CN110472636A - Water gauge E font scale recognition methods based on deep learning - Google Patents
Water gauge E font scale recognition methods based on deep learning Download PDFInfo
- Publication number
- CN110472636A CN110472636A CN201910681936.8A CN201910681936A CN110472636A CN 110472636 A CN110472636 A CN 110472636A CN 201910681936 A CN201910681936 A CN 201910681936A CN 110472636 A CN110472636 A CN 110472636A
- Authority
- CN
- China
- Prior art keywords
- water gauge
- font
- water
- scale
- water level
- 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.)
- Granted
Links
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 water gauge E font scale recognition methods based on deep learning that the present invention relates to a kind of, comprising the following steps: step S1: acquisition water gauge picture, and it is labeled processing, obtain water gauge data collection；Step S2: water gauge scale E font data collection is extracted；Step S3: water level line data set is extracted；Step S4: the Faster RCNN of building extractor characterized by ResNe101, and pre-training；Step S5: the Faster RCNN after pre-training is respectively trained according to water gauge data collection and water gauge scale E font data collection；Step S6: building water level line neural network model, and be trained using water level line data line；Step S7: will sequentially input the first Faster RCNN and the 2nd Faster RCNN to draft picture, obtain the prediction frame data of each complete scale E font on water gauge；Step S8: by the water level line neural network model after the prediction frame data input training of complete scale E font each on obtained water gauge, height of water level is calculated.
Description
Technical field
The water gauge E font scale recognition methods based on deep learning that the present invention relates to a kind of.
Background technique
Water level is the reflection most intuitive factor of water body regimen, its variation causes mainly due to the increase and decrease of water body water
's.Water level refers to that elevation of the table relative to a certain basal plane, the water surface claim the depth of water with a distance from river bed.It is special that water level is higher than some
When fixed height, the disasters such as there will be waterlogging, breach a dyke.That is, this height is the water bodys such as rivers, lake, ocean, reservoir
The safety line of water surface elevation, often be referred to as water level alarm line, different rivers, lake, ocean, reservoir water level alarm line height not
One.Precisely, SEA LEVEL VARIATION is systematically perceived, could be gone after profits and advoided disadvantages during coping with SEA LEVEL VARIATION, tune of adopting an effective measure
Control water level.
Water gauge is the important tool for measuring SEA LEVEL VARIATION, and carrying out real-time monitoring to water gauge by camera is existing monitoring water
One of the main method of position.Existing water gauge scale recognition methods and is manually set by traditional image processing means
The characteristics of image of meter is compared, to identify water gauge scale.This traditional water gauge scale recognition methods has one in application
Fixed disadvantage: because being the characteristics of image of engineer, the multiplicity variation of natural environment can not be effectively adapted to, such as: water gauge inverted image
It is too strong, there is dirt etc. on water gauge.For existing outstanding problem in this respect, the accuracy of water gauge scale identification is improved, accurately
Ground perceives the variation of water level, is the inevitable requirement of water conservancy management work.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of water gauge E font scale identification side based on deep learning
Method is right under changeable natural environment (illumination, block, distort) overcome in a manner of traditional water detection well the shortcomings that
The identification of water gauge scale still has stronger robustness.
To achieve the above object, the present invention adopts the following technical scheme:
A kind of water gauge E font scale recognition methods based on deep learning, comprising the following steps:
Step S1: acquisition water gauge picture, and it is labeled processing, obtain water gauge data collection；
Step S2: according to obtained water gauge data collection, water gauge scale E font data collection is extracted；
Step S3: according to obtained water gauge data collection, water level line data set is extracted；
Step S4: the Faster RCNN of building extractor characterized by ResNe101, and pre-training, after obtaining pre-training
Faster RCNN；
Step S5: the Faster after pre-training is respectively trained according to water gauge data collection and water gauge scale E font data collection
RCNN, the first Faster RCNN and the 2nd Faster RCNN after being trained；
Step S6: building water level line neural network model, and be trained using water level line data set, after being trained
Water level line neural network model；
Step S7: the first Faster RCNN and the 2nd Faster RCNN will be sequentially input to draft picture, will be obtained
Obtain the prediction frame data of each complete scale E font on water gauge；
Step S8: according to the water level line after the prediction frame data of complete scale E font each on obtained water gauge and training
Height of water level is calculated in neural network model.
Further, the content of the water gauge picture mark is to surround the bounding box of entire water gauge.
