CN110472636B - Deep learning-based water gauge E-shaped scale identification method - Google Patents
Deep learning-based water gauge E-shaped scale identification method Download PDFInfo
- Publication number
- CN110472636B CN110472636B CN201910681936.8A CN201910681936A CN110472636B CN 110472636 B CN110472636 B CN 110472636B CN 201910681936 A CN201910681936 A CN 201910681936A CN 110472636 B CN110472636 B CN 110472636B
- Authority
- CN
- China
- Prior art keywords
- water gauge
- water
- data set
- shaped
- scale
- 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.)
- Active
Links
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
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/02—Recognising information on displays, dials, clocks
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)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Character Discrimination (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a water gauge E-shaped scale identification method based on deep learning, which comprises the following steps of: step S1: collecting a water gauge picture, and performing labeling processing to obtain a water gauge data set; step S2: extracting a water gauge scale E-shaped data set; s3, extracting a water line data set; s4, constructing a Faster RCNN with ResNe101 as a feature extractor, and pre-training; s5, respectively training the pre-trained Faster RCNN according to the water gauge data set and the water gauge scale E-shaped data set; s6, constructing a water line neural network model and training by using a water line data line; s7, sequentially inputting the pictures of the water gauge to be measured into a first Faster RCNN and a second Faster RCNN to obtain prediction frame data of each complete scale E shape on the water gauge; and S8, inputting the obtained prediction frame data of each complete scale E shape on the water gauge into the trained water level line neural network model, and calculating to obtain the water level height.
Description
Technical Field
The invention relates to a water gauge E-shaped scale recognition method based on deep learning.
Background
The water level is the most intuitive factor for reflecting the water condition of the water body, and the change of the water level is mainly caused by the increase and decrease of the water quantity of the water body. The water level refers to the elevation of the free water surface relative to a certain base plane, and the distance between the water surface and the river bottom is called the water depth. When the water level is higher than a certain height, disasters such as waterlogging, diking and the like can occur. That is to say, this height is the safety line of the water surface height of the water bodies such as rivers, lakes, oceans, reservoirs, etc., and is often called as a water level warning line, and the water level warning lines of different rivers, lakes, oceans, reservoirs are different in height. The water level change can be accurately and systematically sensed, so that the benefits and the hazards can be avoided in the process of responding to the water level change, and effective measures are taken to regulate and control the water level.
The water gauge is an important tool for measuring water level change, and the real-time monitoring of the water gauge through the camera is one of the main methods for monitoring the water level. The existing water gauge scale identification method is to compare the traditional image processing means with the manually designed image characteristics so as to identify the water gauge scale. The traditional water gauge scale identification method has certain defects in application: because of the artificially designed image characteristics, it is unable to effectively adapt to the various changes of the natural environment, such as: the water gauge has too strong reflection, dirt on the water gauge and the like. Aiming at the outstanding problems existing in the aspect, the accuracy of water gauge scale identification is improved, the change of the water level is accurately sensed, and the water gauge scale identification method is an inevitable requirement for water conservancy management work.
Disclosure of Invention
In view of this, the present invention aims to provide a method for identifying E-shaped scales of a water gauge based on deep learning, so as to well overcome the disadvantages of the conventional ponding detection manner, and under a variable natural environment (illumination, shielding, distortion, etc.), the method still has strong robustness for identifying the scales of the water gauge.
In order to achieve the purpose, the invention adopts the following technical scheme:
a water gauge E-shaped scale recognition method based on deep learning comprises the following steps:
step S1: collecting a water gauge picture, and performing labeling processing to obtain a water gauge data set;
step S2: extracting a water gauge scale E-shaped data set according to the obtained water gauge data set;
s3, extracting a water level line data set according to the obtained water gauge data set;
s4, constructing a Faster RCNN with ResNe101 as a characteristic extractor, and pre-training to obtain the pre-trained Faster RCNN;
step S5, respectively training the pre-trained Faster RCNN according to the water gauge data set and the water gauge scale E-shaped data set to obtain a first fast RCNN and a second fast RCNN after training;
s6, constructing a water line neural network model, and training by using a water line data set to obtain a trained water line neural network model;
s7, inputting the pictures of the water gauge to be tested into the first Faster RCNN and the second Faster RCNN in sequence to obtain
Obtaining the prediction frame data of each complete scale E shape on the water gauge;
and S8, calculating to obtain the water level height according to the obtained prediction frame data of each complete scale E shape on the water gauge and the trained water level line neural network model.
