processing device according to any one of (1) to (12). (14) The processor: Recognizing the code image from a composite image generated using the code image and a cover image; Including, the recognizing step includes recognizing the code image from the composite image using a recognizer that has been trained so that a recognition result of the code image and a recognition result of the composite image are identical; Further comprising: Information processing methods. (15) Computer, a recognition unit that recognizes the code image from a composite image generated using the code image and a cover image; Equipped with the recognition unit recognizes the code image from the composite image using a recognizer that has been trained so that a recognition result of the code image and a recognition result of the composite image are identical. Information processing device, A program to function as a [Explanation of symbols]

[0140] 10 Learning Device 110 Learning Department 120 Storage section 20 Recognition device 210 Recognition part 220 Photography Department 230 Operating Part 240 Storage section

Claims

1. a recognition unit that recognizes the code image from a composite image generated using the code image and a cover image; Equipped with the recognition unit recognizes the code image from the composite image using a recognizer that has been trained so that a recognition result of the code image and a recognition result of the composite image are identical; the recognizer is generated by a first learning for recognizing the code image, and a second learning for making a recognition result of the composite image the same as a recognition result of the code image. Information processing device.

2. the composite image is an image that has a higher pixel-based similarity to the cover image than to the code image; The information processing device according to claim 1 .

3. The cover image is selected according to an environment in which the composite image is to be pasted or displayed. The information processing device according to claim 1 .

4. the cover image includes an image reproducing a texture of an object to which the composite image is attached or displayed; The information processing device according to claim 3 .

5. The code image includes a two-dimensional code. The information processing device according to claim 1 .

6. The code image includes a face image. The information processing device according to claim 1 .

7. The composite image includes an image in which another recognition device capable of recognizing the code image based on a certain image feature cannot recognize the code image from the composite image based on the image feature. The information processing device according to claim 1 .

8. The composite image is generated using a generator that is generated by learning based on the code image and the cover image. The information processing device according to claim 1 .

9. The generator is generated by learning to make an output when the code image is input closer to an output when the cover image is input. The information processing device according to claim 8.

10. The generator is generated by learning to make an output of a predetermined intermediate layer when the code image is input closer to an output of a predetermined intermediate layer when the cover image is input. The information processing device according to claim 9.

11. an imaging unit that captures the composite image; Further comprising: The information processing device according to claim 1 .

12. an operating unit that operates based on the code image recognized by the recognition unit; Further comprising: The information processing device according to claim 1 .

13. The processor: Recognizing the code image from a composite image generated using the code image and a cover image; Including, the recognizing step includes recognizing the code image from the composite image using a recognizer that has been trained so that a recognition result of the code image and a recognition result of the composite image are identical; Further comprising: the recognizer is generated by a first learning for recognizing the code image, and a second learning for making a recognition result of the composite image the same as a recognition result of the code image. Information processing methods.

14. Computer, a recognition unit that recognizes the code image from a composite image generated using the code image and a cover image; Equipped with the recognition unit recognizes the code image from the composite image using a recognizer that has been trained so that a recognition result of the code image and a recognition result of the composite image are identical; the recognizer is generated by a first learning for recognizing the code image, and a second learning for making a recognition result of the composite image the same as a recognition result of the code image. Information processing device, A program to function as a

Images