Further, the step S2 specifically: it is based on water gauge data collection, the bounding box marked using water gauge data collection,
Picture only comprising water gauge is extracted, as water gauge scale E font data collection.
Further, the step S3 specifically: based on water gauge data collection, utilize the encirclement of water gauge data collection mark
Box, using the midpoint of the lower boundary of the bounding box of mark as image midpoint, every picture extracts the water level of one wide a height of 200 pixel
Line image, as water level line data set.
Further, the pre-training carries out pre-training to Faster RCNN using COCO data set.
Further, the step S7 specifically:
Step S71: the first Faster RCNN will be inputted to draft picture, and will obtain the prediction block B of water gaugesg, [xsg,ysg,
wsg,hsg], wherein xsg、ysgRespectively prediction block BsgCross, the ordinate of the upper left corner on the image, wsg、hsgFor water gauge prediction block
It is wide, high；
Step S72: the water gauge prediction block B that will be arrivedsgCorresponding picture g1It individually extracts, and inputs the 2nd Faster
RCNN obtains the prediction block of each complete scale E font on water gauge；Detect k ' E ' font prediction blocks, in the vertical direction,
Nethermost ' E ' font prediction block is Bbe, [xbe,ybe,wbe,hbe], wherein xbe、ybeRespectively prediction block BbeScheming in the upper left corner
As upper cross, ordinate, wbe、hbeWidth, height for prediction block.
Further, the water level line neural network model based on ResNet34, rolled up by 5 by characteristic extraction part
Product process composition.
Further, the step S8 specifically:
Step S81: on the image, withFor the central point of frame, one wide a height of 200 pixel is taken
Image g2, be input in water level line network, feature extracted by ResNet34 feature extractor, after feature be input to connect entirely
Meet neural network, fit slope ag2With the offset b relative to the image lower left cornerg2, to obtain one water level line of output；
Step S82: according to the prediction block B of water level line and water gaugesg, the water level line is obtained in figure g2On midpoint pg2,
Step S83: p is calculatedg2With prediction block BbeThe midpoint of lower boundaryDistanceIt obtains being flooded not
Partial ' E ' font scale, the height to expose the surface areA scale unit；
Step S84: the height of water level H represented based on default water gauge highest pointdb, an E font scale unit it is corresponding existing
Actual distance is from dunit, obtaining actual height of water level is
Compared with the prior art, the invention has the following beneficial effects:
The present invention can be very good the shortcomings that overcoming traditional water detection mode, in changeable natural environment (illumination, screening
Gear, distortion etc.) under, still there is stronger robustness to the identification of water gauge scale.
Detailed description of the invention
Fig. 1 is process principle figure of the present invention；
Fig. 2 is the water gauge data image in one embodiment of the invention；
Fig. 3 is the water gauge prediction block of the water gauge image zooming-out in one embodiment of the invention；
Fig. 4 is the E font prediction block extracted in one embodiment of the invention；
Fig. 5 is water level line image obtained in one embodiment of the invention；
Fig. 6 is a depth residual error method block in one embodiment of the invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
Fig. 1 is please referred to, the present invention provides a kind of water gauge E font scale recognition methods based on deep learning, including following
Step:
Step S1: acquisition water gauge picture, and it is labeled processing, obtain water gauge data collection；
Step S2: according to obtained water gauge data collection, water gauge scale E font data collection is extracted；
Step S3: according to obtained water gauge data collection, water level line data set is extracted；
Step S4: the Faster RCNN of building extractor characterized by ResNe101, and use COCO data set pair
Faster RCNN carries out pre-training, the Faster RCNN after obtaining pre-training；
Step S5: the Faster after pre-training is respectively trained according to water gauge data collection and water gauge scale E font data collection
RCNN, the first Faster RCNN and the 2nd Faster RCNN after being trained；
Step S6: building water level line neural network model, and be trained using water level line data line, after being trained
Water level line neural network model；
Step S7: the first Faster RCNN and the 2nd Faster RCNN will be sequentially input to draft picture, will be obtained
Obtain the prediction frame data of each complete scale E font on water gauge；
Step S8: by the water level line after the prediction frame data input training of complete scale E font each on obtained water gauge
Height of water level is calculated in neural network model.