Furthermore, the marked content of the water gauge picture is an enclosure box enclosing the whole water gauge.
Further, the step S2 specifically includes: based on the water gauge data set, the bounding box marked by the water gauge data set is utilized to extract the picture only containing the water gauge as a water gauge scale E-shaped data set.
Further, the step S3 specifically includes: and taking the water gauge data set as a basis, utilizing the bounding box marked by the water gauge data set, taking the midpoint of the lower boundary of the marked bounding box as the image midpoint, and extracting a water line image with the width and the height of 200 pixels from each image to be used as the water line data set.
Further, the pre-training employs a COCO dataset to pre-train the fast RCNN.
Further, the step S7 specifically includes:
step S71, inputting the picture of the water gauge to be measured into a first Faster RCNN to obtain a prediction frame B of the water gauge sg ，[x sg ,y sg ,w sg ,h sg ]Wherein x is sg 、y sg Are respectively a prediction box B sg Horizontal and vertical coordinates of the upper left corner on the image, w sg 、h sg Predicting the width and height of the frame for the water gauge;
step S72, the obtained water gauge prediction frame B sg Corresponding picture g 1 Is extracted separatelyInputting a second Faster RCNN to obtain a prediction frame of each complete scale E shape on the water gauge; detecting k 'E' -shaped prediction frames, wherein the lowest 'E' -shaped prediction frame is B in the vertical direction be ， [x be ,y be ,w be ,h be ]Wherein x is be 、y be Are respectively a prediction box B be Abscissa and ordinate of the upper left corner on the image, w be 、h be The width and height of the prediction box.
Further, the water line neural network model is based on ResNet34, and the feature extraction part of the water line neural network model is composed of 5 convolution processes.
Further, the step S8 specifically includes:
step S81, on the image, in
Taking an image g with the width and height of 200 pixels as the center point of the frame 2 Inputting the data into a water level network, extracting features through a ResNet34 feature extractor, inputting the features into a fully-connected neural network, and fitting a slope a g2 And an offset b with respect to the lower left corner of the image g2 Thereby obtaining an output water level line;
step S82, according to the water level line and the prediction frame B of the water gauge sg Obtaining the water line in the graph g 2 Upper midpoint p g2 ，
Step S83 of calculating p g2 And the prediction box B be Midpoint of lower boundary
Is a distance of
Obtaining 'E' -shaped scale of the submerged part with the height of the exposed water surface of
A unit of scale;
step S84, based on the water level height H represented by the highest point of the preset water gauge db A realistic distance d corresponding to an E-shaped scale unit unit To obtain an actual water level of
Compared with the prior art, the invention has the following beneficial effects:
the method can well overcome the defects of the traditional accumulated water detection mode, and still has stronger robustness for identifying the scales of the water gauge under variable natural environments (illumination, shielding, distortion and the like).
Drawings
FIG. 1 is a flow diagram of the present invention;
FIG. 2 is a water gauge data image in an embodiment of the present invention;
FIG. 3 is a water gauge prediction box for water gauge image extraction in an embodiment of the present invention;
FIG. 4 is an extracted E-shape prediction box in an embodiment of the present invention;
FIG. 5 is a water line image obtained in an embodiment of the present invention;
fig. 6 is a block diagram of a depth residual method according to an embodiment of the present invention.
Detailed Description
The invention is further explained below with reference to the drawings and the embodiments.
