CN117765540A - Liquid crystal water meter reading identification method and system based on deep learning - Google Patents
Liquid crystal water meter reading identification method and system based on deep learning Download PDFInfo
- Publication number
- CN117765540A CN117765540A CN202410072220.9A CN202410072220A CN117765540A CN 117765540 A CN117765540 A CN 117765540A CN 202410072220 A CN202410072220 A CN 202410072220A CN 117765540 A CN117765540 A CN 117765540A
- Authority
- CN
- China
- Prior art keywords
- liquid crystal
- water meter
- crystal water
- reading frame
- deep learning
- 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
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 89
- 239000004973 liquid crystal related substance Substances 0.000 title claims abstract description 88
- 238000013135 deep learning Methods 0.000 title claims abstract description 23
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000013136 deep learning model Methods 0.000 claims abstract description 31
- 238000002372 labelling Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 4
- 230000008520 organization Effects 0.000 abstract description 2
- 230000000007 visual effect Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image reading, in particular to a liquid crystal water meter reading identification method and a liquid crystal water meter reading identification system based on deep learning, wherein the identification method comprises the following steps of establishing a yolov5 deep learning model I capable of identifying a liquid crystal water meter reading frame and a dial plate; establishing a yolov5 deep learning model II capable of identifying digital texts in a reading frame; correcting the image of the liquid crystal water meter to be identified through the model I to obtain a corrected liquid crystal water meter reading frame picture; and carrying out digital text recognition on the reading frame picture by adopting the model II. The invention utilizes the computer vision technology and the deep learning method to accurately and intelligently identify the reading value of the liquid crystal water meter in place of the manual mode, thereby saving the cost and improving the efficiency. The invention has strong scene adaptability and high accuracy, and provides reliable metering basis for related organization personnel.
Description
Technical Field
The invention relates to the technical field of image reading, in particular to a liquid crystal water meter reading identification method and system based on deep learning.
Background
Computer vision technology is a great important technical direction in the field of machine learning in recent years. The research content is to explore how a computer extracts important information through visual image means and uses the information. Research goal of computer vision technology: firstly, an image algorithm simulating human is designed by analyzing the acquired image information, and the system is intelligent and automatic in an intelligent substitution manual mode; secondly, the collected image information is directly analyzed end to end by using a deep learning algorithm which is popular in recent years as a tool. The information collected by the camera contains rich features. Under this age of intelligence, analysis of visual images has also become an important application of deep learning algorithms. Visual algorithms based on deep learning bring great convenience to modern society, both for intelligent management and for future data and trend prediction. The liquid crystal water meter digital identification method based on deep learning is introduced herein, and aims to reduce labor cost and inaccuracy of manual meter reading.
At present, the numerical value of the water meter is read manually, namely, a meter reader reads the numerical value by going to the door, and because of factors of some environments, the risk of error reading of the numerical value is caused, and the labor cost is also caused; on the other hand, the traditional image algorithm is used for identifying the liquid crystal number on the liquid crystal water meter, and the liquid crystal number is greatly influenced by the environment, so that the instability of the algorithm is enhanced. In summary, it is proposed herein to use a deep learning algorithm to identify readings of a liquid crystal water meter.
Disclosure of Invention
In order to solve the technical problems, the invention provides a liquid crystal water meter reading identification method based on deep learning, which comprises the following steps,
s1, establishing a yolov5 deep learning model I capable of identifying a reading frame and a dial of a liquid crystal water meter;
s2, establishing a yolov5 deep learning model II capable of identifying digital texts in a reading frame;
s3, correcting the image of the liquid crystal water meter to be identified through the model I to obtain a corrected liquid crystal water meter reading frame picture; and carrying out digital text recognition on the reading frame picture by adopting the model II.
Further, the yolov5 deep learning model I is established by the following method,
acquiring an original liquid crystal water meter picture;
marking data of original liquid crystal water meter pictures, and marking a reading frame and a dial plate for each water meter picture to form a marking file I;
the original liquid crystal water meter picture and the labeling file are input into a yolov5 deep learning model for training to obtain a yolov5 deep learning model I, a plurality of rounds of iterative training can be carried out in the training process, each round of iteration can obtain a residual error value, and the parameter obtained by one round of training with the minimum residual error value is the optimal model parameter.
Further, the yolov5 deep learning model II is established by the following method,
the training yolov5 deep learning model I is used for identifying a reading frame and a dial plate of a liquid crystal water meter picture to be processed, and coordinate information of the reading frame and the identification frame of the dial plate is obtained;
calculating the center points of the reading frame and the dial according to the coordinate information;
calculating the angle difference between the symmetry axis of the liquid crystal water meter picture to be processed and the Y axis of the image according to the calculated center points of the reading frame and the dial plate, translating the center point of the reading frame to coincide with the center point of the image, correcting the liquid crystal water meter picture, and obtaining the corrected picture of the liquid crystal water meter reading frame; in this scheme, the picture refers to a file, and the image is the content on the picture.
Screening out clear pictures of the reading frame of the liquid crystal water meter after the correction, marking the pictures to form a marking file II,
and inputting the picture and the labeling file II of the reading frame of the liquid crystal water meter after the correction into a yolov5 deep learning model for training, wherein the model has the optimal model parameters.
Further, the pictures of the reading frame of the liquid crystal water meter are marked to form a marking file II, wherein the marking file II comprises numerals from 0 to 9 and 10 numerals with decimal points from 0 to 9; "0." - "9." means the sign of the decimal place, i.e. the unit digit, of the liquid crystal meter reading.
Further, the labeling rule is that "0" is labeled 0, "1" is labeled 1, "2" is labeled 2, "3" is labeled 3, "4" is labeled 4, "5" is labeled 5, "6" is labeled 6, "7" is labeled 7, "8" is labeled 8, "9" is labeled 9; "0." is denoted by 10, "1." is denoted by 11, "2." is denoted by 12, "3." is denoted by 13, "4." is denoted by 14, "5." is denoted by 15, "6." is denoted by 16, "7." is denoted by 17, "8." is denoted by 18, and "9." is denoted by 19.
According to another aspect of the invention, a liquid crystal water meter reading identification system based on deep learning is provided, which comprises a model I building unit, a yolov5 deep learning model I capable of identifying a liquid crystal water meter reading frame and a dial plate is built;
the model II building unit is used for building a yolov5 deep learning model II capable of identifying the digital text in the reading frame;
the identification unit is used for correcting the image of the liquid crystal water meter to be identified through the model I to obtain a corrected liquid crystal water meter reading frame picture; and carrying out digital text recognition on the reading frame picture by adopting the model II.
The invention utilizes the computer vision technology and the deep learning method to accurately and intelligently identify the reading value of the liquid crystal water meter in place of the manual mode, thereby saving the cost and improving the efficiency. The invention has strong scene adaptability and high accuracy, and provides reliable metering basis for related organization personnel.
Drawings
Fig. 1 is a schematic flow chart of the present invention.
Detailed Description
The principles and features of the present invention are described below with reference to the drawings, the examples are illustrated for the purpose of illustrating the invention and are not to be construed as limiting the scope of the invention.
Referring to fig. 1, the invention provides a liquid crystal water meter reading identification method based on deep learning, which comprises the following steps,
s1, establishing a yolov5 deep learning model I capable of identifying a reading frame and a dial of a liquid crystal water meter; specifically comprises the following steps of,
and screening the obtained original liquid crystal water meter picture, selecting a clear and complete picture sample, and ensuring the diversity and the balance of the sample.
Marking data of original liquid crystal water meter pictures, and marking a reading frame and a dial plate for each water meter picture to form a marking file I;
and inputting the original liquid crystal water meter picture and the labeling file I into a yolov5 deep learning model for training to obtain the yolov5 deep learning model I capable of identifying the reading frame and the dial of the liquid crystal water meter, wherein the model I has optimal model parameters.
S2, establishing a yolov5 deep learning model II capable of identifying digital texts in a reading frame;
the yolov5 deep learning model II is built by,
the training yolov5 deep learning model I is used for identifying a reading frame and a dial plate of a liquid crystal water meter picture to be processed, and coordinate information of the reading frame and the identification frame of the dial plate is obtained;
calculating the center points of the reading frame and the dial according to the coordinate information;
calculating the angle difference between the symmetry axis of the liquid crystal water meter picture to be processed and the Y axis of the image according to the calculated center points of the reading frame and the dial plate, translating the center point of the reading frame to coincide with the center point of the image, correcting the liquid crystal water meter picture, and obtaining the corrected picture of the liquid crystal water meter reading frame;
screening out clear pictures of the reading frame of the liquid crystal water meter after the correction, marking the pictures to form a marking file II,
the picture of the reading frame of the liquid crystal water meter is marked to form a marking file II, which comprises numbers marked with 0-9 and 10 numbers with decimal points from '0' - '9'; "0." - "9." means the sign of the decimal place, i.e. the unit digit, of the liquid crystal meter reading.