In the present embodiment, the content of the water gauge picture mark is to surround the bounding box of entire water gauge.In data set
Picture considers following element during collection: the 1. interference of hot spot；2. the different interference of intensity of illumination；3. spray is dry
It disturbs；4. the interference of background；5. the interference 7. conditions such as interference of different weather condition of water stain interference 6. inverted image.
In the present embodiment, it is based on water gauge data collection, the bounding box marked using water gauge data collection will only include water gauge
Picture extract, as water gauge scale E font data collection.The picture concentrated to data is labeled.The content of mark are as follows:
Surround the bounding box of each ' E ' font water gauge scale.Bounding box has and the water gauge scale of only one complete ' E ' font.
In the present embodiment, the step S3 specifically: based on water gauge data collection, utilize water gauge data collection mark
Bounding box, using the midpoint of the lower boundary of the bounding box of mark as image midpoint, every picture extracts one wide a height of 200 pixel
Water level line image, as water level line data set.The picture concentrated to data is labeled.The content of mark are as follows: with water level line phase
As straight line (slope, offset).
In the present embodiment, the step S7 specifically:
Step S71: the first Faster RCNN will be inputted to draft picture, and will obtain the prediction block B of water gaugesg, [xsg,ysg,
wsg,hsg], wherein xsg、ysgRespectively prediction block BsgCross, the ordinate of the upper left corner on the image, wsg、hsgFor water gauge prediction block
It is wide, high；
Step S72: the water gauge prediction block B that will be arrivedsgCorresponding picture g1It individually extracts, and inputs the 2nd Faster
RCNN obtains the prediction block of each complete scale E font on water gauge；Detect k ' E ' font prediction blocks, in the vertical direction,
Nethermost ' E ' font prediction block is Bbe, [xbe,ybe,wbe,hbe], wherein xbe、ybeRespectively prediction block BbeScheming in the upper left corner
As upper cross, ordinate, wbe、hbeWidth, height for prediction block.
In the present embodiment, the water level line neural network model is based on ResNet34, characteristic extraction part
It is made of 5 convolution process.Water level line neural network model is as follows:
Table 1
The construction of block is as shown in fig. 6, the subsequent number of block is the storehouse quantity of block in table 1.Down-sampling by Conv3_1,
Conv4_1 and Conv5_1 is executed, stride 2.
As shown in table 1, water level line neural network model is based on ResNet34, and the characteristic extraction part of the network is by 5
A convolution process composition.
Wherein all convolution operations, the step-length of pondization operation are all 2.In Conv_1, with size for 7 × 7, port number 64
Convolution kernel carry out convolution operation.In Conv2_x, 3 × 3 maximum pondizations are carried out first and are operated, be followed by containing 3 storehousesResidual error method block carries out convolution operation.Residual error method block details are as shown in Figure 6: it is 3 that input I, which successively carries out size,
× 3, the convolution operation that port number is 64, the operation of relu nonlinear activation, size is that the convolution operation that 3 × 3, port number is 64 obtains
To M.I is followed by upper relu activation, the Feature Mapping exported by corresponding be added in channel with the finger of M again.Conv3_x,
The realization of Conv4_x, Conv5_x with it is described above consistent, only the parameter of residual error method block and the number of residual error method block be not
Equally, this can refer to table 1.Down-sampling operation, step-length are executed in addition, having after Conv3_1, Conv4_1 and Conv5_1
It is 2.What Conv3_1 was represented is first residual error method block of Conv3_x.After Conv1-Conv5_x characteristic extraction step,
It will obtain the advanced features mapping of image.
After carrying out convolution operation, obtained Feature Mapping is separately input to 2 branches.2 branches structure,
It is consistent in design: the Feature Mapping of input being subjected to the average pond that step-length is 2 first, is followed by the full connection of one 1 dimension
Layer, obtains output valve.Loss function is the MSE (mean square error) of network output and mark value.It is trained, 2 branches are final
It will be fitted output offset amount and slope respectively.
In the present embodiment, the step S8 specifically:
Step S81: on the image, withFor the central point of frame, one wide a height of 200 pixel is taken
Image g2, be input in water level line network, feature extracted by ResNet34 feature extractor, after feature be input to connect entirely
Meet neural network, fit slope ag2With the offset b relative to the image lower left cornerg2, to obtain one water level line of output；
Step S82: according to the prediction block B of water level line and water gaugesg, the water level line is obtained in figure g2On midpoint pg2,
Step S73: p is calculatedg2With prediction block BbeThe midpoint of lower boundaryDistanceIt obtains being flooded not
Partial ' E ' font scale, the height to expose the surface areA scale unit；
Step S74: the height of water level H represented based on default water gauge highest pointdb, ' E ' font scale unit it is corresponding
Reality distance dunit, obtaining actual height of water level is
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, is all covered by the present invention.