Referring to fig. 1, the invention provides a method for identifying E-shaped scales of a water gauge based on deep learning, which comprises the following steps:
step S1: collecting a water gauge picture, and performing labeling processing to obtain a water gauge data set;
step S2: extracting a water gauge scale E-shaped data set according to the obtained water gauge data set;
s3, extracting a water line data set according to the obtained water gauge data set;
s4, constructing a Faster RCNN with ResNe101 as a feature extractor, and pre-training the Faster RCNN by adopting a COCO data set to obtain the pre-trained fast RCNN;
step S5, respectively training the pre-trained Faster RCNN according to the water gauge data set and the water gauge scale E-shaped data set to obtain a first fast RCNN and a second fast RCNN after training;
s6, constructing a water line neural network model, and training by using a water line data line to obtain a trained water line neural network model;
s7, inputting the pictures of the water gauge to be tested into the first Faster RCNN and the second Faster RCNN in sequence to obtain
Obtaining the prediction frame data of each complete scale E shape on the water gauge;
and S8, inputting the obtained prediction frame data of each complete scale E shape on the water gauge into the trained water level line neural network model, and calculating to obtain the water level height.
In this embodiment, the content marked by the water gauge picture is an enclosure surrounding the whole water gauge. The pictures in the data set take into account the following elements in the collection process: (1) interference of light spots; (2) interference of different illumination intensities; (3) interference of water splash; (4) interference of the background; (5) interference of water stains (6) interference of reflection (7) interference of different weather conditions and the like.
In the embodiment, based on the water gauge data set, the bounding box marked by the water gauge data set is used for extracting the picture only containing the water gauge as the water gauge scale E-shaped data set. And labeling the pictures in the data set. The marked content is: a bounding box enclosing each 'E' shaped water gauge scale. The bounding box has and has only one complete 'E' shaped water gauge scale.
In this embodiment, the step S3 specifically includes: and taking the water gauge data set as a basis, utilizing the bounding box marked by the water gauge data set, taking the midpoint of the lower boundary of the marked bounding box as the midpoint of the image, and extracting a water line image with the width and the height of 200 pixels from each image to be used as the water line data set. And marking the pictures in the data set. The content of the annotation is: a line (slope, offset) similar to the water line.
In this embodiment, the step S7 specifically includes:
step S71, inputting the picture of the water gauge to be measured into a first Faster RCNN to obtain a prediction frame B of the water gauge sg ，[x sg ,y sg ,w sg ,h sg ]Wherein x is sg 、y sg Are respectively a prediction box B sg Horizontal and vertical coordinates of the upper left corner on the image, w sg 、h sg Predicting the width and height of the frame for the water gauge;
step S72, the obtained water gauge prediction frame B sg Corresponding picture g 1 Independently extracting the E-shaped prediction frame, and inputting the E-shaped prediction frame into a second Faster RCNN to obtain the E-shaped prediction frame of each complete scale on the water gauge; detecting k 'E' -shaped prediction frames, wherein the lowest 'E' -shaped prediction frame is B in the vertical direction be ， [x be ,y be ,w be ,h be ]Wherein x is be 、y be Are respectively a prediction box B be Abscissa and ordinate of the upper left corner on the image, w be 、h be The width and height of the prediction box.
In this embodiment, the water line neural network model is based on the ResNet34, and the feature extraction part thereof is composed of 5 convolution processes. The water line neural network model is as follows:
TABLE 1
The construction of the blocks in table 1 is shown in fig. 6, and the number following the block is the number of stacks of blocks. The downsampling is performed by Conv3_1, conv4_1 and Conv5_1 with a step size of 2.
As shown in table 1, the water line neural network model is based on ResNet34, and the feature extraction part of the network consists of 5 convolution processes.
Wherein the step size of all convolution operations, pooling operations is 2. In Conv _1, the convolution operation is performed with a convolution kernel of size 7 × 7 and channel number 64. Conv2_ x, first 3 × 3 max pooling is followed by 3 stacks
The residual method block performs a convolution operation. Residual method block details are shown in fig. 6: and (3) sequentially performing convolution operation with the size of 3 multiplied by 3 and the number of channels of 64 and relu nonlinear activation operation on the input I, and performing convolution operation with the size of 3 multiplied by 3 and the number of channels of 64 to obtain M. And correspondingly adding the fingers of the I and the M according to the channels and then connecting with relu activation to obtain the output feature mapping. The implementations of Conv3_ x, conv4_ x and Conv5_ x are the same as described above, and table 1 may refer to this only if the parameters of the residual method blocks are different from the number of residual method blocks. In addition, after Conv3_1, conv4_1 and Conv5_1, a downsampling operation is performed with a step size of 2.Conv3_1 represents the first residual method block of Conv3_ x. After the Conv1-Conv5_ x feature extraction step, the high-level feature mapping of the image is obtained.
