CN112001383A - Water meter code intelligent identification method based on convolutional neural network technology - Google Patents
Water meter code intelligent identification method based on convolutional neural network technology Download PDFInfo
- Publication number
- CN112001383A CN112001383A CN202010795175.1A CN202010795175A CN112001383A CN 112001383 A CN112001383 A CN 112001383A CN 202010795175 A CN202010795175 A CN 202010795175A CN 112001383 A CN112001383 A CN 112001383A
- Authority
- CN
- China
- Prior art keywords
- water meter
- image
- code
- meter code
- upright
- 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 173
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 17
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims abstract description 9
- 230000009466 transformation Effects 0.000 claims abstract description 8
- 238000007635 classification algorithm Methods 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 33
- 238000012795 verification Methods 0.000 claims description 21
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 230000003416 augmentation Effects 0.000 claims description 4
- 238000013145 classification model Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 4
- 238000004519 manufacturing process Methods 0.000 claims description 3
- 239000000428 dust Substances 0.000 description 3
- 238000012545 processing Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/146—Aligning or centring of the image pick-up or image-field
- G06V30/1475—Inclination or skew detection or correction of characters or of image to be recognised
- G06V30/1478—Inclination or skew detection or correction of characters or of image to be recognised of characters or characters lines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Abstract
The invention discloses an intelligent water meter code identification method based on a convolutional neural network technology, which comprises the following steps of: step one, carrying out image acquisition on one surface of the water meter with a water meter code; secondly, identifying and judging whether the water meter image is an upright image or an inverted image by using the collected water meter code in the water meter image as a standard through a classification algorithm, and obtaining model parameters for upright and inverted judgment of the water meter image; thirdly, recognizing a mask image of a water meter code of the water meter image, calculating an external torque, and performing affine transformation rotation according to the angle of the external torque to obtain a water meter code image in which the water meter code is placed according to a horizontal position; and fourthly, identifying the water meter code image to obtain the coordinates and numerical value information of each number in the water meter code. The intelligent water meter numerical value identification is realized, the error rate is low, and the identification is accurate.
Description
Technical Field
The invention relates to a meter reading method, in particular to an intelligent water meter code identification method based on a convolutional neural network technology.
Background
The traditional algorithm for identifying the water meter codes is very easy to be interfered by an external complex environment, and if a better accuracy rate is expected, the traditional algorithm has very high requirements and shooting limits on the dial plate pictures of the water meters.
Compared with a classical image processing algorithm, the image processing technology based on the convolutional neural network obtains great breakthrough and progress in various fields, various external factor interferences such as weather, environment, dust and the like often occur in the actual meter reading process, great challenge is provided for automatic water meter code identification, certain limitation often exists in manual design algorithm feature extraction, and the water meter code features can be extracted well by using the convolutional neural network and a back propagation algorithm through a deep learning technology and combining a large number of on-site irregular pictures.
Disclosure of Invention
The invention aims to provide an intelligent water meter code identification method based on a convolutional neural network technology.
The technical scheme for realizing the purpose of the invention is as follows: a water meter code intelligent identification method based on a convolutional neural network technology comprises the following steps:
step one, carrying out image acquisition on one surface of the water meter with a water meter code;
secondly, identifying and judging whether the water meter image is an upright image or an inverted image by using the collected water meter code in the water meter image as a standard through a classification algorithm, and obtaining model parameters for upright and inverted judgment of the water meter image;
thirdly, recognizing a mask image of a water meter code of the water meter image, calculating an external torque, and performing affine transformation rotation according to the angle of the external torque to obtain a water meter code image in which the water meter code is placed according to a horizontal position;
and fourthly, identifying the water meter code image to obtain the coordinates and numerical value information of each number in the water meter code.
The second step comprises the following specific steps:
s1, making upright and inverted classification data sets according to the water meter images collected in the step I.