Claims (8)
1. a kind of water gauge E font scale recognition methods based on deep learning, which comprises the following steps:
Step S1: acquisition water gauge picture, and it is labeled processing, obtain water gauge data collection；
Step S2: according to obtained water gauge data collection, water gauge scale E font data collection is extracted；
Step S3: according to obtained water gauge data collection, water level line data set is extracted；
Step S4: the Faster RCNN of building extractor characterized by ResNe101, and pre-training, after obtaining pre-training
Faster RCNN；
Step S5: the Faster RCNN after pre-training is respectively trained according to water gauge data collection and water gauge scale E font data collection,
The first Faster RCNN and the 2nd Faster RCNN after being trained；
Step S6: building water level line neural network model, and be trained using water level line data set, the water level after being trained
Line neural network model；
Step S7: the first Faster RCNN and the 2nd Faster RCNN will be sequentially input to draft picture, obtains water
The prediction frame data of each complete scale E font on ruler；
Step S8: according to the water level line nerve after the prediction frame data of complete scale E font each on obtained water gauge and training
Height of water level is calculated in network model.
2. the water gauge E font scale recognition methods according to claim 1 based on deep learning, it is characterised in that: described
The content of water gauge picture mark is to surround the bounding box of entire water gauge.
3. the water gauge E font scale recognition methods according to claim 2 based on deep learning, which is characterized in that described
Step S2 specifically: be based on water gauge data collection, the bounding box marked using water gauge data collection will only include the picture extraction of water gauge
Out, as water gauge scale E font data collection.
4. the water gauge E font scale recognition methods according to claim 2 based on deep learning, which is characterized in that described
Step S3 specifically: based on water gauge data collection, the bounding box marked using water gauge data collection, under the bounding box of mark
The midpoint on boundary is image midpoint, and every picture extracts the water level line image of one wide a height of 200 pixel, as water level line data
Collection.
5. the water gauge E font scale recognition methods according to claim 1 based on deep learning, it is characterised in that: described
Pre-training carries out pre-training to Faster RCNN using COCO data set.
6. the water gauge E font scale recognition methods according to claim 1 based on deep learning, which is characterized in that described
Step S7 specifically:
Step S71: the first FasterRCNN will be inputted to draft picture, and will obtain the prediction block B of water gaugesg, [xsg,ysg,wsg,
hsg], wherein xsg、ysgRespectively prediction block BsgCross, the ordinate of the upper left corner on the image, wsg、hsgFor water gauge prediction block width,
It is high；
Step S72: the water gauge prediction block B that will be arrivedsgCorresponding picture g1It individually extracts, and inputs the 2nd Faster RCNN,
Obtain the prediction block of each complete scale E font on water gauge；Detect k ' E ' font prediction blocks, it is in the vertical direction, most lower
' E ' the font prediction block in face is Bbe, [xbe,ybe,wbe,hbe], wherein xbe、ybeRespectively prediction block BbeThe upper left corner is on the image
Cross, ordinate, wbe、hbeWidth, height for prediction block.
7. the water gauge E font scale recognition methods according to claim 6 based on deep learning, it is characterised in that: described
Based on ResNet34, characteristic extraction part is made of water level line neural network model 5 convolution process.