After performing the convolution operation, the resulting feature maps are input to 2 branches, respectively. The 2 branches are consistent in structure and design: firstly, the input feature mapping is subjected to average pooling with the step length of 2, and then a 1-dimensional full-connection layer is connected to obtain an output value. The loss function is the MSE (mean square error) of the network output and the index value. Training is carried out, and the 2 branches are finally fitted with the output offset and the slope respectively.
In this embodiment, the step S8 specifically includes:
step S81, on the image, in
Taking an image g with the width and height of 200 pixels as the center point of the frame 2 Inputting the data into a water level network, extracting features through a ResNet34 feature extractor, inputting the features into a fully-connected neural network, and fitting a slope a g2 And an offset b with respect to the lower left corner of the image g2 Thereby obtaining an output water level line;
step S82, according to the water level line and the prediction frame B of the water gauge sg Obtaining the water line in the graph g 2 Upper midpoint p g2 ，
Step S73 of calculating p g2 And prediction block B be Midpoint of lower boundary
Of (2) is
Obtaining 'E' -shaped scale of the submerged part with the height of the exposed water surface of
A unit of scale;
step S74, based on the water level height H represented by the highest point of the preset water gauge db A realistic distance d corresponding to the scale unit of the E shape unit Obtaining the actual water level as
The above description is only a preferred embodiment of the present invention, and all the equivalent changes and modifications made according to the claims of the present invention should be covered by the present invention.
Claims (8)
1. A water gauge E-shaped scale recognition method based on deep learning is characterized by comprising the following steps:
step S1: collecting a water gauge picture, and performing labeling processing to obtain a water gauge data set;
step S2: extracting a water gauge scale E-shaped data set according to the obtained water gauge data set;
s3, extracting a water line data set according to the obtained water gauge data set;
s4, constructing a Faster RCNN with ResNe101 as a characteristic extractor, and pre-training to obtain the pre-trained Faster RCNN;
step S5, respectively training the pre-trained Faster RCNN according to the water gauge data set and the water gauge scale E-shaped data set to obtain a first fast RCNN and a second fast RCNN after training;
s6, constructing a water line neural network model, and training by using a water line data set to obtain a trained water line neural network model;
s7, sequentially inputting the pictures of the water gauge to be measured into a first Faster RCNN and a second Faster RCNN to obtain prediction frame data of each complete scale E shape on the water gauge;
and S8, calculating to obtain the water level height according to the obtained prediction frame data of each complete scale E shape on the water gauge and the trained water level line neural network model.
2. The deep learning-based water gauge E-shaped scale recognition method according to claim 1, characterized in that: and the marked content of the water gauge picture is an enclosure box enclosing the whole water gauge.
3. The deep learning-based water gauge E-shaped scale recognition method according to claim 2, wherein the step S2 specifically comprises: and based on the water gauge data set, extracting the picture only containing the water gauge by using the bounding box marked by the water gauge data set as a water gauge scale E-shaped data set.
4. The deep learning-based water gauge E-shaped scale recognition method according to claim 2, wherein the step S3 specifically comprises: and taking the water gauge data set as a basis, utilizing the bounding box marked by the water gauge data set, taking the midpoint of the lower boundary of the marked bounding box as the image midpoint, and extracting a water line image with the width and the height of 200 pixels from each image to be used as the water line data set.
5. The deep learning-based water gauge E-shaped scale recognition method according to claim 1, characterized in that: the pre-training uses a COCO dataset to pre-train the Faster RCNN.
