# US 2026/0065652 A1 · Non-transitory computer-readable recording medium, generation method, and information processing apparatus > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2026/0065652 A1 - Document ID: US20260065652A1 - Status: Published application - Kind: A1 - Application number: 19/385,230 - Filed: 2025-11-11 - Published: 2026-03-05 - Earliest claimed: 2023-05-29 - Assignees: Fujitsu Limited - Applicants: Fujitsu Limited - Inventors: Takashi KIKUCHI; Shun KOHATA - Web page: https://patentgrep.com/patent/US20260065652A1 ## CPC - G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting - G06T 7/00: Image analysis - G06T 7/11: Region-based segmentation - G06T 7/73: using feature-based methods - G06T 11/60: Creating or editing images; Combining images with text - G06V 10/82: using neural networks - G06V 20/52: Surveillance or monitoring of activities, e.g. for recognising suspicious objects (recognising microscopic objects G06V20/69) - G06V 40/103: {Static body considered as a whole, e.g. static pedestrian or occupant recognition} - G06V 40/23: {Recognition of whole body movements, e.g. for sport training} - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30196: Human being; Person - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A non-transitory computer-readable recording medium has stored therein a generation program that causes a computer to execute a process including acquiring an image that includes a person extracting an object that is used by the person included in the image by analyzing the acquired image generating a composite image in which the extracted object is arranged at a position that satisfies a predetermined condition on a basis of a position of the object that is used by the person included in the acquired image and generating, by using the generated composite image, a machine learning model that has been trained to identify the person who uses the object. ## Claims ### Claim 1 A non-transitory computer-readable recording medium having stored therein a generation program that causes a computer to execute a process comprising: acquiring an image that includes a person; extracting an object that is used by the person included in the image by analyzing the acquired image; generating a composite image in which the extracted object is arranged at a position that satisfies a predetermined condition on a basis of a position of the object that is used by the person included in the acquired image; and generating, by using the generated composite image, a machine learning model that has been trained to identify the person who uses the object. ### Claim 2 (depends on 1) The non-transitory computer-readable recording medium according to claim 1, wherein the process further includes receiving setting of a parameter based on a distance between a coordinate position of the person and a coordinate position of the object held by the person, generating, based on the set parameter, a coordinate position of an arrangement candidate for the image of the extracted object, determining whether or not the generated coordinate position is included in an area related to a size of the object, and generating, based on a determined result, the composite image in which the image of the object is arranged on the acquired image. ### Claim 3 (depends on 2) The non-transitory computer-readable recording medium according to claim 2, wherein the process further includes generating a first composite image based on a first object that is included in a first image, and generating a second composite image based on a second object that is included in a second image, training an encoder included in the machine learning model such that an output result obtained when the first image is input to the encoder and an output result obtained when the second image is input to the encoder approach each other, training the encoder such that the output result obtained when the first image is input to the encoder and an output result obtained when the first composite image is input to the encoder diverge from each other, and training the encoder such that the output result obtained when the first composite image is input to the encoder and an output result obtained when the second composite image is input to the encoder approach each other. ### Claim 4 (depends on 1) The non-transitory computer-readable recording medium according to claim 1, wherein the process further includes identifying a behavior of the person taking out a commodity product from a commodity product shelf by inputting an image that has been captured by a camera provided in an inside of a store and that includes both of the person and the commodity product shelf that accommodates the commodity products to the machine learning model. ### Claim 5 (depends on 1) The non-transitory computer-readable recording medium according to claim 1, wherein the process further includes extracting skeleton information on the person included in the image by analyzing the acquired image, and extracting, based on the skeleton information, the object that is used by the person. ### Claim 6 A generation method comprising: acquiring an image that includes a person; extracting an object that is used by the person included in the image by analyzing the acquired image; generating a composite image in which the extracted object is arranged at a position that satisfies a predetermined condition on a basis of a position of the object that is used by the person included in the acquired image; and generating, by using the generated composite image, a machine learning model that has been trained to identify the person who uses the object, by using a processor. ### Claim 7 (depends on 6) The generation method according to claim 6, further including receiving setting of a parameter based on a distance between a coordinate position of the person and a coordinate position of the object held by the person, generating, based on the set parameter, a coordinate position of an arrangement candidate for the image of the extracted object, determining whether or not the generated coordinate position is included in an area related to a size of the object, and generating, based on a determined result, the composite image in which the image of the object is arranged on the acquired image. ### Claim 8 (depends on 7) The generation method according to claim 7, further including generating a first composite image based on a first object that is included in a first image, and generating a second composite image based on a second object that is included in a second image, and training an encoder included in the machine learning model such that an output result obtained when the first image is input to the encoder and an output result obtained when the second image is input to the encoder approach each other, training the encoder such that the output result obtained when the first image is input to the encoder and an output result obtained when the first composite image is input to the encoder diverge from each other, and training the encoder such that the output result obtained when the first composite image is input to the encoder and an output result obtained when the second composite image is input to the encoder approach each other. ### Claim 9 (depends on 6) The generation method according to claim 6, further including identifying a behavior of the person taking out a commodity product from a commodity product shelf by inputting an image that has been captured by a camera provided in an inside of a store and that includes both of the person and the commodity product shelf that accommodates the commodity products to the machine learning model. ### Claim 10 (depends on 6) The generation method according to claim 6, further including extracting skeleton information on the person included in the image by analyzing the acquired image, and extracting, based on the skeleton information, the object that is used by the person. ### Claim 11 An information processing apparatus comprising: a memory; and a processor coupled to the memory and configured to: acquire an image that includes a person; extract an object that is used by the person included in the image by analyzing the acquired image; generate a composite image in which the extracted object is arranged at a position that satisfies a predetermined condition on a basis of a position of the object that is used by the person included in the acquired image; and generate, by using the generated composite image, a machine learning model that has been trained to identify the person who uses the object. ### Claim 12 (depends on 11) The information processing apparatus according to claim 11, wherein the processor is further configured to receive setting of a parameter based on a distance between a coordinate position of the person and a coordinate position of the object held by the person, generate, based on the set parameter, a coordinate position of an arrangement candidate for the image of the extracted object, determine whether or not the generated coordinate position is included in an area related to a size of the object, and generate, based on a determined result, the composite image in which the image of the object is arranged on the acquired image. ### Claim 13 (depends on 12) The information processing apparatus according to claim 12, wherein the processor is further configured to generate a first composite image based on a first object that is included in a first image, generate a second composite image based on a second object that is included in a second image, train an encoder included in the machine learning model such that an output result obtained when the first image is input to the encoder and an output result obtained when the second image is input to the encoder approach each other, train the encoder such that the output result obtained when the first image is input to the encoder and an output result obtained when the first composite image is input to the encoder diverge from each other, and train the encoder such that the output result obtained when the first composite image is input to the encoder and an output result obtained when the second composite image is input to the encoder approach each other. ### Claim 14 (depends on 11) The information processing apparatus according to claim 11, wherein the processor is further configured to identify a behavior of the person taking out a commodity product from a commodity product shelf by inputting an image that has been captured by a camera provided in an inside of a store and that includes both of the person and the commodity product shelf that accommodates the commodity products to the machine learning model. ### Claim 15 (depends on 11) The information processing apparatus according to claim 11, wherein the processor is further configured to extract skeleton information on the person included in the image by analyzing the acquired image, and extract, based on the skeleton information, the object that is used by the person.