8. the water gauge E font scale recognition methods according to claim 7 based on deep learning, which is characterized in that described
Step S8 specifically:
Step S81: on the image, withFor the central point of frame, the image of one wide a height of 200 pixel is taken
g2, be input in water level line network, feature extracted by ResNet34 feature extractor, after feature is input to full connection nerve
Network, fit slope ag2With the offset b relative to the image lower left cornerg2, to obtain one water level line of output；
Step S82: according to the prediction block B of water level line and water gaugesg, the water level line is obtained in figure g2On midpoint pg2,
Step S83: p is calculatedg2With prediction block BbeThe midpoint of lower boundaryDistanceIt obtains being flooded not
Partial ' E ' font scale, the height to expose the surface areA scale unit；
Step S84: the height of water level H represented based on default water gauge highest pointdb, the corresponding reality of an E font scale unit away from
From dunit, obtaining actual height of water level is
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 true CN110472636A (en) | 2019-11-19 |
| CN110472636B 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) |
Cited By (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 |
| CN111310671A (en) * | 2020-02-19 | 2020-06-19 | 中冶赛迪重庆信息技术有限公司 | Heating furnace bottom sump abnormity 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 |
| CN111680606A (en) * | 2020-06-03 | 2020-09-18 | 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 |
| CN112465061A (en) * | 2020-12-10 | 2021-03-09 | 厦门四信通信科技有限公司 | Water level identification method, device, equipment and storage medium |
| CN112598001A (en) * | 2021-03-08 | 2021-04-02 | 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 |
| CN114067095A (en) * | 2021-11-29 | 2022-02-18 | 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 |
| CN115880683A (en) * | 2023-03-02 | 2023-03-31 | 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 |
| CN117048773A (en) * | 2023-08-01 | 2023-11-14 | 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 |
|---|
| 胡正委等: "基于改进Faster RCNN与Grabcut的商品图像检测", 《计算机系统应用》 * |
Cited By (14)
| 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 |
| CN111310671A (en) * | 2020-02-19 | 2020-06-19 | 中冶赛迪重庆信息技术有限公司 | Heating furnace bottom sump abnormity identification method, system and equipment based on deep learning |
| 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 |
| CN111680606A (en) * | 2020-06-03 | 2020-09-18 | 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 |
| CN112465061A (en) * | 2020-12-10 | 2021-03-09 | 厦门四信通信科技有限公司 | Water level identification method, device, equipment and storage medium |
| CN112598001A (en) * | 2021-03-08 | 2021-04-02 | 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 |
| CN114067095A (en) * | 2021-11-29 | 2022-02-18 | 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 |
| 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 |
| CN115880683A (en) * | 2023-03-02 | 2023-03-31 | 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 |
| CN117048773A (en) * | 2023-08-01 | 2023-11-14 | 黄岛检验认证有限公司 | Automatic tracking water gauge light supplementing double-shaft camera and water gauge observation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110472636B (en) | 2022-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110472636A (en) | Water gauge E font scale recognition methods based on deep learning | |
| CN113537177B (en) | Flood disaster monitoring and disaster situation analysis method based on visual Transformer | |
| CN108921165B (en) | Water level identification method based on water gauge image | |
| CN106557764A (en) | A kind of water level recognition methodss based on binary-coded character water gauge and image procossing | |
| CN105761286A (en) | Water color exception object extraction method and system based on multi-spectral remote sensing image | |
| CN109764930B (en) | Water gauge water line visual detection method suitable for complex illumination conditions | |
| CN106682664A (en) | Water meter disc area detection method based on full convolution recurrent neural network | |
| CN107833203A (en) | A kind of horizontal plane identification and water level method for real-time measurement based on image procossing | |
| CN109376740A (en) | A kind of water gauge reading detection method based on video | |
| CN102156984B (en) | Method for determining optimal mark image by adaptive threshold segmentation | |
| CN107631782A (en) | A kind of level testing methods based on Harris Corner Detections | |
| CN112418124B (en) | Intelligent fish monitoring method based on video image | |
| CN106503695A (en) | A kind of tobacco plant identification and method of counting based on Aerial Images | |
| CN103292792A (en) | Actual measurement SVP reconstruction method suitable for submarine detection and pseudo-landform processing | |
| CN108053412A (en) | The remote sensing image optimum segmentation result for merging more quantitative assessing index determines method | |
| CN110889840A (en) | Effectiveness detection method of high-resolution 6 # remote sensing satellite data for ground object target | |
| CN114639064B (en) | Water level identification method and device | |
| CN111882573B (en) | Cultivated land block extraction method and system based on high-resolution image data | |
| CN105405138A (en) | Water surface target tracking method based on saliency detection | |
| CN115880683B (en) | Urban waterlogging ponding intelligent water level detection method based on deep learning | |
| CN110211128A (en) | Loess plateau terraced fields extracting method based on remote sensing image and DEM | |
| CN113469097A (en) | SSD (solid State disk) network-based real-time detection method for water surface floating object multiple cameras | |
| CN116630352A (en) | Rock core measurement method and device based on bidirectional cascade pixel differential network | |
| CN102542259B (en) | Identification method for near-shore on-land water body | |
| CN107067384B (en) | A kind of loess plateau terraced fields extracting method based on ray method |
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 |