6. The deep learning-based water gauge E-shaped scale recognition method according to claim 1, wherein the step S7 specifically comprises:
step S71, treatInputting the picture of the water gauge into a first FasterRCNN to obtain a prediction frame B of the water gauge sg ，[x sg ,y sg ,w sg ,h sg ]Wherein x is sg 、y sg Are respectively a prediction box B sg Abscissa and ordinate of the upper left corner on the image, w sg 、h sg Predicting the width and height of the frame for the water gauge;
step S72, the obtained water gauge prediction frame B sg Corresponding picture g 1 Independently extracting the E-shaped prediction frame of each complete scale on the water gauge, and inputting the E-shaped prediction frame into a second Faster RCNN; detecting k 'E' -shaped prediction frames, wherein the lowest 'E' -shaped prediction frame is B in the vertical direction be ，[x be ,y be ,w be ,h be ]Wherein x is be 、y be Are respectively a prediction box B be Horizontal and vertical coordinates of the upper left corner on the image, w be 、h be The width and height of the prediction box.
7. The deep learning-based water gauge E-shaped scale recognition method according to claim 6, characterized in that: the water line neural network model is based on ResNet34, and the characteristic extraction part of the water line neural network model consists of 5 convolution processes.
8. The deep learning-based water gauge E-shaped scale recognition method according to claim 7, wherein the step S8 specifically comprises:
step S81, on the image, in
Taking an image g with the width and height of 200 pixels as the center point of the frame 2 Inputting the characteristic into a water level network, extracting the characteristic through a ResNet34 characteristic extractor, inputting the characteristic into a full-connection neural network, and fitting the slope a g2 And an offset b with respect to the lower left corner of the image g2 Thereby obtaining an output water line;
step S82, according to the water level line and the prediction frame B of the water gauge sg Obtaining the water line in the graph g 2 Middle point p on g2 ，
Step S83 of calculating p g2 And prediction block B be Midpoint of lower boundary
Of (2) is
Obtaining 'E' -shaped scale of the submerged part, the height of the exposed water surface is
A unit of scale;
step S84, based on the water level height H represented by the highest point of the preset water gauge db A realistic distance d corresponding to one E-shaped scale unit unit To obtain an actual water level of
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910681936.8A CN110472636B (en) | 2019-07-26 | 2019-07-26 | Deep learning-based water gauge E-shaped scale identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910681936.8A CN110472636B (en) | 2019-07-26 | 2019-07-26 | Deep learning-based water gauge E-shaped scale identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110472636A CN110472636A (en) | 2019-11-19 |
| CN110472636B true CN110472636B (en) | 2022-10-14 |
Family
ID=68508411
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910681936.8A Active CN110472636B (en) | 2019-07-26 | 2019-07-26 | Deep learning-based water gauge E-shaped scale identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110472636B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111259890A (en) * | 2020-01-19 | 2020-06-09 | 深圳市宏电技术股份有限公司 | Water level identification method, device and equipment of water level gauge |
| CN111310671B (en) * | 2020-02-19 | 2023-04-28 | 中冶赛迪信息技术(重庆)有限公司 | Heating furnace bottom water accumulation pit anomaly identification method, system and equipment based on deep learning |
| CN111488846A (en) * | 2020-04-16 | 2020-08-04 | 上海芯翌智能科技有限公司 | Method and equipment for identifying water level |
| CN111680606B (en) * | 2020-06-03 | 2021-11-23 | 淮河水利委员会水文局(信息中心) | Low-power-consumption water level remote measuring system based on artificial intelligence cloud identification water gauge |
| CN112329644A (en) * | 2020-11-06 | 2021-02-05 | 中冶赛迪重庆信息技术有限公司 | Reservoir water level monitoring method and system, medium and electronic terminal |
| CN112465061B (en) * | 2020-12-10 | 2023-03-24 | 厦门四信通信科技有限公司 | Water level identification method, device, equipment and storage medium |
| CN112598001B (en) * | 2021-03-08 | 2021-06-25 | 中航金城无人系统有限公司 | Automatic ship water gauge reading identification method based on multi-model fusion |
