Generation program, generation method, and information processing device.

Abstract

The challenge is to generate accurate information regarding measures to be applied to stores. [Solution] The information processing device acquires purchase information about a product or service. The information processing device detects the area of ​​the store where the product is located when the analysis results of the acquired purchase information over a predetermined period of time satisfy predetermined conditions. The information processing device extracts the analysis results of the video corresponding to the predetermined period from the video captured in the detected area, and stores the analysis results of the extracted video in the storage unit each time an analysis result for the video corresponding to the predetermined period is extracted. The information processing device inputs the analysis results of the video stored in the storage unit during the first period, along with prompts including requests regarding measures to be applied to the store, into a language model to generate information related to measures to be applied to the store as a response to the requests, and outputs information related to the generated measures.

Claims

1. On the computer, We obtain purchase information regarding products or services, The analysis results of the acquired purchase information over a predetermined period of time satisfy predetermined conditions, and the area of ​​the store where the product is located is detected. From the video footage taken in the detected area, the analysis results of the video corresponding to a predetermined period are extracted. Each time an analysis result of the video corresponding to the predetermined period is extracted, the extracted video analysis result is stored in the storage unit. During the first period, prompts including the analysis results of the video stored in the memory unit are input to the language model to generate information related to measures to be applied to the store. Outputs information related to the generated measures. A generation program characterized by executing a process.

2. The generation program according to claim 1, characterized in that the generation process generates information related to measures to be applied to the store as a response to the request by inputting the analysis results of the video stored in the memory unit during the first period and prompts including a request for measures to be applied to the store into a language model.

3. By searching the graph data of the video footage taken in the detected area, results indicating interaction information associated with a person are identified. By inputting domain knowledge about the store, the results showing the identified interaction information, and the requirements regarding measures to be applied to the store into a language model, information related to measures to be applied to the store is generated as a response to the requirements. The generation program according to claim 2, characterized in that it is to be executed by the aforementioned computer.

4. The system acquires input data to which the object to be detected is associated with attribute information of the object or interaction information between the objects. By analyzing the video footage of the detected area using the input data, a first object representing the detected object is detected from among the video frames constituting the video. Based on the input data and the detected first object, a result is generated showing attribute information of the first object or interaction information related to the first object. The generated results are associated with the detected first object, and graph data is generated. The generation program according to claim 3, characterized in that it is to be executed by the aforementioned computer.

5. Information regarding the structure of the graph data to be searched and the people included in the video of the detected area are obtained. Based on the information regarding the structure of the graph data to be searched, a search query for the graph data is generated. Based on the aforementioned search query, the graph data is searched, which associates the person's attribute information or interaction information between the person and the object with the person included in the video. By inputting the search results of the retrieved graph data and prompts including requests regarding measures to be applied at the store into the language model, information related to the measures to be applied at the store is generated as a response to the requests. The generation program according to claim 1, characterized in that it is to be executed by the aforementioned computer.

6. The generation program according to claim 1, characterized in that the language model is a transformer-based model trained using a first token set generated from a token set obtained by masking some tokens from a plurality of tokens.

7. By analyzing video footage of the area where the product is placed when the predetermined conditions are met, customer interaction information in the detected area during the first period is identified. Using the identified customer interaction information, the language model generates information regarding product display. Based on coaching information that associates the period after the first period with information regarding product display, the measures to be applied to the period after the first period include displaying information regarding product display on employees' terminals. The generation program according to claim 1, characterized in that it is to be executed by the aforementioned computer.

8. By analyzing video footage of the area where the product is placed when the predetermined conditions are met, customer interaction information in the detected area during the first period is identified. Using the identified customer interaction information, the language model generates guidance information indicating customer guidance within the store. Based on coaching information that associates the period after the first period with the guidance information indicating customer guidance, the generated guidance information is displayed on the digital signage installed in the store during the period after the first period. The generation program according to claim 1, characterized in that it is to be executed by the aforementioned computer.

9. By analyzing video footage of the area where the product is placed when the aforementioned predetermined conditions are met, the congestion level of the area is identified. Using the congestion status of the identified area, the language model generates information on measures related to employee allocation. The generation program according to claim 1, characterized in that it is to be executed by the aforementioned computer.

10. The aforementioned purchase information is POS (Point of Sales) data. The aforementioned POS data is data on customer consumption behavior acquired by a POS system, and the data items of consumption behavior include any of the following: the date and time the product was purchased, the number of products purchased, the name of the product purchased, the price of the product purchased, the gender of the person who purchased the product, or the age group of the person who purchased the product. The products that meet the aforementioned predetermined conditions are those that have increased in value or quantity by a predetermined amount or more during the aforementioned predetermined period. The generation program according to feature 1.

11. The generation program according to claim 1, characterized in that when a target is given to an AI agent, the AI ​​agent generates tasks to achieve the target, collects the results of video analysis from the storage unit as information necessary to have the language model execute the generated tasks, and inputs the collected video analysis results into the language model to generate information related to measures to be applied to the store.

12. Computers We obtain purchase information regarding products or services, The analysis results of the acquired purchase information over a predetermined period of time satisfy predetermined conditions, and the area of ​​the store where the product is located is detected. From the video footage taken in the detected area, the analysis results of the video corresponding to a predetermined period are extracted. Each time an analysis result of the video corresponding to the predetermined period is extracted, the extracted video analysis result is stored in the storage unit. During the first period, prompts including the analysis results of the video stored in the memory unit are input to the language model to generate information related to measures to be applied to the store. Outputs information related to the generated measures. A generation method characterized by performing a process.

13. We obtain purchase information regarding products or services, The analysis results of the acquired purchase information over a predetermined period of time satisfy predetermined conditions, and the area of ​​the store where the product is located is detected. From the video footage taken in the detected area, the analysis results of the video corresponding to a predetermined period are extracted. Each time an analysis result of the video corresponding to the predetermined period is extracted, the extracted video analysis result is stored in the storage unit. During the first period, prompts including the analysis results of the video stored in the memory unit are input to the language model to generate information related to measures to be applied to the store. Outputs information related to the generated measures. An information processing device characterized by having a control unit.

Images