And S2, training the network by using a GoogleNet network as a basis, modifying the last classification layer of the classification data set into 2 types, reducing the number of part of low-layer networks, halving the number on the basis of the original number, and not changing the classification data set from the fourth layer, training a classification model by using a back propagation algorithm, and expanding the classification data set by using random augmentation operation in combination with a label in the training process.
And S3, performing convolution operation on the collected water meter images and the trained network, and classifying the collected water meter images into output results by softmax.
And step two, manufacturing upright and inverted classification data sets according to the water meter images, wherein the upright and inverted judgment standards are as follows: the lower part of the water meter code is taken as a horizontal line, the angle of the horizontal line is inverted when the angle and the image horizontal line are between 90 and 180 degrees, and the other parts are upright.
The third step comprises the following specific steps:
and S1, marking the outlines of the water meter code, the pointer character wheel and the water meter dial in the water meter image in a mask mode, storing a rectangular coordinate storage file containing the outlines in a form of length, width and height by using the initial coordinates, and storing the coordinates of each outline point by using the outline coordinates.
S2, extracting features by using a resnet50 as a framework, training a network, and identifying the water meter code, the pointer character wheel and the outline result of the water meter dial by using a Mask-RCNN algorithm.
S3, combining the standard condition of the water meter characteristics and the upright or inverted classification result of the water meter image obtained in the step two;
if the image is the upright water meter image, the image is input;
if the water meter image is the inverted water meter image, calculating the external moment of the water meter code, and according to the water meter code and the horizontal line angle of the image; and rotating the inverted water meter image to a standard posture of the water meter code through affine transformation rotation and outputting a picture.
The fourth step comprises the following specific steps:
and S1, acquiring upright water meter images, manually marking the digital position of each picture to form a training set and a verification set, wherein the training set and the verification set respectively account for 90% and 10% of the total number of the pictures, and training by using a deep residual error network (ResNet-18).
And S2, selecting the parameters which are best represented on the verification set by the trained model for actual detection, and acquiring the coordinates and numerical values of the target character through the pictures output in the step three.
The fourth step further comprises the following specific steps:
s3, sorting all the character results according to the coordinates, calculating the interval between every two characters, considering the water meter code characteristics, the distances between the water meter codes are basically equal or proportional, and removing the obvious abnormal character interval according to the standard difference value by calculating the standard difference of the interval.
S4, according to the reasonable character coordinate array obtained from S3, if some coordinates are very close, the situation represents that a plurality of results are judged or hyphens appear, the two situations are processed, if no hyphen exists, a number with a high Faster-RCNN prediction probability is selected, if the hyphen appears, a larger number in the hyphen, such as 5_6, is directly selected, and the current number is 6.
By adopting the technical scheme, the invention has the following beneficial effects: the invention identifies the key area of the water meter code of the upright or inverted water meter image by acquiring the water meter image data and taking the digital code as the standard for upright and inverted judgment, takes out the water meter code by an external moment method, rotates the water meter code to the standard position by affine transformation, identifies a pointer character wheel on the water meter for digital code verification, further determines the upright and inverted conditions of the water meter by combining the image relationship of the character wheel and the water meter code, further corrects the rotation mode of the water meter code, obtains the digital position and classification result of each water meter code by using the image identification technology based on the convolutional neural network according to the rotated water meter code, corrects the final result by combining the priori knowledge of the water meter and using a linear fitting method, calculates the water meter code value, realizes intelligent water meter value identification, and has low error rate and accurate identification.
Drawings
In order that the present disclosure may be more readily and clearly understood, reference is now made to the following detailed description of the present disclosure taken in conjunction with the accompanying drawings, in which
FIG. 1 is a block flow diagram of the present invention;
FIG. 2 is a schematic view of the water meter code of the present invention rotated to a standard position;
FIG. 3 is a schematic representation of a hyphenated water meter code according to the present invention;
Detailed Description
Example one
Referring to fig. 1, the water meter code intelligent identification method based on the convolutional neural network technology of the embodiment includes the following steps:
step one, carrying out image acquisition on one surface of the water meter with a water meter code; shoot the water gauge of arbitrary angle through cell-phone and meter reading pole, include outdoor water gauge and indoor water gauge among them, contain the water gauge of environmental factor such as various weather, dust, mud.
Secondly, identifying and judging whether the water meter image is an upright image or an inverted image by using the collected water meter code in the water meter image as a standard through a classification algorithm, and obtaining model parameters for upright and inverted judgment of the water meter image;
thirdly, recognizing a mask image of a water meter code of the water meter image, calculating an external torque, and performing affine transformation rotation according to the angle of the external torque to obtain a water meter code image in which the water meter code is placed according to a horizontal position;
and fourthly, identifying the water meter code image to obtain the coordinates and numerical value information of each number in the water meter code.
The second step comprises the following specific steps:
s1, making upright and inverted classification data sets according to the water meter images collected in the step I. And defining a label of the upright or inverted picture according to the standard position of the water meter code, and finally distributing the picture according to the proportion of 80% of training set and 20% of verification set.
And S2, training the network by using a GoogleNet network as a basis, modifying the last classification layer of the classification data set into 2 types, reducing the number of part of low-layer networks, halving the number on the basis of the original number, and not changing the classification data set from the fourth layer, training a classification model by using a back propagation algorithm, and expanding the classification data set by using random augmentation operation in combination with a label in the training process.
And S3, performing convolution operation on the collected water meter images and the trained network, and classifying the collected water meter images into output results by softmax. And selecting a model with the optimal performance in the verification set, and obtaining the parameters of the upright and inverted judgment model of the water meter image.
And step two, manufacturing upright and inverted classification data sets according to the water meter images, wherein the upright and inverted judgment standards are as follows: the lower part of the water meter code is taken as a horizontal line, the angle of the horizontal line is inverted when the angle and the image horizontal line are between 90 and 180 degrees, and the other parts are upright.
The third step comprises the following specific steps:
and S1, marking the outlines of the water meter code, the pointer character wheel and the water meter dial in the water meter image in a mask mode, storing a file containing rectangular coordinates of the areas, storing the file in a form of length, width and height by using the initial coordinates, storing the coordinates of each outline point by using the outline coordinates, and storing the file by using the values of the coordinates in a json format. And selecting 90% of labeled data as a training set and 10% of data as a verification set.
S2, extracting features by using a resnet50 as a framework, training a network, training the network by using a Mask-RCNN framework, and identifying the water meter code, the pointer character wheel and the contour result of the water meter dial. The validation set leaves the model parameters that perform optimally.
S3, combining the standard condition of the water meter characteristics and the upright or inverted classification result of the water meter image obtained in the step two;
if the image is the upright water meter image, the image is input;
if the water meter image is the inverted water meter image, calculating the external moment of the water meter code, and according to the water meter code and the horizontal line angle of the image; and rotating the inverted water meter image to a standard posture of the water meter code through affine transformation rotation and outputting a picture.
The fourth step comprises the following specific steps:
and S1, acquiring upright water meter images, manually marking the digital position of each picture to form a training set and a verification set, wherein the training set and the verification set respectively account for 90% and 10% of the total number of the pictures, and training by using a deep residual error network (ResNet-18). It should be noted that, in addition to marking conventional 0-9 numbers, hyphens are marked, the marked content includes 0_1,1_2,2_3,3_4,4_5,5_6,6_7,7_8,8_9,9_0, etc., a data set is made, and 90% of data is selected as a training set and 10% of data is selected as a verification set.
And S2, selecting the parameters which are best represented on the verification set by the trained model for actual detection, and acquiring the coordinates and numerical values of the target character through the pictures output in the step three.
The fourth step further comprises the following specific steps:
s3, sorting all the character results according to the coordinates, calculating the interval between every two characters, considering the water meter code characteristics, the distances between the water meter codes are basically equal or proportional, and removing the obvious abnormal character interval according to the standard difference value by calculating the standard difference of the interval.
S4, according to the reasonable character coordinate array obtained from S3, if some coordinates are very close, the situation represents that a plurality of results are judged or hyphens appear, the two situations are processed, if no hyphen exists, a number with a high Faster-RCNN prediction probability is selected, if the hyphen appears, a larger number in the hyphen, such as 5_6, is directly selected, and the current number is 6.
The intelligent water meter code identification method based on the convolutional neural network technology of the embodiment is implemented specifically as follows:
step 1: shoot the water gauge of arbitrary angle through cell-phone and meter reading pole, wherein contain outdoor water gauge and indoor water gauge, contain the visible water gauge picture of the abundant water gauge sign indicating number of variety such as various weather, dust, mud. And defining a label of the upright or inverted picture according to the standard position of the water meter code, and finally distributing the picture according to the proportion of 80% of training set and 20% of verification set.
Step 2: the network training uses GoogleNet network as the basis, modifies the last classification layer into 2 types, reduces the number of part of low-level networks, reduces the number by half on the basis of the original number, does not change from the fourth layer, and trains the classification model by using a back propagation algorithm. And in the training process, random augmentation operation is used in combination with the label to expand the data set.
And step 3: and selecting a model with the optimal performance in the verification set, and obtaining the parameters of the upright and inverted judgment models of the pictures.
And 4, step 4: and marking the picture data in the step 1 by using a mask mode, wherein the picture data comprises a water meter code, a pointer character wheel and contour data of a dial plate, and an area rectangular coordinate storage file comprising the picture data, the pointer character wheel and the contour data is stored in a starting coordinate, length, width and height mode, the contour coordinate stores the coordinate of each contour point, and the values of the coordinates are stored in a file in a json format. And selecting 90% of data as a training set and 10% of data as a verification set.
And 5: and (3) extracting features by using a resnet50 as a framework, training a network by using a Mask-RCNN framework, and storing the model in a verification set to represent the optimal model.
Step 6: and (5) inputting all data set pictures by using the model parameters in the step 5, acquiring the outline coordinates of the water meter code, calculating the external moment, and rotating to a standard position by using an affine transformation method, as shown in figure 2.
And 7: the water meter code for marking the standard position comprises a rectangular coordinate of the water meter code and a label numerical value, wherein a training set and a verification set are marked manually on the numerical position of each picture, which respectively account for 90% and 10% of the total number of the pictures, it is noted that a hyphen is marked in addition to marking conventional 0-9 numbers, and the marking content comprises 0_1,1_2,2_3,3_4,4_5,5_6,6_7,7_8,8_9,9_0 and the like, as shown in fig. 3, a data set is made, 90% of data is selected as the training set, and 10% of data is selected as the verification set.
And 8: training is based on the fast-RCNN architecture using a deep residual network (ResNet-18). The trained model selects the parameters which have the best performance on the verification set for actual detection.
And step 9: and (3) storing the three model parameters trained in the steps (3), (5) and (8), inputting a water meter picture shot by a meter reading rod or a mobile phone at any angle, judging whether the water meter picture is upright or inverted by the model in the step (3), and acquiring a water meter code, a pointer character wheel, a dial contour coordinate and an external moment by the input picture through the model in the step (5).
Step 10: if the pointer character wheel exists, the upright and inverted conditions of the water meter are judged through the priori knowledge that the pointer character wheel is located below the water meter code, if the upright and inverted results are consistent with the model classification result in the step 3, the upright and inverted results can be confirmed, the first-level confidence coefficient is set to be 99%, if the upright and inverted results are inconsistent, the confidence coefficient of the pointer character wheel result is greater than the classification confidence coefficient of the model in the step 3, the result judged according to the position of the pointer character wheel is taken, the first-level confidence coefficient is set to be the confidence coefficient of the pointer character wheel, and if the inverted. And (4) according to the results of the upright and inverted positions, stacking the rotary water meter to a standard position.
Step 11: and (4) predicting the water meter code picture obtained in the step (10) by using the model obtained in the step (8) to obtain a target character coordinate and a numerical value.
Step 12: and (3) sequencing the results of the step (11) left and right according to coordinates, calculating the interval between every two characters, taking the characteristics of the digital codes of the water meter into consideration, wherein the distances between the digital codes are basically equal or proportional, and removing the obvious abnormal character interval according to the standard difference value by calculating the standard difference of the interval, wherein the common abnormal character interval can be misjudgment caused by the existence of interference.
Step 13: according to the reasonable character coordinate array obtained in the step 12, if some coordinates are very close, the situation represents that a plurality of results are judged or hyphens appear, the two situations are processed, if no hyphen exists, a number with higher Faster-RCNN prediction probability is selected, if the hyphen appears, a larger number in the hyphen is directly selected, for example, 5_6, and the current number is determined as 6.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention, and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. An intelligent water meter code identification method based on a convolutional neural network technology is characterized by comprising the following steps:
step one, carrying out image acquisition on one surface of the water meter with a water meter code;
secondly, identifying and judging whether the water meter image is an upright image or an inverted image by using the collected water meter code in the water meter image as a standard through a classification algorithm, and obtaining model parameters for upright and inverted judgment of the water meter image;
thirdly, recognizing a mask image of a water meter code of the water meter image, calculating an external torque, and performing affine transformation rotation according to the angle of the external torque to obtain a water meter code image in which the water meter code is placed according to a horizontal position;
and fourthly, identifying the water meter code image to obtain the coordinates and numerical value information of each number in the water meter code.
2. The intelligent water meter code identification method based on the convolutional neural network technology as claimed in claim 1, wherein: the second step comprises the following specific steps:
s1, according to the step one
And making classification data sets of upright and inverted water meter images.
And S2, training the network by using a GoogleNet network as a basis, modifying the last classification layer of the classification data set into 2 types, reducing the number of part of low-layer networks, halving the number on the basis of the original number, and not changing the classification data set from the fourth layer, training a classification model by using a back propagation algorithm, and expanding the classification data set by using random augmentation operation in combination with a label in the training process.
And S3, performing convolution operation on the collected water meter images and the trained network, and classifying the collected water meter images into output results by softmax.
3. The intelligent water meter code identification method based on the convolutional neural network technology as claimed in claim 2, wherein: and step two, manufacturing upright and inverted classification data sets according to the water meter images, wherein the upright and inverted judgment standards are as follows: the lower part of the water meter code is taken as a horizontal line, the angle of the horizontal line is inverted when the angle and the image horizontal line are between 90 and 180 degrees, and the other parts are upright.
4. The intelligent water meter code identification method based on the convolutional neural network technology as claimed in claim 2, wherein: the third step comprises the following specific steps:
and S1, marking the outlines of the water meter code, the pointer character wheel and the water meter dial in the water meter image in a mask mode, storing a rectangular coordinate storage file containing the outlines in a form of length, width and height by using the initial coordinates, and storing the coordinates of each outline point by using the outline coordinates.
S2, extracting features by using a resnet50 as a framework, training a network, and identifying the water meter code, the pointer character wheel and the outline result of the water meter dial by using a Mask-RCNN algorithm.
S3, combining the standard condition of the water meter characteristics and the upright or inverted classification result of the water meter image obtained in the step two;
if the image is the upright water meter image, the image is input;
if the water meter image is the inverted water meter image, calculating the external moment of the water meter code, and according to the water meter code and the horizontal line angle of the image; and rotating the inverted water meter image to a standard posture of the water meter code through affine transformation rotation and outputting a picture.
5. The intelligent water meter code identification method based on the convolutional neural network technology as claimed in claim 4, wherein: the fourth step comprises the following specific steps:
and S1, acquiring upright water meter images, manually marking the digital position of each picture to form a training set and a verification set, wherein the training set and the verification set respectively account for 90% and 10% of the total number of the pictures, and training by using a deep residual error network (ResNet-18).
And S2, selecting the parameters which are best represented on the verification set by the trained model for actual detection, and acquiring the coordinates and numerical values of the target character through the pictures output in the step three.
6. The intelligent water meter code identification method based on the convolutional neural network technology as claimed in claim 5, wherein: the fourth step further comprises the following specific steps:
s3, sorting all the character results according to the coordinates, calculating the interval between every two characters, considering the water meter code characteristics, the distances between the water meter codes are basically equal or proportional, and removing the obvious abnormal character interval according to the standard difference value by calculating the standard difference of the interval.
S4, according to the reasonable character coordinate array obtained from S3, if some coordinates are very close, the situation represents that a plurality of results are judged or hyphens appear, the two situations are processed, if no hyphen exists, a number with a high Faster-RCNN prediction probability is selected, if the hyphen appears, a larger number in the hyphen, such as 5_6, is directly selected, and the current number is 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010795175.1A CN112001383A (en) | 2020-08-10 | 2020-08-10 | Water meter code intelligent identification method based on convolutional neural network technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010795175.1A CN112001383A (en) | 2020-08-10 | 2020-08-10 | Water meter code intelligent identification method based on convolutional neural network technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112001383A true CN112001383A (en) | 2020-11-27 |
Family
ID=73462943
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010795175.1A Pending CN112001383A (en) | 2020-08-10 | 2020-08-10 | Water meter code intelligent identification method based on convolutional neural network technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112001383A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112200160A (en) * | 2020-12-02 | 2021-01-08 | 成都信息工程大学 | Deep learning-based direct-reading water meter reading identification method |
| CN112818983A (en) * | 2021-01-22 | 2021-05-18 | 常州友志自动化科技有限公司 | Method for judging character inversion by using picture acquaintance |
| CN113516110A (en) * | 2021-09-13 | 2021-10-19 | 成都千嘉科技有限公司 | Gas meter character wheel coordinate extraction method based on image segmentation |
| CN113743405A (en) * | 2021-09-07 | 2021-12-03 | 南方电网数字电网研究院有限公司 | Intelligent meter reading method and device for electric energy meter |
| WO2022127384A1 (en) * | 2020-12-15 | 2022-06-23 | 中兴通讯股份有限公司 | Character recognition method, electronic device and computer-readable storage medium |
| CN117274971A (en) * | 2023-11-20 | 2023-12-22 | 深圳拓安信物联股份有限公司 | Image processing method applied to water meter data extraction and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108416412A (en) * | 2018-01-23 | 2018-08-17 | 浙江瀚镪自动化设备股份有限公司 | A kind of logistics compound key recognition methods based on multitask deep learning |
| CN108875739A (en) * | 2018-06-13 | 2018-11-23 | 深圳市云识科技有限公司 | A kind of accurate detecting method of digital displaying meter reading |
| CN109299717A (en) * | 2018-09-13 | 2019-02-01 | 网易（杭州）网络有限公司 | Text region model foundation and character recognition method, device, medium and equipment |
| WO2019169532A1 (en) * | 2018-03-05 | 2019-09-12 | 深圳前海达闼云端智能科技有限公司 | License plate recognition method and cloud system |
-
2020
- 2020-08-10 CN CN202010795175.1A patent/CN112001383A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108416412A (en) * | 2018-01-23 | 2018-08-17 | 浙江瀚镪自动化设备股份有限公司 | A kind of logistics compound key recognition methods based on multitask deep learning |
| WO2019169532A1 (en) * | 2018-03-05 | 2019-09-12 | 深圳前海达闼云端智能科技有限公司 | License plate recognition method and cloud system |
| CN108875739A (en) * | 2018-06-13 | 2018-11-23 | 深圳市云识科技有限公司 | A kind of accurate detecting method of digital displaying meter reading |
| CN109299717A (en) * | 2018-09-13 | 2019-02-01 | 网易（杭州）网络有限公司 | Text region model foundation and character recognition method, device, medium and equipment |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112200160A (en) * | 2020-12-02 | 2021-01-08 | 成都信息工程大学 | Deep learning-based direct-reading water meter reading identification method |
| WO2022127384A1 (en) * | 2020-12-15 | 2022-06-23 | 中兴通讯股份有限公司 | Character recognition method, electronic device and computer-readable storage medium |
| CN112818983A (en) * | 2021-01-22 | 2021-05-18 | 常州友志自动化科技有限公司 | Method for judging character inversion by using picture acquaintance |
| CN113743405A (en) * | 2021-09-07 | 2021-12-03 | 南方电网数字电网研究院有限公司 | Intelligent meter reading method and device for electric energy meter |
| CN113516110A (en) * | 2021-09-13 | 2021-10-19 | 成都千嘉科技有限公司 | Gas meter character wheel coordinate extraction method based on image segmentation |
| CN113516110B (en) * | 2021-09-13 | 2021-12-21 | 成都千嘉科技有限公司 | Gas meter character wheel coordinate extraction method based on image segmentation |
| CN117274971A (en) * | 2023-11-20 | 2023-12-22 | 深圳拓安信物联股份有限公司 | Image processing method applied to water meter data extraction and electronic equipment |
| CN117274971B (en) * | 2023-11-20 | 2024-04-12 | 深圳拓安信物联股份有限公司 | Image processing method applied to water meter data extraction and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112001383A (en) | Water meter code intelligent identification method based on convolutional neural network technology | |
| CN108920580B (en) | Image matching method, device, storage medium and terminal | |
| CN107944450B (en) | License plate recognition method and device | |
| CN111461190A (en) | Deep convolutional neural network-based non-equilibrium ship classification method | |
| CN110991435A (en) | Express waybill key information positioning method and device based on deep learning | |
| CN111967470A (en) | Text recognition method and system based on decoupling attention mechanism | |
| CN103154972A (en) | Text-based 3D augmented reality | |
| CN111860233B (en) | SAR image complex building extraction method and system based on attention network selection | |
| CN113947766B (en) | Real-time license plate detection method based on convolutional neural network | |
| CN108664860B (en) | Method and device for identifying room plan | |
| CN103778436A (en) | Pedestrian gesture inspecting method based on image processing | |
| CN110659637A (en) | Electric energy meter number and label automatic identification method combining deep neural network and SIFT features | |
| CN110647956A (en) | Invoice information extraction method combined with two-dimensional code recognition | |
| CN109800756A (en) | A kind of text detection recognition methods for the intensive text of Chinese historical document | |
| EP3561729A1 (en) | Method for detecting and recognising long-range high-density visual markers | |
| CN111160107B (en) | Dynamic region detection method based on feature matching | |
| CN111783608A (en) | Face changing video detection method | |
| CN115265545A (en) | Map matching navigation method, device, equipment and storage medium based on decision analysis | |
| CN113095316B (en) | Image rotation target detection method based on multilevel fusion and angular point offset | |
| CN115880683B (en) | Urban waterlogging ponding intelligent water level detection method based on deep learning | |
| CN111652244A (en) | Pointer type meter identification method based on unsupervised feature extraction and matching | |
| CN108133213A (en) | A kind of embedded digital recognition methods imaged towards shell of gas meter formula | |
| CN112241736A (en) | Text detection method and device | |
| CN115690770A (en) | License plate recognition method based on space attention characteristics in non-limited scene | |
| CN113128461B (en) | Pedestrian re-recognition performance improving method based on human body key point mining full-scale features |
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 |