| CN112884753A (en) * | 2021-03-10 | 2021-06-01 | 杭州申昊科技股份有限公司 | Track fastener detection and classification method based on convolutional neural network |
| CN114067095B (en) * | 2021-11-29 | 2023-11-10 | 黄河勘测规划设计研究院有限公司 | Water level identification method based on water gauge character detection and identification |
| CN115880683B (en) * | 2023-03-02 | 2023-05-16 | 江西省水利科学院(江西省大坝安全管理中心、江西省水资源管理中心) | Urban waterlogging ponding intelligent water level detection method based on deep learning |
| CN117048773B (en) * | 2023-08-01 | 2024-09-10 | 黄岛检验认证有限公司 | Automatic tracking water gauge light supplementing double-shaft camera and water gauge observation method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018214195A1 (en) * | 2017-05-25 | 2018-11-29 | 中国矿业大学 | Remote sensing imaging bridge detection method based on convolutional neural network |
| CN109145830A (en) * | 2018-08-24 | 2019-01-04 | 浙江大学 | A kind of intelligence water gauge recognition methods |
| CN109522889A (en) * | 2018-09-03 | 2019-03-26 | 中国人民解放军国防科技大学 | Hydrological ruler water level identification and estimation method based on image analysis |
-
2019
- 2019-07-26 CN CN201910681936.8A patent/CN110472636B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018214195A1 (en) * | 2017-05-25 | 2018-11-29 | 中国矿业大学 | Remote sensing imaging bridge detection method based on convolutional neural network |
| CN109145830A (en) * | 2018-08-24 | 2019-01-04 | 浙江大学 | A kind of intelligence water gauge recognition methods |
| CN109522889A (en) * | 2018-09-03 | 2019-03-26 | 中国人民解放军国防科技大学 | Hydrological ruler water level identification and estimation method based on image analysis |
Non-Patent Citations (1)
| Title |
|---|
| 基于改进Faster RCNN与Grabcut的商品图像检测;胡正委等;《计算机系统应用》;20181114(第11期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110472636A (en) | 2019-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110472636B (en) | Deep learning-based water gauge E-shaped scale identification method | |
| CN111626190B (en) | Water level monitoring method for scale recognition based on clustering partition | |
| CN113469177B (en) | Deep learning-based drainage pipeline defect detection method and system | |
| CN109145830B (en) | Intelligent water gauge identification method | |
| CN107506798B (en) | Water level monitoring method based on image recognition | |
| CN106875381B (en) | Mobile phone shell defect detection method based on deep learning | |
| CN110619258B (en) | Road track checking method based on high-resolution remote sensing image | |
| CN108921165B (en) | Water level identification method based on water gauge image | |
| CN112215125A (en) | Water level identification method based on YOLOv3 | |
| CN111598098A (en) | Water gauge water line detection and effectiveness identification method based on full convolution neural network | |
| CN115995056A (en) | Automatic bridge disease identification method based on deep learning | |
| CN116844147A (en) | Pointer instrument identification and abnormal alarm method based on deep learning | |
| CN113469097A (en) | SSD (solid State disk) network-based real-time detection method for water surface floating object multiple cameras | |
| CN114639064B (en) | Water level identification method and device | |
| CN115375925A (en) | Underwater sonar image matching algorithm based on phase information and deep learning | |
| CN115272826A (en) | Image identification method, device and system based on convolutional neural network | |
| CN115578695B (en) | Water gauge water level machine vision detection method and device with free shooting visual angle | |
| CN113822105B (en) | Artificial intelligence water level monitoring system based on online two classifiers of SVM water scale | |
| CN115965568A (en) | Bathroom ceramic surface defect detection method and system | |
| CN112528885A (en) | Identification method for platform staff in intelligent zoning | |
| CN112862820B (en) | Image recognition technology-based intelligent acquisition device and method for water depth of manometric tube group | |
| CN117952981B (en) | Intelligent indoor lamp detection device and method based on CNN convolutional neural network | |
| CN116109869B (en) | Method and device for identifying reservoir water level fluctuation based on water bank image matching | |
| WO2024092968A1 (en) | Pavement crack detection method, medium, and system | |
| CN113554018B (en) | Meter automatic modeling method based on machine vision |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |