CN114743201A - Multimeter reading identification method and system based on rotating target detection - Google Patents
Multimeter reading identification method and system based on rotating target detection Download PDFInfo
- Publication number
- CN114743201A CN114743201A CN202210410911.6A CN202210410911A CN114743201A CN 114743201 A CN114743201 A CN 114743201A CN 202210410911 A CN202210410911 A CN 202210410911A CN 114743201 A CN114743201 A CN 114743201A
- Authority
- CN
- China
- Prior art keywords
- reading
- rotating
- angle
- result
- change
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 29
- 230000009466 transformation Effects 0.000 claims abstract description 25
- 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 23
- 238000004364 calculation method Methods 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 8
- 230000010354 integration Effects 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 10
- 102100032202 Cornulin Human genes 0.000 claims description 7
- 101000920981 Homo sapiens Cornulin Proteins 0.000 claims description 7
- 238000013518 transcription Methods 0.000 claims description 6
- 230000035897 transcription Effects 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 230000007306 turnover Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 15
- 238000010586 diagram Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000002457 bidirectional effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 125000004432 carbon atom Chemical group C* 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- 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
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/02—Affine transformations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a multimeter reading identification method and a system based on rotating target detection, wherein the method comprises the following steps: processing an image to be detected based on the improved YOLOv5 model, and outputting a reading area rotating frame, a reading area rotating angle and a change-over switch rotating angle; calculating the actual rotation angle of the change-over switch according to the rotation angle of the reading area and the rotation angle of the change-over switch; matching the actual rotation angle of the change-over switch with unit matching information to obtain a reading unit result; performing affine transformation and identification on the reading area rotating frame to obtain a reading digital result; and integrating the reading number result and the reading unit result to obtain a complete reading result. The system comprises: the device comprises a detection module, an actual rotation angle calculation module, a unit matching module, a digital identification module and a reading integration module. By using the invention, the complete reading result of the multimeter with the change-over switch at any rotation angle can be detected and identified.
Description
Technical Field
The invention relates to the field of intelligent manufacturing in China, in particular to a multimeter reading identification method and a multimeter reading identification system based on rotary target detection.
Background
The automation and intellectualization of reading identification of industrial instruments are in great trend. The reading identification technology of the instrument based on computer vision can automatically identify the numerical information of the collected instrument and quickly input the numerical information into a service system, effectively solves the problems of error reading, missing reading and the like in the manual reading process, improves the reading efficiency, reduces the manual input workload, reduces the labor cost of enterprises and realizes the automation of the data input of the instrument and the instrument.
The existing multimeter reading identification method mainly includes that a reading detection area is firstly cut, and then reading identification is carried out on a cut reading image, but most of the methods only pay attention to detection and identification of horizontal reading, and identification of reading units is often ignored, so that application value and application range are limited. The method based on the traditional image processing requires a single environment background and is greatly interfered by environmental changes such as illumination and the like, so that the universality is poor. In contrast, YOLOv5 is an One-stage deep learning target detection algorithm without an RPN structure, can well detect a reading area, has a small backbone structure such as tiny-dark net, can lighten the model through a series of optimization measures, is suitable for being deployed at a mobile end as an engineering algorithm, and has higher practical application value. However, YOLOv5, as a general target detection algorithm, is not suitable for detecting a rotating target with a large inclination angle, and even cannot determine the target orientation and angle.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a multimeter reading identification method and a multimeter reading identification system based on rotating target detection, which can detect and identify a complete reading result of a multimeter with a change-over switch at any rotating angle.
The first technical scheme adopted by the invention is as follows: a multimeter reading identification method based on rotating target detection comprises the following steps:
processing an image to be detected based on the improved YOLOv5 model, and outputting a reading area rotating frame, a reading area rotating angle and a change-over switch rotating angle;
calculating the actual rotation angle of the change-over switch according to the rotation angle of the reading area and the rotation angle of the change-over switch;
matching the actual rotation angle of the change-over switch with unit matching information to obtain a reading unit result;
performing affine transformation and identification on the reading area rotating frame to obtain a reading digital result;
and integrating the reading number result and the reading unit result to obtain a complete reading result.
Further, the training step of the improved YOLOv5 model specifically includes:
acquiring a training image, labeling a rotating frame, and coding coordinates of four corners of the rotating frame into geometric elements to obtain a labeled image;
performing data enhancement on the marked image to obtain a training set;
inputting the training set into a YOLOv5 model;
sequentially performing feature extraction, feature fusion and frame regression to obtain rotating frame information;
and (4) updating the model parameters by combining the real value label calculation loss to obtain an improved YOLOv5 model.
Further, the rotating frame information includes a center point coordinate, a width and a height, an angle, a confidence, bias information, each category probability, and each orientation category probability of the rotating frame.
Further, the output formula of the detection head of the improved YOLOv5 model is as follows:
in the above formula, nhIn order to detect the number of heads,bsfor the number of images to be measured of the input model, naNumber of anchor points of preset frame for grid cell, HiIs high at the output of the ith detection head, WiWidth output for the ith detection head, (x, y) center coordinates of the prediction box, (w, h) width and height of the prediction box, conf confidence of the prediction target, C probability of the prediction box belonging to each class, ncFor the predicted number of object classes, rjFor the regression parameters of the rotating frame, OkThe target orientation category probabilities.
Further, the step of matching the actual rotation angle of the change-over switch with the unit matching information to obtain the reading unit result specifically includes:
acquiring unit matching information to obtain a unit information data set and an angle information data set;
matching the actual rotation angle of the change-over switch with the angle information data set to obtain an angle index value;
and matching the angle index value with the unit information data set to obtain a reading unit result.
Further, the step of performing affine transformation and identification on the reading area rotation frame to obtain a reading digital result specifically includes:
performing affine transformation on the reading area rotating frame to obtain a horizontal reading area graph;
the affine transformation comprises the combined transformation of translation, scaling, rotation, turnover and miscut of a reading area rotation frame;
and sequentially carrying out CNN feature extraction, RNN sequence modeling and CTC transcription on the horizontal reading area graph based on the CRNN reading identification model to obtain a reading number result.
Further, the matrix formula of the affine transformation is as follows:
in the above formula, (x, y) is coordinates of the rotation frame point of the reading area, and (u, v) is obtained by affine transformation of the rotation frame point of the reading areaA rotating frame point of0、a1、a2、b0、b1And b2M is an affine transformation matrix.
The second technical scheme adopted by the invention is as follows: a multimeter reading identification system based on rotating target detection, comprising:
the detection module is used for processing the image to be detected based on the improved YOLOv5 model and outputting a reading area rotating frame, a reading area rotating angle and a change-over switch rotating angle;
the actual rotating angle calculating module is used for calculating the actual rotating angle of the change-over switch according to the rotating angle of the reading area and the rotating angle of the change-over switch;
the unit matching module is used for matching the actual rotation angle of the change-over switch with the unit matching information to obtain a reading unit result;
the digital identification module is used for carrying out affine transformation and identification on the reading area rotating frame to obtain a reading digital result;
and the reading integration module is used for integrating the reading number result and the reading unit result to obtain a complete reading result.
The method and the system have the beneficial effects that: the method solves the reading identification difficulty of the multimeter at any rotation angle by improving the YOLOv5 algorithm, and provides an identification method of reading units for a series of multimeters with change-over switches, so that the reading information is more complete, and the method is suitable for a class of digital multimeters with change-over switches, and the application range and the application value are expanded.
Drawings
FIG. 1 is a flow chart of the steps of a multimeter reading identification method of the present invention based on rotating target detection;
FIG. 2 is a block diagram of a multimeter reading identification system based on rotating target detection in accordance with the present invention;
FIG. 3 is a schematic diagram of a modified YOLOv5 rotation region detection model according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the improved YOLOv5 output spin box of an embodiment of the present invention;
FIG. 5 is a schematic diagram of a transfer switch according to an embodiment of the present invention;
FIG. 6 is a schematic view of a reading zone correction according to an embodiment of the present invention;
fig. 7 is a block diagram of a CRNN reading recognition model according to an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and the specific embodiments. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
Referring to fig. 1, the invention provides a multimeter reading identification method based on rotating target detection, which comprises the following steps:
s1, referring to FIG. 3, training a Yolov5 model to obtain an improved Yolov5 model;
s1.1, acquiring a training image, labeling a rotating frame, and coding coordinates of four corners of the rotating frame into geometric elements to obtain a labeled image;
specifically, the coordinates of the four corners of the rotating frame are encoded as geometric elements in the manner described with reference to fig. 4, and the specific operations are as follows:
||Rj||＝w×Sigmoid(rj)，j∈{0,1}；
in the above formula, RjThe distance between the position of the j point of the predicted rotating frame and the position of the j point of the predicted horizontal frame is taken as the distance; w is the width of the horizontal box predicted by the grid cell, Sigmoid (r)j) Is the normalized value of the regression parameter of the bounding box, rjThe regression parameters are offset for the bounding box.
