CN112763041A - Fresh food identification weighing electronic scale and identification method - Google Patents
Fresh food identification weighing electronic scale and identification method Download PDFInfo
- Publication number
- CN112763041A CN112763041A CN202110006668.7A CN202110006668A CN112763041A CN 112763041 A CN112763041 A CN 112763041A CN 202110006668 A CN202110006668 A CN 202110006668A CN 112763041 A CN112763041 A CN 112763041A
- Authority
- CN
- China
- Prior art keywords
- camera
- android
- tray
- identification
- weight
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01G—WEIGHING
- G01G19/00—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups
- G01G19/40—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups with provisions for indicating, recording, or computing price or other quantities dependent on the weight
- G01G19/413—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups with provisions for indicating, recording, or computing price or other quantities dependent on the weight using electromechanical or electronic computing means
- G01G19/414—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups with provisions for indicating, recording, or computing price or other quantities dependent on the weight using electromechanical or electronic computing means using electronic computing means only
- G01G19/415—Weighing apparatus or methods adapted for special purposes not provided for in the preceding groups with provisions for indicating, recording, or computing price or other quantities dependent on the weight using electromechanical or electronic computing means using electronic computing means only combined with recording means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention relates to the field of fresh food identification weighing equipment, in particular to a fresh food identification weighing electronic scale which comprises a tray, wherein a weight sensor is connected to the bottom end of the tray, the weight sensor is connected with an android board, the android board and the weight sensor are installed in a base, the base is fixedly provided with a cylindrical installation frame, a code printer is installed on the cylindrical installation frame, the code printer is connected with an android display screen, a camera is rotatably connected to the cylindrical installation frame, and the camera is arranged towards the right lower side. The device is used based on the identification technology of video stream, and does not need an AI chip, a chip module or an AI server. Besides the camera, no additional hardware is provided, and the cost advantage is very obvious.
Description
Technical Field
The invention relates to the field of fresh food identification weighing equipment, in particular to a fresh food identification weighing electronic scale and an identification method.
Background
For a traditional electronic scale, what needs to be done when a weigher takes a commodity is: 1. the brain identifies which item is; 2. then putting the commodity on an electronic scale; 3. thinking the PLU code corresponding to the commodity; 4. inputting a commodity PLU code; 5. and finally, labeling.
In the whole link, the factors of human participation are more, and a steelyard needs to carry 100 plus 200 bar codes or even more. When more customers are available in rush hours, the traditional electronic scale can only be operated by a weigher, so that the queuing phenomenon is serious and the efficiency is influenced. In view of this, electronic scales based on artificial intelligence are introduced in the industry, and are limited by the factor of "AI requires huge calculation force", and such electronic scales are mainly realized by two schemes: (1) using an AI server, wherein a plurality of electronic scales correspond to one AI server, and the electronic scales are in data interaction with the AI server through a network; (2) the AI chip is additionally arranged on the electronic scale to cooperate with a local CPU for calculation, which is commonly called as 'edge calculation'.
The first solution is costly and cumbersome to communicate. And factors such as algorithm model upgrading are considered, so that the software maintenance cost is high. The second scheme is that either a chip or a chip module is additionally introduced, wherein the former circuit hardware structure of the electronic scale is changed, the cost is huge, and the additionally introduced driving program is troublesome to change. The latter chip module has similar disadvantages as the first solution.
Disclosure of Invention
The invention aims to provide a fresh food identification weighing electronic scale to solve the problems of the two schemes pointed out in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme:
the utility model provides a give birth to bright discernment electronic scale of weighing, includes the tray, the tray bottom is connected with weighing sensor, weighing sensor connects tall and erect the board of ann, tall and erect board and weighing sensor and install in the base, the base is fixed with the cylindricality mounting bracket, install the coding machine on the cylindricality mounting bracket, the coding machine is connected with tall and erect display screen of ann, it is connected with the camera to rotate on the cylindricality mounting bracket, the camera sets up under towards.
As a further scheme of the invention: and the weight sensor is connected with a usb interface of the android board through a serial port usb.
As a further scheme of the invention: the android board and the camera are communicated by adopting a standard UVC protocol, the camera is installed at the top end of the cylindrical mounting frame, the camera adopts a ninety-degree lens, and the camera lens shooting range covers the whole tray.
As a further scheme of the invention: the classification algorithm of the android is based on deep learning CNN network ALexNet and further cutting a network structure to the extent that the algorithm can run on a common arm chip.
As a further scheme of the invention: the camera is installed in the tall and erect display screen bottom of ann, the camera adopts the ninety degree camera lens, the whole tray is covered to camera lens scope of making a video recording.
A fresh food identification method uses the electronic scale, and the method comprises the following steps: weight sensor (7) detect tray (1) weight, when weight transmission changes, weight sensor (7) send signal for android board (8), android board (8) control camera (6) camera begins work and shoots the article on tray (1), and with the video stream data transmission of shooing for android board (8), android board (8) carry out the quality analysis of target with the video stream of receiving first, filter behind the unqualified image, discern the image that will meet the demands through giving birth to bright kind identification algorithm and map into corresponding edible material name with the recognition result, and match with the weight data that weight sensor transmitted, send for coding machine (4) and android display screen (5) after generating corresponding price, coding machine (4) print the bar code, android display screen (5) show commodity kind, weight and price.
As a further scheme of the invention: the weight sensor (7) detects the weight of the tray (1), when the weight is sent and changed, the added value of the weight of the tray (1) needs to be compared with a set value, and when the added value is larger than the set value, the android board (8) controls the camera of the camera (6) to start working.
As a further scheme of the invention: the specific process of identifying the images meeting the requirements through the fresh food type identification algorithm and mapping the identification result into the corresponding food material name is as follows:
extracting the characteristics { xi } of the image which meets the requirement through a CNN network ALexNet, calculating a confidence coefficient formula (1) by using softmax, sampling RGB pixel values of the image which meets the requirement at intervals, and calculating the confidence coefficient by using softmax through a full connection layer. Then, two confidence values are obtained, and the final confidence value is obtained in a weighted summation mode as shown in a formula (2);
Pi＝a·Si+β·S′i (2)
where (i, j ═ 1,2.. n), n is the dimension representing CNN feature { xi } and also the dimension representing RGB color feature { yi }; in the actual use scene, n represents the number of the categories of the food materials; si represents the probability, S ', when the food material in the current picture is identified as the category i by using the ' CNN network ALexNet 'iThe probability of identifying the food material in the current picture as the category i by using the RGB color pixel value is shown, and Pi represents the comprehensive probability obtained by combining the two schemes;
α and β are empirical values that represent weighting coefficients and are preset.
The fresh food identification weighing electronic scale provided by the embodiment of the invention is used based on the identification technology of video stream, and does not need an AI chip, a chip module or an AI server. Besides the camera, no additional hardware is provided, and the cost advantage is very obvious.
Drawings
FIG. 1 is a flow chart of a system operation of a fresh food identification weighing electronic scale according to an embodiment of the present invention;
FIG. 2 is a block diagram;
FIG. 3 is a flow chart of a business logic algorithm;
fig. 4 is an algorithm flow chart.
In the figure: 1. a tray; 2. a base; 3. a cylindrical mounting bracket; 4. a coding machine; 5. an android display screen; 6. a camera; 7. a weight sensor; 8. and (5) performing android plate.
Detailed Description
The technical solution of the present patent will be described in further detail with reference to the following embodiments.
Reference will now be made in detail to embodiments of the present patent, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are exemplary only for the purpose of explaining the present patent and are not to be construed as limiting the present patent.
Example (b):
as shown in fig. 1-4, in the embodiment of the invention, a fresh food identification weighing electronic scale comprises a tray 1, a weight sensor 7 is connected to the bottom end of the tray 1, the weight sensor 7 is connected with an android board 8, the android board 8 and the weight sensor 7 are installed in a base 2, a cylindrical mounting rack 3 is fixed on the base 2, a code printer 4 is installed on the cylindrical mounting rack 3, the code printer 4 is connected with an android display screen 5, a camera 6 is rotatably connected to the cylindrical mounting rack 3, and a camera 6 is arranged under the camera.
And the weight sensor 7 is connected with a usb interface of the android board 8 through a serial port usb.
Android board 8 and camera 6 adopt standard UVC agreement to communicate, camera 6 installs in 3 tops of cylindricality mounting bracket, camera 6 adopts the ninety degrees camera lens, whole tray 1 is covered to 6 camera lens scope of making a video recording of camera.
The classification algorithm of the android board 8 is based on deep learning CNN network ALexNet and further cutting a network structure to the algorithm which can be operated on a common arm chip.
The camera 6 is installed in 5 bottoms of tall and erect display screens of ann, camera 6 adopts the ninety degree camera lens, the camera 6 camera lens scope of making a video recording covers whole tray 1.
In the embodiment of the invention, the whole set of hardware system is arranged in an embedded mode according to the structure of an actual product. The system work flow chart is shown in figure 1: (1) after the system is powered on, the android board starts to work to detect signals of 'fresh weighing', and the state is set as 'state A'. (2) Suppose that a living thing is weighed at a certain moment, and after the android board acquires a signal of the sensor, the camera is started randomly to shoot. (3) The camera transmits the video stream data to the android board frame by frame. (4) The android board performs quality analysis on the received video stream firstly, and the purpose is mainly to filter some unqualified images. (5) And when the image quality of a certain frame meets a set threshold value, the frame is sent to a fresh type identification algorithm for identification. (6) The android board is responsible for mapping the received identification result into a corresponding food material name, converting the weight data transmitted by the weight sensor into price, displaying the price through the APP interface, and informing the bar code printing equipment of printing the bar code. (7) The android panel resets to state a again.
The relationship between the android board and the weight sensor is as follows: the weight sensor is connected to the usb port of the android board through the serial port transfer usb, the weight sensor sends weight data all the time after power-on, and the android board is obtained in real time.
The relationship between the android board and the camera is as follows: the android board and the camera adopt a standard UVC protocol for communication. The camera adopts a 90-degree lens, so that the shooting range can cover the width of the whole scale pan. Chip parameters of the camera are optimized to a certain extent, exposure time is set to be very short, and motion blur during photographing is prevented; the aperture of the lens is set to be larger to ensure the light incoming quantity, thereby ensuring the brightness of the picture. There are two main ways of positioning the camera. Wherein the camera is mounted in the position shown in figure 2.
The main logic of the android board is shown in fig. 3, regarding the part of the logic of the software algorithm:
step one, sending the frame image to a quality judgment algorithm, training the quality judgment algorithm based on another CNN network to obtain a quality value C, comparing the value C with a manually preset threshold value, and identifying the fresh type if the value C is larger than a certain value, wherein the quality of the frame image is in line with the expected quality.
And step two, detecting the fresh position of the current video frame by adopting a target detection algorithm based on deep learning CNN, acquiring the position, and then, obtaining the region crop of the position, and further identifying and the like. It is worth to be noted that the quality judgment algorithm is a very small deep learning network, so that the real-time performance is guaranteed.
The camera recognition algorithm:
the classification algorithm is based on deep learning CNN network ALexNet, and further the network structure is cut to the extent that the algorithm can run on a common arm chip. It should be noted that the target region crop is output in advance before being input into the classification algorithm.
The algorithm is shown in FIG. 4, the original graph is subjected to CNN network ALexNet to extract features { xi } and then the confidence coefficient formula 1 is calculated by using softmax; and the original image adopts the RGB pixel values sampled at intervals, and the confidence coefficient is calculated through a full connection layer and softmax. There are two confidence values, and we use weighted summation to get the final confidence as formula 2. The purpose of this is that the classification information is more accurate by using a method based on artificial 'strengthening' color features because the identification of fresh fruits and vegetables depends on the color features to a great extent.