The labeling rule is that "0" on a liquid crystal water meter reading frame picture is labeled 0, "1" is labeled 1, "2" is labeled 2, "3" is labeled 3, "4" is labeled 4, "5" is labeled 5, "6" is labeled 6, "7" is labeled 7, "8" is labeled 8, and "9" is labeled 9; "0." is denoted by 10, "1." is denoted by 11, "2." is denoted by 12, "3." is denoted by 13, "4." is denoted by 14, "5." is denoted by 15, "6." is denoted by 16, "7." is denoted by 17, "8." is denoted by 18, and "9." is denoted by 19.
S3, correcting the image of the liquid crystal water meter to be identified through the model I to obtain a corrected liquid crystal water meter reading frame picture; and carrying out digital text recognition on the reading frame picture by adopting the model II.
According to another aspect of the present invention, there is provided a liquid crystal water meter reading identification system based on deep learning, comprising
The model I building unit is used for building a yolov5 deep learning model I capable of identifying a reading frame of the liquid crystal water meter and a dial plate;
the model II building unit is used for building a yolov5 deep learning model II capable of identifying the digital text in the reading frame;
the identification unit is used for correcting the image of the liquid crystal water meter to be identified through the model I to obtain a corrected liquid crystal water meter reading frame picture; and carrying out digital text recognition on the reading frame picture by adopting the model II.
In this scheme, the first purpose of this patent is discernment reading frame and the dial plate of liquid crystal water gauge, translates the reading frame central point of liquid crystal water gauge to the picture center and just turns over.
The technical scheme for realizing the first purpose of the invention is as follows: marking and training a reading frame and a dial of a liquid crystal water meter by using a deep learning method, fully learning characteristics of a sample to obtain a target model, identifying the reading frame and the dial of the liquid crystal water meter according to the model, determining the angle difference between a symmetry axis and an image y axis of the water meter according to the central positions of the reading frame and the dial, translating a central point of the reading frame to a position overlapped with the central point of the image, and then rotating a picture to enable the reading frame to be positioned at the central point of the image and correct the picture of the water meter so as to facilitate the reading identification at the back;
the second purpose of this patent is to perform text detection and text recognition on the readings of the liquid crystal water meter.
The technical scheme for realizing the second purpose of the invention is as follows: based on the first purpose, the numerical values in the reading frame are marked and trained by using a deep learning network, and the characteristics of the sample are fully learned to obtain a target model. And then the model is used for processing according to the image of the liquid crystal water meter to be identified, and a series of image processing methods are used for accurately obtaining the identification result of the liquid crystal water meter and providing the identification result for related institution personnel to serve as reference measurement. Practice proves that the reading identification method of the liquid crystal water meter can accurately identify the reading value of the liquid crystal water meter on the premise that the image of the liquid crystal water meter is not particularly unclear.
The foregoing description of the preferred embodiments of the invention is not intended to limit the invention to the precise form disclosed, and any such modifications, equivalents, and alternatives falling within the spirit and scope of the invention are intended to be included within the scope of the invention.
Claims (6)
1. A liquid crystal water meter reading identification method based on deep learning is characterized by comprising the following steps,
s1, establishing a yolov5 deep learning model I capable of identifying a reading frame and a dial of a liquid crystal water meter;
s2, establishing a yolov5 deep learning model II capable of identifying digital texts in a reading frame;
s3, correcting the image of the liquid crystal water meter to be identified through the model I to obtain a corrected liquid crystal water meter reading frame picture; and carrying out digital text recognition on the reading frame picture by adopting the model II.
2. The method for recognizing readings of a liquid crystal water meter based on deep learning according to claim 1, wherein the yolov5 deep learning model I is established by the following method,
acquiring an original liquid crystal water meter picture;
marking data of original liquid crystal water meter pictures, and marking a reading frame and a dial plate for each water meter picture to form a marking file I;
and inputting the original liquid crystal water meter picture and the labeling file into a yolov5 deep learning model for training to obtain a yolov5 deep learning model I, and obtaining optimal model parameters.
3. The method for recognizing readings of liquid crystal water meter based on deep learning as set forth in claim 1, wherein said yolov5 deep learning model II is established by,
the training yolov5 deep learning model I is used for identifying a reading frame and a dial plate of a liquid crystal water meter picture to be processed, and coordinate information of the reading frame and the identification frame of the dial plate is obtained;
calculating the center points of the reading frame and the dial according to the coordinate information;
calculating the angle difference between the symmetry axis of the liquid crystal water meter picture to be processed and the Y axis of the image according to the calculated center points of the reading frame and the dial plate, translating the center point of the reading frame to coincide with the center point of the image, correcting the liquid crystal water meter picture, and obtaining the corrected picture of the liquid crystal water meter reading frame;
screening out clear pictures of the reading frame of the liquid crystal water meter after the correction, marking the pictures to form a marking file II,
and inputting the picture and the labeling file II of the reading frame of the liquid crystal water meter after the correction into a yolov5 deep learning model for training, wherein the yolov5 deep learning model II has the optimal model parameters.
4. The method for recognizing the reading of the liquid crystal water meter based on the deep learning according to claim 1, wherein the labeling of the pictures of the reading frame of the liquid crystal water meter forms a labeling file II, and the labeling includes labeling numbers of 0-9 and numbers with decimal points of ' 0 ' - ' 9 ' - ' 10; "0." - "9." means the sign of the decimal place, i.e. the unit digit, of the liquid crystal meter reading.
5. The method for identifying readings of the liquid crystal water meter based on deep learning according to claim 4, wherein the labeling rule is that "0" on a picture of a reading frame of the liquid crystal water meter is labeled 0, "1" is labeled 1, "2" is labeled 2, "3" is labeled 3, "4" is labeled 4, "5" is labeled 5, "6" is labeled 6, "7" is labeled 7, "8" is labeled 8, "9" is labeled 9; "0." is denoted by 10, "1." is denoted by 11, "2." is denoted by 12, "3." is denoted by 13, "4." is denoted by 14, "5." is denoted by 15, "6." is denoted by 16, "7." is denoted by 17, "8." is denoted by 18, and "9." is denoted by 19.
6. A liquid crystal water meter reading identification system based on deep learning is characterized by comprising
The model I building unit is used for building a yolov5 deep learning model I capable of identifying a reading frame of the liquid crystal water meter and a dial plate;
the model II building unit is used for building a yolov5 deep learning model II capable of identifying the digital text in the reading frame;
the identification unit is used for correcting the image of the liquid crystal water meter to be identified through the model I to obtain a corrected liquid crystal water meter reading frame picture; and carrying out digital text recognition on the reading frame picture by adopting the model II.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410072220.9A CN117765540A (en) | 2024-01-18 | 2024-01-18 | Liquid crystal water meter reading identification method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410072220.9A CN117765540A (en) | 2024-01-18 | 2024-01-18 | Liquid crystal water meter reading identification method and system based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117765540A true CN117765540A (en) | 2024-03-26 |
Family
ID=90322024
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410072220.9A Pending CN117765540A (en) | 2024-01-18 | 2024-01-18 | Liquid crystal water meter reading identification method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117765540A (en) |
-
2024
- 2024-01-18 CN CN202410072220.9A patent/CN117765540A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110659636B (en) | Pointer instrument reading identification method based on deep learning | |
| CN102521560B (en) | Instrument pointer image identification method of high-robustness rod | |
| CN109635806B (en) | Ammeter value identification method based on residual error network | |
| CN112949564A (en) | Pointer type instrument automatic reading method based on deep learning | |
| CN111950396B (en) | Meter reading neural network identification method | |
| CN110175502A (en) | A kind of backbone Cobb angle measuring method, device, readable storage medium storing program for executing and terminal device | |
| CN112801094B (en) | Pointer instrument image inclination correction method | |
| CN108764134A (en) | A kind of automatic positioning of polymorphic type instrument and recognition methods suitable for crusing robot | |
| CN111639629B (en) | Pig weight measurement method and device based on image processing and storage medium | |
| CN116434266A (en) | Automatic extraction and analysis method for data information of medical examination list | |
| CN111368906A (en) | Pointer type oil level indicator reading identification method based on deep learning | |
| CN108197564A (en) | A kind of assessment system and method for drawing clock experiment | |
| CN110427943A (en) | A kind of intelligent electric meter technique for partitioning based on R-CNN | |
| CN115984862A (en) | Deep learning-based remote water meter digital identification method | |
| CN113283466A (en) | Instrument reading identification method and device and readable storage medium | |
| CN115019294A (en) | Pointer instrument reading identification method and system | |
| CN112529003A (en) | Instrument panel digital identification method based on fast-RCNN | |
| CN112115834A (en) | Standard certificate photo detection method based on small sample matching network | |
| CN109784257B (en) | Transformer thermometer detection and identification method | |
| CN117765540A (en) | Liquid crystal water meter reading identification method and system based on deep learning | |
| CN109886320B (en) | Human femoral X-ray intelligent recognition method and system | |
| CN114926841B (en) | Electronic water meter reading image identification method and device based on improved threading method | |
| CN116612265A (en) | Target encoding and decoding-based dial pointer reading identification method | |
| CN115588196A (en) | Pointer type instrument reading method and device based on machine vision | |
| CN113076941A (en) | Single pointer dial reading identification method based on video frame fusion |
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 |