S1.2, performing data enhancement on the marked image to obtain a training set;
s1.3, inputting a training set into a YOLOv5 model;
s1.4, sequentially performing feature extraction, feature fusion and frame regression to obtain rotating frame information;
specifically, the rotation frame information includes center point coordinates, width and height, angle, confidence, bias information, each category probability, and each orientation category probability of the rotation frame.
The calculation operation of the angle of the rotating frame is as follows:
when r isjNot equal to 0, i.e. the rotation angle is calculated in a clockwise direction starting from the vertically upward direction in the presence of a rotating frame
In the above formula, the first and second carbon atoms are,
the coordinates of the four vertices of the annotated image,
is the angle of the rotating frame.
When r isjIf 0, i.e. only horizontal boxes are present, the orientation can be predicted by
Calculating the rotation angle of the target
Wherein, the first and the second end of the pipe are connected with each other,
for the direction with the maximum probability in the direction class probabilities, the specific calculation operation is as follows:
in the above formula, P (O)k) For each orientation category probability, the specific calculation operation is as follows:
P(Ok)＝Sigmoid(Ok),k∈{0,1,2,3}。
and S1.5, updating the model parameters by combining the loss calculated by the true value label to obtain an improved YOLOv5 model.
Specifically, the model's penalty function calculation operates as follows:
L＝λboxLbox+λobjLobj+λclsLcls+λrboxLrbox+λrotLrot；
in the above formula, λbox、λobj、λcls、λrboxAnd λrotAre respectively Lbox、Lobj、Lcls、LrboxAnd LrotThe weight coefficient of (a); l isboxUsing a CIOU Loss function as a Loss function of the horizontal frame bbox; l isobjUsing a Focal local Loss function as a target confidence coefficient Loss function; l is a radical of an alcoholclsFor predicting the category Loss function, using a Focal local Loss function; l is a radical of an alcoholrboxTo rotate the frame offset regression loss function, Smooth was usedL1A loss function; l isrotFor each orientation classification Loss function, the Focal local Loss function is used.
Wherein the rotating frame is offset from the regression loss function LrboxIs represented as follows:
in the above formula, rj' is rjThe normalized value of (a) is calculated,
is rjTrue value of rjSmooth for rotating frame offset regression parametersL1To calculate rj' and
the loss function of the regression error loss between the following operations is specifically performed:
loss function for each orientation classification LrotIs represented as follows:
in the above formula, αkTo balance the weight coefficients of the positive and negative samples, γ is a sample used to balance simple difficulty.
S2, processing the image to be measured based on the improved YOLOv5 model, and outputting a reading area rotating frame, a reading area rotating angle and a change-over switch rotating angle;
specifically, the improved YOLOv5 model is composed of an input end, a Backbone network for Backbone feature extraction, a Neck feature enhancement network and a Head detection Head.
The output of the improved YOLOv5 model detection head is specifically calculated as follows:
in the above formula, nhTo detect the number of heads, bsFor the size of the number of images to be measured of the input model, naNumber of anchor points of preset frame for grid cell, HiIs high at the output of the ith detection head, WiWidth output for the ith detection head, (x, y) center coordinates of the prediction box, (w, h) width and height of the prediction box, conf confidence of the prediction target, C probability of the prediction box belonging to each class, ncFor the predicted number of object classes, rjFor the regression parameters of the rotating frame, OkThe target orientation category probabilities.
S3, calculating the actual rotation angle of the change-over switch according to the rotation angle of the reading area and the rotation angle of the change-over switch;
specifically, the specific calculation operation of the actual rotation angle of the change-over switch is:
in the above formula, θdTo read the angular rotation of the field, θgTo change over the switch rotation angle, θrTo change the actual angle of rotation of the switch.
Wherein, thetadAnd thetagIs that
S4, referring to FIG. 5, matching the actual rotation angle of the change-over switch with unit matching information to obtain a reading unit result;
s4.1, acquiring unit matching information to obtain a unit information data set and an angle information data set;
specifically, unit information is stored in an array U, U ═ Ull∈{1,2,...,nu}}；
In the above formula, UlDenotes the l unit, nuThe unit type number of the change-over switch is expressed;
each UlThe corresponding angle information is stored in an array theta, theta being { theta ═ thetall∈{1,2,...,nu}}：
In the above formula, θlIndicating unit U in change-over switchlThe corresponding rotation angle.
S4.2, matching the actual rotation angle of the change-over switch with the angle information data set to obtain an angle index value;
specifically, the matching operation formula of the actual rotation angle of the change-over switch and the angle information data set is as follows:
m＝argminl|θr-θl|；
in the above formula, m is theta and thetarThe index value of the closest angle.
And S4.3, matching the angle index value with the unit information data set to obtain a reading unit result.
Specifically, the matching operation formula of the angle index value and the unit information data set is as follows:
s5, referring to FIG. 6, performing affine transformation and identification on the reading area rotating frame to obtain a reading digital result;
s5.1, performing affine transformation on the reading area rotating frame to obtain a horizontal reading area graph;
specifically, the original rotation frame point coordinates (x, y) are affine transformed to the coordinates (u, v), and the specific operation is as follows:
the matrix representation is as follows:
in the above formula, M is an affine transformation matrix, and can perform translation, scaling, rotation, turnover and miscut combined transformation on the reading area rotating frame, a0、a1、a2、b0、b1And b2Is the M parameter.
S5.2, referring to FIG. 7, sequentially performing CNN feature extraction, RNN sequence modeling and CTC transcription on the horizontal reading area graph based on the CRNN reading identification model to obtain a reading number result;
s5.2.1, inputting the horizontal reading area graph into a CRNN reading identification model;
specifically, the CRNN reading recognition model is composed of a CNN feature extraction module, an RNN sequence modeling module and a CTC transcription module, and is a stable end-to-end text recognition framework.
S5.2.2, performing feature extraction on the horizontal reading area map based on a CNN feature extraction module to obtain a convolution feature map;
specifically, the horizontal reading area image is input into a MobileNetv3 lightweight convolution neural network to extract overall characteristic information, and a convolution characteristic map is output.
S5.2.3, performing sequence modeling on the convolution characteristic diagram based on an RNN sequence modeling module to obtain a posterior probability matrix;
specifically, in order to enhance the extraction capability of the sequence features before and after the RNN, the method uses a deep bidirectional LSTM network to extract the sequence features before and after the reading of the multimeter; the eigen-map size of the input Bi-LSTM is (1, T, D), its height is 1, the maximum time length is T, each input vector dimension is D, and the Bi-LSTM output y is a posterior probability matrix.
S5.2.4, transcribing the posterior probability matrix based on the CTC transcription module to obtain a reading number result.
Specifically, the CTC transcription module accesses each time slice of the posterior probability matrix y to softmax to obtain a prediction sequence, removes repeated tags in the sequence, and removes invalid space tags in the sequence, thereby translating the prediction sequence containing spaces into text information; suppose that the predicted sequence output by the bidirectional LSTM network is "- - -11-2-22-3" - "and becomes the final prediction result" 1223 "after being transcribed by CTC.
And S6, integrating the reading number result and the reading unit result to obtain a complete reading result.
Referring to FIG. 2, a multimeter reading identification system based on rotating target detection comprises:
the detection module is used for processing the image to be detected based on the improved YOLOv5 model and outputting a reading area rotating frame, a reading area rotating angle and a change-over switch rotating angle;
the actual rotating angle calculation module is used for calculating the actual rotating angle of the change-over switch according to the reading area rotating angle and the change-over switch rotating angle;
the unit matching module is used for matching the actual rotating angle of the change-over switch with the unit matching information to obtain a reading unit result;
the digital identification module is used for carrying out affine transformation and identification on the reading area rotating frame to obtain a reading digital result;
and the reading integration module is used for integrating the reading number result and the reading unit result to obtain a complete reading result.
The contents in the above method embodiments are all applicable to the present system embodiment, the functions specifically implemented by the present system embodiment are the same as those in the above method embodiment, and the beneficial effects achieved by the present system embodiment are also the same as those achieved by the above method embodiment.
The beneficial effects of the invention specifically comprise:
1) the reading unit result under any rotation angle of the multimeter can be detected and identified: expanding based on YOLOv5, adding target rotating frame offset regression parameters and output of target orientation probability parameters to a detection head of the detection head, and completing detection of a target rotating frame and an angle by using only one model; based on the multimeter reading area and the pose relation characteristic of the change-over switch, the actual angle of the change-over switch can be calculated by combining the reading area angle and the change-over switch angle to obtain a reading unit.
2) The multimeter reading digital result identification precision is improved, and the reading identification steps are simplified: and performing affine transformation on the detected reading area rotating frame into a horizontal frame so as to perform reading identification. The conventional reading recognition method comprises two steps of reading character segmentation and recognition, and the CRNN algorithm does not need to perform reading character segmentation and can finish reading recognition end to end.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (8)
1. A multimeter reading identification method based on rotating target detection is characterized by comprising the following steps:
processing an image to be detected based on the improved YOLOv5 model, and outputting a reading area rotating frame, a reading area rotating angle and a change-over switch rotating angle;
calculating the actual rotation angle of the change-over switch according to the rotation angle of the reading area and the rotation angle of the change-over switch;
matching the actual rotation angle of the change-over switch with unit matching information to obtain a reading unit result;
performing affine transformation and identification on the reading area rotating frame to obtain a reading digital result;
and integrating the reading number result and the reading unit result to obtain a complete reading result.
2. The method as claimed in claim 1, wherein the training step of the improved YOLOv5 model specifically comprises:
acquiring a training image, labeling a rotating frame, and coding coordinates of four corners of the rotating frame into geometric elements to obtain a labeled image;
performing data enhancement on the marked image to obtain a training set;
inputting the training set into a YOLOv5 model;
sequentially performing feature extraction, feature fusion and frame regression to obtain rotating frame information;
and (4) updating the model parameters by combining the real value label calculation loss to obtain an improved YOLOv5 model.
3. The method of claim 2, wherein the rotating frame information comprises center point coordinates, width and height, angle, confidence, bias information, class probabilities and orientation class probabilities of the rotating frame.
4. A multimeter reading identification method based on rotating object detection as recited in claim 1, wherein the output formula of the improved YOLOv5 model is expressed as follows:
in the above formula, nhTo detect the number of heads, bsFor the number of images to be measured of the input model, naNumber of anchor points of preset frame for grid cell, HiIs high at the output of the ith detection head, WiWidth output for the ith detection head, (x, y) center coordinates of the prediction box, (w, h) width and height of the prediction box, conf confidence of the prediction target, C probability of the prediction box belonging to each class, ncFor the predicted number of object classes, rjFor the regression parameters of the rotating frame, OkThe target orientation category probabilities.
5. The method for identifying the reading of the multimeter based on the rotating target detection as claimed in claim 1, wherein the step of matching the actual rotation angle of the change-over switch with unit matching information to obtain the reading unit result specifically comprises:
acquiring unit matching information to obtain a unit information data set and an angle information data set;
matching the actual rotation angle of the change-over switch with the angle information data set to obtain an angle index value;
and matching the angle index value with the unit information data set to obtain a reading unit result.
6. The method for identifying the reading of the multimeter based on the rotating target detection as claimed in claim 1, wherein the step of performing affine transformation and identification on the reading area rotating frame to obtain a reading digital result specifically comprises:
performing affine transformation on the reading area rotating frame to obtain a horizontal reading area graph;
the affine transformation comprises the combined transformation of translation, scaling, rotation, turnover and miscut of a reading area rotation frame;
and sequentially carrying out CNN feature extraction, RNN sequence modeling and CTC transcription on the horizontal reading area graph based on the CRNN reading identification model to obtain a reading number result.
7. A multimeter reading identification method based on rotating target detection as recited in claim 6, wherein the affine transformation matrix formula is as follows:
in the above formula, (x, y) is coordinates of a reading area rotation frame point, (u, v) is a rotation frame point obtained by affine transformation of the reading area rotation frame point, and a0、a1、a2、b0、b1And b2M is an affine transformation matrix.
8. A multimeter reading identification system based on rotating target detection, comprising:
the detection module is used for processing the image to be detected based on the improved YOLOv5 model and outputting a reading area rotating frame, a reading area rotating angle and a change-over switch rotating angle;
the actual rotating angle calculation module is used for calculating the actual rotating angle of the change-over switch according to the reading area rotating angle and the change-over switch rotating angle;
the unit matching module is used for matching the actual rotating angle of the change-over switch with the unit matching information to obtain a reading unit result;
the digital identification module is used for carrying out affine transformation and identification on the reading area rotating frame to obtain a reading digital result;
and the reading integration module is used for integrating the reading number result and the reading unit result to obtain a complete reading result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210410911.6A CN114743201A (en) | 2022-04-19 | 2022-04-19 | Multimeter reading identification method and system based on rotating target detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210410911.6A CN114743201A (en) | 2022-04-19 | 2022-04-19 | Multimeter reading identification method and system based on rotating target detection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114743201A true CN114743201A (en) | 2022-07-12 |
Family
ID=82280642
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210410911.6A Pending CN114743201A (en) | 2022-04-19 | 2022-04-19 | Multimeter reading identification method and system based on rotating target detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114743201A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115082922A (en) * | 2022-08-24 | 2022-09-20 | 济南瑞泉电子有限公司 | Water meter digital picture processing method and system based on deep learning |
| CN116740704A (en) * | 2023-06-16 | 2023-09-12 | 安徽农业大学 | Wheat leaf phenotype parameter change rate monitoring method and device based on deep learning |
-
2022
- 2022-04-19 CN CN202210410911.6A patent/CN114743201A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115082922A (en) * | 2022-08-24 | 2022-09-20 | 济南瑞泉电子有限公司 | Water meter digital picture processing method and system based on deep learning |
| CN116740704A (en) * | 2023-06-16 | 2023-09-12 | 安徽农业大学 | Wheat leaf phenotype parameter change rate monitoring method and device based on deep learning |
| CN116740704B (en) * | 2023-06-16 | 2024-02-27 | 安徽农业大学 | Wheat leaf phenotype parameter change rate monitoring method and device based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108427924B (en) | Text regression detection method based on rotation sensitive characteristics | |
| CN111462120B (en) | Defect detection method, device, medium and equipment based on semantic segmentation model | |
| CN114743201A (en) | Multimeter reading identification method and system based on rotating target detection | |
| CN114694165B (en) | PID drawing intelligent recognition and redrawing method | |
| CN112347284B (en) | Combined trademark image retrieval method | |
| Shen et al. | Vehicle detection in aerial images based on lightweight deep convolutional network and generative adversarial network | |
| CN112949338A (en) | Two-dimensional bar code accurate positioning method combining deep learning and Hough transformation | |
| CN115439458A (en) | Industrial image defect target detection algorithm based on depth map attention | |
| CN112836650A (en) | Semantic analysis method and system for quality inspection report scanning image table | |
| CN112016638A (en) | Method, device and equipment for identifying steel bar cluster and storage medium | |
| CN111476210A (en) | Image-based text recognition method, system, device and storage medium | |
| CN113159215A (en) | Small target detection and identification method based on fast Rcnn | |
| CN113762269A (en) | Chinese character OCR recognition method, system, medium and application based on neural network | |
| CN114332007A (en) | Transformer-based industrial defect detection and identification method | |
| CN115019103A (en) | Small sample target detection method based on coordinate attention group optimization | |
| CN112418207B (en) | Weak supervision character detection method based on self-attention distillation | |
| CN111144469B (en) | End-to-end multi-sequence text recognition method based on multi-dimensional associated time sequence classification neural network | |
| CN114782827B (en) | Object capture point acquisition method and device based on image | |
| CN110889418A (en) | Gas contour identification method | |
| Li et al. | Few-shot meta-learning on point cloud for semantic segmentation | |
| CN112199984B (en) | Target rapid detection method for large-scale remote sensing image | |
| CN115187896A (en) | Narrow-mouth bottle inclination angle detection method for intelligent experiment evaluation | |
| CN115345932A (en) | Laser SLAM loop detection method based on semantic information | |
| CN114998216A (en) | Method and device for rapidly detecting surface defects of transparent part | |
| CN114332715A (en) | Method, device and equipment for identifying snow through automatic meteorological observation and storage medium |
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 |