Pi＝α·Si+β·S′j (2)
Where (i, j ═ 1,2.. n), n is the dimension representing CNN feature { xi } and also the dimension representing RGB color feature { yi }; in the actual use scene, n represents the number of the categories of the food materials; si represents the probability, S ', when the food material in the current picture is identified as the category i by using the ' CNN network ALexNet 'iThe probability of identifying the food material in the current picture as the category i by using the RGB color pixel value is shown, and Pi represents the comprehensive probability obtained by combining the two schemes;
α and β are empirical values that represent weighting coefficients and are preset. In practice, if the color characteristics of a batch of samples are particularly obvious, β can be set to be larger. In actual use, the app presents the top 8 categories with the highest confidence. The user selects the corresponding correct category.
The embodiment of the invention provides a fresh food identification weighing electronic scale, which is used based on the identification technology of video stream, and does not need an AI chip, a chip module or an AI server. Besides the camera, no additional hardware is provided, and the cost advantage is very obvious.
The above is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, it is possible to make several variations and modifications without departing from the concept of the present invention, and these should be considered as the protection scope of the present invention, which will not affect the effect of the implementation of the present invention and the utility of the patent.
Claims (8)
1. The utility model provides a give birth to bright discernment electronic scale of weighing, includes tray (1), its characterized in that, tray (1) bottom is connected with weighing sensor (7), tall and erect board (8) of ann is connected weighing sensor (7), install in base (2) tall and erect board (8) and weighing sensor (7), base (2) are fixed with cylindricality mounting bracket (3), install coding machine (4) on cylindricality mounting bracket (3), coding machine (4) are connected with tall and erect display screen of ann (5), it is connected with camera (6) to rotate on cylindricality mounting bracket (3), camera (6) camera sets up under towards.
2. The fresh food identification weighing electronic scale according to claim 1, wherein the weight sensor (7) is connected with a usb interface of an android board (8) through a serial port usb.
3. Fresh food identification weighing electronic scale according to claim 2, characterized in that the android board (8) and the camera (6) communicate using a standard UVC protocol, the camera (6) is mounted on the top of the cylindrical mounting frame (3), the camera (6) uses a ninety-degree lens, and the camera (6) has a lens range covering the whole tray (1).
4. Fresh identification weighing electronic scale according to claim 3, characterized in that the classification algorithm of the android board (8) is based on deep learning CNN network ALexNet and further tailoring the network structure to algorithms that can run on common arm chips.
5. The fresh food identification weighing electronic scale according to claim 1, wherein the camera (6) is mounted at the bottom end of an android display screen (5), the camera (6) adopts a ninety-degree lens, and the shooting range of the camera (6) covers the whole tray (1).
6. A fresh food identification method using the electronic scale according to any one of claims 1 to 6, the method comprising: weight sensor (7) detect tray (1) weight, when weight transmission changes, weight sensor (7) send signal for android board (8), android board (8) control camera (6) camera begins work and shoots the article on tray (1), and with the video stream data transmission of shooing for android board (8), android board (8) carry out the quality analysis of target with the video stream of receiving first, filter behind the unqualified image, discern the image that will meet the demands through giving birth to bright kind identification algorithm and map into corresponding edible material name with the recognition result, and match with the weight data that weight sensor transmitted, send for coding machine (4) and android display screen (5) after generating corresponding price, coding machine (4) print the bar code, android display screen (5) show commodity kind, weight and price.
7. The freshness identification method according to claim 6, wherein: the weight sensor (7) detects the weight of the tray (1), when the weight is sent and changed, the added value of the weight of the tray (1) needs to be compared with a set value, and when the added value is larger than the set value, the android board (8) controls the camera of the camera (6) to start working.
