A method for saving traffic and settling accounts based on cloud recognition image cabinet
By using local AI image recognition algorithm in vending machines to detect whether the hand reaches in, collect and save data only when necessary, the problem of useless data in video files is solved, traffic and time savings are achieved, and order settlement is significantly improved efficiency.
Patent Information
- Application Number
- CN202211423941.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-11-15
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2042-11-15
AI Technical Summary
Existing vending machines contain a large amount of useless data in the video files between users' door opening and closing, resulting in large video files, long upload time, more 4G traffic, and increasing the time-consuming of cloud AI image recognition.
When the device opens the door, start the camera and use the local AI image recognition algorithm to detect whether the hand is in. Only collect and save data when the hand is in, otherwise it will not be saved until the video file is uploaded for cloud recognition and settlement after the door is closed.
By filtering useless data, it reduces the size of video files, shortens the upload time and 4G traffic consumption, reduces the time spent on AI image recognition in cloud services, and significantly shortens the order settlement time.
Smart Images

Figure CN115767250B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of automatic vending machines, and in particular refers to a flow-saving settlement method based on a cloud recognition image cabinet. Background Art
[0002] After the user scans the code (or uses facial recognition) to open the door, the terminal device (vending machine) turns on the camera to start recording and saves the video locally. After the user closes the door, the terminal device turns off the camera and stops recording. Then, the terminal device reports the door closing to the cloud service and uploads the video file. The cloud service sends the video file to the AI algorithm for recognition and settles the order after the result is recognized. Figure 1 shown.
[0003] The main drawbacks are:
[0004] (1) The video file from the time the user opens the door to the time the user closes the door actually contains a lot of useless data, such as the data when the user does not perform any operation during shopping;
[0005] (2) Video files are relatively large, and uploading to cloud services takes a long time, consuming more 4G traffic and also increasing the cloud AI image recognition time. Summary of the invention
[0006] The purpose of the present invention is to provide a traffic-saving settlement method based on a cloud recognition image cabinet that can segment and optimize local data and save traffic.
[0007] The object of the present invention is achieved in that:
[0008] A traffic saving settlement method based on cloud recognition image cabinet, comprising:
[0009] Step 1: When the door of the device is opened, start the camera to start video acquisition;
[0010] Step 2: The device starts the local AI image recognition algorithm to detect whether the hand is inserted;
[0011] If a hand is detected, the data collected by the camera is saved locally;
[0012] If no hand is detected, the data collected by the camera will not be saved locally;
[0013] Step 3: When the user closes the door, the camera and local AI image recognition algorithm are turned off, the door closing is reported to the cloud server, and the locally saved video file is uploaded;
[0014] Step 4: The cloud service sends the received video file to the cloud AI image recognition algorithm for product identification;
[0015] Step 5: Settle the order based on the recognition results.
[0016] Preferably, an algorithm recognition area is set in the video image captured by the camera, and when a hand is recognized to enter the algorithm recognition area, the captured video data is written to the local storage card in real time; when a hand is recognized to exit the algorithm recognition area, writing the captured video data to the local storage card is stopped.
[0017] Preferably, the device is a vending machine; the camera is arranged on the upper part of the cabinet and shoots downward; when the door of the vending machine is opened, the door and the cabinet form a certain angle; the camera captures the image of the angle area.
[0018] Preferably, one side of the algorithm recognition area is the storage area of the vending machine, and the other side of the algorithm recognition area is the external standing area; during a retrieval process,
[0019] The process of the human hand entering the algorithm recognition area from the external standing area is marked as t1;
[0020] The process of the human hand entering the storage area from the algorithm recognition area is marked as t2;
[0021] The process of the human hand entering the algorithm recognition area from the storage area is marked as t3;
[0022] The process of the human hand entering the external standing area from the algorithm recognition area is marked as t4;
[0023] The vending machine uses the local AI image recognition algorithm to determine whether the hand is empty or holding goods during the t1 process; if the hand is empty, only the video data of the t4 process is written into the local storage card in real time.
[0024] Compared with the prior art, the present invention has the following outstanding and beneficial technical effects:
[0025] This patent uses local AI image algorithms in video acquisition to automatically filter out useless data (such as parts where the user has no operation), thereby reducing the size of video files, reducing upload time and 4G traffic consumption, while also shortening the time required for cloud service AI image recognition, greatly shortening the order settlement time. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a schematic diagram of the order settlement process in the prior art.
[0027] Figure 2 This is a schematic diagram of the order settlement process of this patent.
[0028] Figure 3 It is a schematic diagram of one-time object extraction of this patent. DETAILED DESCRIPTION
[0029] The present invention will be further described below in conjunction with specific embodiments:
[0030] A traffic saving settlement method based on cloud recognition image cabinet, comprising:
[0031] Step 1: When the door of the device is opened, start the camera to start video acquisition;
[0032] Step 2: The device starts the local AI image recognition algorithm to detect whether the hand is inserted;
[0033] If a hand is detected, the data collected by the camera is saved locally;
[0034] If no hand is detected, the data collected by the camera will not be saved locally;
[0035] Step 3: When the user closes the door, the camera and local AI image recognition algorithm are turned off, the door closing is reported to the cloud server, and the locally saved video file is uploaded;
[0036] Step 4: The cloud service sends the received video file to the cloud AI image recognition algorithm for product identification;
[0037] Step 5: Settle the order based on the recognition results.
[0038] Preferably, an algorithm recognition area is set in the video image captured by the camera, and when a hand is recognized to enter the algorithm recognition area, the captured video data is written to the local storage card in real time; when a hand is recognized to exit the algorithm recognition area, writing the captured video data to the local storage card is stopped.
[0039] Preferably, the device is a vending machine; the camera 300 is arranged on the upper part of the cabinet 200 and shoots downward; when the door 100 of the vending machine is opened, the door 100 and the cabinet 200 form a certain angle; the camera collects images of the angle area.
[0040] Preferably, one side of the algorithm recognition area A2 is the storage area A3 of the vending machine, and the other side of the algorithm recognition area is the external standing area A1; during a retrieval process,
[0041] The process of the human hand entering the algorithm recognition area A2 from the external standing area A1 is marked as t1;
[0042] The process of the human hand entering the storage area A3 from the algorithm recognition area A2 is marked as t2;
[0043] The process of the human hand entering the algorithm recognition area A2 from the storage area A3 is marked as t3;
[0044] The process of the human hand entering the external standing area A1 from the algorithm recognition area A2 is marked as t4;
[0045] The vending machine uses the local AI image recognition algorithm to determine whether the hands are empty or holding goods during the t1 process; in the case of empty hands, only the video data of the t4 process is written to the local storage card in real time, thereby further reducing the video file size.
[0046] The video file from the time the user opens the door to the time the user closes the door actually contains a lot of useless data, such as data when the user does not perform any operation during shopping; the video file is relatively large, and it takes a long time to upload to the cloud service, which consumes more traffic and also increases the cloud AI image recognition time; after the user scans the code (or scans the face) to open the door, the terminal device turns on the camera to start recording, and at the same time turns on the local AI image recognition to detect whether the hand is inserted. When the hand is not detected, the data collected by the camera is not saved locally, and the data collected by the camera is not saved locally until the hand is detected. The above steps are repeated until the user closes the door, then the camera is turned off and the closing of the door and the collected video file are reported and uploaded. When the cloud service receives the closing of the door and the video file, it is sent to the cloud AI image recognition calculation, and finally the order is settled. By detecting whether the hand is inserted through local AI image recognition, the camera collection can automatically filter out useless data, thereby reducing the size of the video file saved locally.
[0047] The above embodiments are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for saving traffic based on cloud recognition image cabinet, characterized in that: include: Step 1: When the door of the device is opened, start the camera to start video acquisition; Step 2: The device starts the local AI image recognition algorithm to detect whether the hand is inserted; If a hand is detected, the data collected by the camera is saved locally; If no hand is detected, the data collected by the camera will not be saved locally; Step 3: When the user closes the door, the camera and local AI image recognition algorithm are turned off, the door closing is reported to the cloud server, and the locally saved video file is uploaded; Step 4: The cloud service sends the received video file to the cloud AI image recognition algorithm for product identification; Step 5: settle the order based on the recognition result; An algorithm recognition area is set in the video image captured by the camera, and when a hand is recognized to enter the algorithm recognition area, the captured video data is written to the local storage card in real time; when a hand is recognized to exit the algorithm recognition area, the captured video data is stopped from being written to the local storage card; The device is an automatic vending machine; the camera is arranged on the upper part of the cabinet and shoots downward; when the door of the automatic vending machine is opened, the door and the cabinet form a certain angle; the camera collects images of the angle area; One side of the algorithm recognition area is the storage area of the vending machine, and the other side of the algorithm recognition area is the external standing area; in the process of taking an object, the process of a human hand entering the algorithm recognition area from the external standing area is marked as t1; The process of the human hand entering the storage area from the algorithm recognition area is marked as t2; The process of the human hand entering the algorithm recognition area from the storage area is marked as t3; The process of a human hand entering the external standing area from the algorithm recognition area is marked as t4; the vending machine determines through the local AI image recognition algorithm whether the hand is empty or holding goods during the t1 process; in the case of empty hands, only the video data of the t4 process is written into the local storage card in real time.