8. The fresh food identification method according to claim 6, wherein the specific process of identifying the image meeting the requirement by the fresh food type identification algorithm and mapping the identification result to the corresponding food material name is as follows:
extracting features { xi } (i is 0,1,2. n) of the image meeting the requirements through a CNN network ALexNet, then calculating a confidence coefficient formula (1) by using softmax, sampling RGB pixel values of the image meeting the requirements at intervals, and calculating the confidence coefficient through a full connecting layer and softmax. Then, two confidence values are obtained, and the final confidence value is obtained in a weighted summation mode as shown in a formula (2);
Pi＝α·Si+β·S′i (2)
where (i, j ═ 1,2.. n), n is the dimension representing CNN feature { xi } and also the dimension representing RGB color feature { yi }; in the actual use scene, n represents the number of the categories of the food materials; si represents the probability, S ', when the food material in the current picture is identified as the category i by using the ' CNN network ALexNet 'iThe probability of identifying the food material in the current picture as the category i by using the RGB color pixel value is shown, and Pi represents the comprehensive probability obtained by combining the two schemes;
α and β are empirical values that represent weighting coefficients and are preset.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110006668.7A CN112763041A (en) | 2021-01-05 | 2021-01-05 | Fresh food identification weighing electronic scale and identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110006668.7A CN112763041A (en) | 2021-01-05 | 2021-01-05 | Fresh food identification weighing electronic scale and identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112763041A true CN112763041A (en) | 2021-05-07 |
Family
ID=75699218
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110006668.7A Pending CN112763041A (en) | 2021-01-05 | 2021-01-05 | Fresh food identification weighing electronic scale and identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112763041A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114157813A (en) * | 2022-02-07 | 2022-03-08 | 深圳市慧为智能科技股份有限公司 | Electronic scale camera motion control method and device, control terminal and storage medium |
-
2021
- 2021-01-05 CN CN202110006668.7A patent/CN112763041A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114157813A (en) * | 2022-02-07 | 2022-03-08 | 深圳市慧为智能科技股份有限公司 | Electronic scale camera motion control method and device, control terminal and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109863530B (en) | Method, system and computer-readable storage medium for determining grain weight of ears | |
| US8619340B2 (en) | Augmenting page orientation direction detection in a digital document processing environment | |
| CN108564079B (en) | Portable character recognition device and method | |
| CN106815560A (en) | It is a kind of to be applied to the face identification method that self adaptation drives seat | |
| CN112763041A (en) | Fresh food identification weighing electronic scale and identification method | |
| WO2024000927A1 (en) | Deep learning based method for automatic geological disaster extraction from unmanned aerial vehicle image | |
| CN110493595A (en) | The detection method and device of camera, storage medium and electronic device | |
| CN115302963B (en) | Bar code printing control method, system and medium based on machine vision | |
| CN111766248A (en) | Steel seal on-line detection system and method based on color CCD | |
| CN101751512A (en) | Recipe management system applied to communication device and method | |
| CN114612897A (en) | Intelligent fruit and vegetable weighing and ticketing method and device, electronic equipment and storage medium | |
| CN214066313U (en) | Give birth to bright discernment electronic scale of weighing | |
| CN102248254A (en) | Data processing system and method | |
| CN111507325A (en) | Industrial visual OCR recognition system and method based on deep learning | |
| CN116403197A (en) | Intelligent weighing method and system based on AI image recognition | |
| JPH09262027A (en) | Automatically diagnostic system for plant | |
| CN101151882A (en) | System and method of processing scan data | |
| US6493597B1 (en) | Print monitoring system and method using slave signal processor/master processor arrangement | |
| US20210176375A1 (en) | Information processing device, information processing system, information processing method and program | |
| CN114235122A (en) | Weighing settlement method and system of electronic scale based on AI image recognition | |
| CN115456984A (en) | High-speed image recognition defect detection system based on two-dimensional code | |
| US20230104061A1 (en) | Methods and apparatuses for determining statuses of positions on shelves, shelves and non-transitory computer-readable storage media | |
| CN111639928B (en) | Bulk commodity metering device and method based on convolutional neural network | |
| CN109660863B (en) | Visual attention area detection method, device, equipment and computer storage medium | |
| CN116843397A (en) | Management system and method of fruit and vegetable weighing and pricing device based on visual identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |