CN110334228A - A kind of Internet Problems map screening method based on deep learning - Google Patents
A kind of Internet Problems map screening method based on deep learning Download PDFInfo
- Publication number
- CN110334228A CN110334228A CN201910616225.2A CN201910616225A CN110334228A CN 110334228 A CN110334228 A CN 110334228A CN 201910616225 A CN201910616225 A CN 201910616225A CN 110334228 A CN110334228 A CN 110334228A
- Authority
- CN
- China
- Prior art keywords
- map
- picture
- internet
- deep learning
- screening method
- 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
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000012216 screening Methods 0.000 title claims abstract description 18
- 238000013135 deep learning Methods 0.000 title claims abstract description 14
- 238000004140 cleaning Methods 0.000 claims abstract description 8
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 6
- 230000008676 import Effects 0.000 claims abstract description 5
- 238000012549 training Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 10
- 239000000284 extract Substances 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 230000006855 networking Effects 0.000 claims description 3
- 238000012797 qualification Methods 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 2
- 238000013507 mapping Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 241000083879 Polyommatus icarus Species 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000000465 moulding Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
-
- 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
Abstract
The invention discloses a kind of Internet Problems map screening method based on deep learning, it is main include collect sample database, cleaning sample database obtains document data set, is calculated by depth convolutional neural networks algorithm document data set training and obtains cartographic model, imports cartographic model file and carry out screening identification judgement, label problem map to website picture and export.The present invention can assist Internet map supervision department, carry out quick-searching to mass network picture, identify map picture, and judge whether map is qualified；Good network and the Map Market environment are built, internet and Publication Enterprises national territory consciousness and geography information awareness of safety is strengthened conscientiously, safeguards the country's territorial sovereignty, safety and interests.
Description
Technical field
The invention belongs to image identification technical fields, and in particular to a kind of Internet Problems map based on deep learning
Screening method.
Background technique
Machine learning is learnt by analyzing mass data, is found the feature mode in data and is predicted, data
Its forecasting accuracy can be continuously improved in the promotion of richness and the accumulation of data volume, and has in the data of reply complex situations
Stronger learning ability and inclusiveness.Wherein, deep learning is as in an important research in machine learning field in recent years
Hold, has played great function in fields such as image recognition, intelligent search, Language Processing, intelligent controls.
" problem map " largely exists in internet, refers in particular to no check of drawings number mark, mistake non-national region by national table
Show, the wrong map drawn leakage and draw state boundary.Specification is published and is used, and investigating and prosecuting unlawful practice is mapping geography information supervisor at different levels
Department is duty-bound, and artificial Internet picture of consulting is the current major way for carrying out " problem map " focus efforts on special areas, workload
Greatly, inefficiency there is no a kind of method of automatic identification problem map at present.
Summary of the invention
It is an object of the invention to be directed to the deficiencies in the prior art, a kind of interconnection based on deep learning is provided
Net problem map screening method.This method can quickly recognize " the problem map " for not meeting map publication requirement, can batch
Processing identification network picture, recognition efficiency are high.
To achieve the goals above, the invention adopts the following technical scheme:
A kind of Internet Problems map screening method based on deep learning, it is main to be obtained including collecting sample database, cleaning sample database
Document data set is obtained, acquisition cartographic model is calculated document data set training, imports cartographic model file to the progress of website picture
Screening identification judgement, label problem map simultaneously export；
Specific step is as follows:
Step 1 collects a large amount of picture samples, then draws laws and regulations according to the map (such as " People's Republic of China's mapping
Method ", the laws and regulations such as " management map regulations ") requirement carry out arrangement samples pictures, samples pictures are classified as two major classes again
And it is marked: qualified standard map sample database, underproof " problem map " sample database；
Step 2 distinguishes the qualified standard map sample database obtained in step 1 and underproof " problem map " sample database
It is cleaned, is then carried out the standard map sample database of the qualification after cleaning and underproof " problem map " sample database respectively
Binary coding is compressed into two datasets file；
The cleaning step includes:
1) it deletes size and is less than 10KB or the wide high picture for being both less than 50 pixels of picture；
2) delete that clarity is high, the picture more than smudgy or watermark；
3) all remaining pictures are subjected to gray proces, linear stretch, random noise are added；
4) picture after step 3) processing is subjected to form modifying, is unified for jpg format；
Step 3, the document data set that step 2 is made are trained calculating, that is, pass through depth convolutional neural networks algorithm
(DCNN) convolutional calculation is carried out to the image array in document data set, obtains corresponding eigenmatrix, rolled up by being repeated as many times
Product calculating operation, extracts the characteristic information of image data in document data set；
The characteristic information extracted and label (two labels: qualified map, problem map) are compared into calculated result essence
Degree, and the error amount calculated progress backpropagation is gone back DCNN network parameter, after repeated multiple times calculating, if knot
Fruit precision is not improving, and as reaches fitting；Network parameter all in DCNN is saved after fitting, is saved
Data file is model file, including standard map model and " problem map " model；
Step 4 retrieves website picture, the model file that training is calculated in steps for importing three, to the net according to given network address
Picture of standing carries out identification judgement, specifically:
1) website picture is cut according to 512*512 pixel specification；
2) the website picture after cutting successively imports resnet residual error network operations；
3) it extracts and exports characteristic value；The feature that i.e. resnet residual error network can do very well according to input picture extracts,
Obtain the dimensional matrix data for indicating the characteristic value of the picture；
4) judge whether to meet map feature value；
By the 3) characteristic value that step is extracted be converted to one and be greater than 0 numerical value less than 1, if this numerical value is greater than 0.5,
It may determine that such picture is map, indicate that this picture is not map less than 0.5；
5) it is judged as NO, is then labeled as non-map picture, operation terminates；
6) it is judged as YES, is then labeled as map picture, carry out 7) step operation；
7) map picture and problem map feature are compared；
It carries out trained " problem map " model in the dimensional matrix data and step 3 extracted in 3) step to ask poor calculating,
Judge whether to be fitted " problem map " model；
8) it is judged as YES, is then labeled as problem map, operation terminates；
Step 5 exports recognition result report, the website picture of doubtful " problem map " is marked, entire mutual after label
Networking issue map screening process terminates.
The present invention further illustrates, in step 3, the characteristic information refers to the distinctive information of picture, including road network,
The information such as water system, house, text annotation.The network parameter includes the biasing of the numerical value of the weight of convolutional calculation, calculating
The critical parameter informations such as size, learning rate size.
The present invention further illustrates, in step 4, the noticeable feature refers to contour line (the i.e. map of picture
Principal moulding) or picture dominant hue (the common blue of such as map, red, green).
The present invention further illustrates, in described " problem map " model containing by " there is no Hainan on map of China ", " in
There is no sea area on state's domain " dimensional matrix data that is obtained after matrix calculates of characteristics of image.
The present invention further illustrates, in step 1, the picture sample includes obtaining from internet or camera shooting
Qualified map picture and " problem map " picture.
In the present invention, standard map model is used to judge whether the picture of input to be map picture, " problem map " mould
Type apply in next step, after being judged as YES map picture, by characteristic value check map picture there are the problem of, determine whether
For problem map.
Advantages of the present invention:
1. the present invention can assist Internet map supervision department, quick-searching is carried out to mass network picture, identifies Map
Piece, and judge whether map is qualified；Good network and the Map Market environment are built, internet and Publication Enterprises state are strengthened conscientiously
Family's domain consciousness and geography information awareness of safety, safeguard the country's territorial sovereignty, safety and interests.
2. the present invention makes full use of the advantage of image convolution, mould caused by illumination variation, color fading, movement is reduced
The influence of the factors to image recognition such as paste, complicated background, partial occlusion, improves anti-interference ability, and recognition accuracy is high, accidentally
Discrimination is low.
Detailed description of the invention
Fig. 1 is the flow chart that picture recognition is realized in one embodiment of the invention application.
Specific embodiment
The following further describes the present invention with reference to the drawings.
Embodiment 1:
A kind of Internet Problems map screening method based on deep learning, the specific steps are as follows:
Step 1 collects a large amount of picture samples, then draws laws and regulations according to the map (such as " People's Republic of China's mapping
Method ", the laws and regulations such as " management map regulations ") requirement carry out arrangement samples pictures, samples pictures are classified as two major classes again
And it is marked: qualified standard map sample database, underproof " problem map " sample database；
Step 2 distinguishes the qualified standard map sample database obtained in step 1 and underproof " problem map " sample database
It is cleaned, is then carried out the standard map sample database of the qualification after cleaning and underproof " problem map " sample database respectively
Binary coding is compressed into two datasets file；
The cleaning step includes:
1) it deletes size and is less than 10KB or the wide high picture for being both less than 50 pixels of picture；
2) delete that clarity is high, the picture more than smudgy or watermark；
3) all remaining pictures are subjected to gray proces, linear stretch, random noise are added；
4) picture after step 3) processing is subjected to form modifying, is unified for jpg format；
Step 3, the document data set that step 2 is made are trained calculating, that is, pass through depth convolutional neural networks algorithm
(DCNN) convolutional calculation is carried out to the image array in document data set, obtains corresponding eigenmatrix, rolled up by being repeated as many times
Product calculating operation, extracts the characteristic information of image data in document data set；
The characteristic information extracted and label (two labels: qualified map, problem map) are compared into calculated result essence
Degree, and the error amount calculated progress backpropagation is gone back DCNN network parameter, after repeated multiple times calculating, if knot
Fruit precision is not improving, and as reaches fitting；Network parameter all in DCNN is saved after fitting, is saved
Data file is model file, including standard map model and " problem map " model；
Step 4 retrieves website picture, the model file that training is calculated in steps for importing three, to the net according to given network address
Picture of standing carries out identification judgement, specially (as shown in Figure 1):
1) website picture is cut according to 512*512 pixel specification；
2) the website picture after cutting successively imports resnet residual error network operations；
3) it extracts and exports characteristic value；The feature that i.e. resnet residual error network can do very well according to input picture extracts,
Obtain the dimensional matrix data for indicating the characteristic value of the picture；
4) judge whether to meet map feature value；
By the 3) characteristic value that step is extracted be converted to one and be greater than 0 numerical value less than 1, if this numerical value is greater than 0.5,
It may determine that such picture is map, indicate that this picture is not map less than 0.5；
5) it is judged as NO, is then labeled as non-map picture, operation terminates；
6) it is judged as YES, is then labeled as map picture, carry out 7) step operation；
7) map picture and problem map feature are compared；
It carries out trained " problem map " model in the dimensional matrix data and step 3 extracted in 3) step to ask poor calculating,
Judge whether to be fitted " problem map " model；
8) it is judged as YES, is then labeled as problem map, operation terminates；
Step 5 exports recognition result report, the website picture of doubtful " problem map " is marked, entire mutual after label
Networking issue map screening process terminates.
Obviously, above-described embodiment is just for the sake of clearly illustrating examples made by the present invention, and is not the limit to implementation
It is fixed.For the those of ordinary skill in the field, it can also make on the basis of the above description other various forms of
Variation changes.There is no need and unable to all embodiment party's formulas with exhaustion.And it thus amplifies out obvious
Variation or variation are still in the protection scope of this invention.
Claims (6)
1. a kind of Internet Problems map screening method based on deep learning, which comprises the following steps:
Step 1 collects a large amount of picture samples, and the requirement for then drawing laws and regulations according to the map carries out arrangement samples pictures, will
Samples pictures are classified as two major classes again and are marked: qualified standard map sample database, underproof " problem map " sample
Library；
Step 2 distinguishes the qualified standard map sample database obtained in step 1 and underproof " problem map " sample database
It is cleaned, is then carried out the standard map sample database of the qualification after cleaning and underproof " problem map " sample database respectively
Binary coding is compressed into two datasets file；
The cleaning step includes:
1) it deletes size and is less than 10KB or the wide high picture for being both less than 50 pixels of picture；
2) delete that clarity is high, the picture more than smudgy or watermark；
3) all remaining pictures are subjected to gray proces, linear stretch, random noise are added；
4) picture after step 3) processing is subjected to form modifying, is unified for jpg format；
Step 3, the document data set that step 2 is made are trained calculating, that is, pass through depth convolutional neural networks algorithm
Convolutional calculation is carried out to the image array in document data set, corresponding eigenmatrix is obtained, by being repeated as many times convolutional calculation
Operation, extracts the characteristic information of image data in document data set；
The characteristic information extracted and label are compared into computational solution precision, and the error amount calculated is carried out instead
It goes back the network parameter of percentage regulation convolutional neural networks algorithm to propagation, after repeated multiple times calculating, if result precision
It no longer improves, as reaches fitting；Network parameter all in depth convolutional neural networks algorithm is saved after fitting,
Saving obtained data file is model file, including standard map model and " problem map " model；
Step 4 retrieves website picture, the model file that training is calculated in steps for importing three, to the net according to given network address
Picture of standing carries out identification judgement, specifically:
1) website picture is cut according to 512*512 pixel specification；
2) the website picture after cutting successively imports resnet residual error network operations；
3) it extracts and exports characteristic value；The feature that i.e. resnet residual error network can do very well according to input picture extracts,
Obtain the dimensional matrix data for indicating the characteristic value of the picture；
4) judge whether to meet map feature value；
By the 3) characteristic value that step is extracted be converted to one and be greater than 0 numerical value less than 1, if this numerical value is greater than 0.5,
It may determine that such picture is map, indicate that this picture is not map less than 0.5；
5) it is judged as NO, is then labeled as non-map picture, operation terminates；
6) it is judged as YES, is then labeled as map picture, carry out 7) step operation；
7) map picture and problem map feature are compared；
It carries out trained " problem map " model in the dimensional matrix data and step 3 extracted in 3) step to ask poor calculating,
Judge whether to be fitted " problem map " model；
8) it is judged as YES, is then labeled as problem map, operation terminates；
Step 5 exports recognition result report, the website picture of doubtful " problem map " is marked, entire mutual after label
Networking issue map screening process terminates.
2. the Internet Problems map screening method according to claim 1 based on deep learning, which is characterized in that in step
In rapid three, the characteristic information refers to the distinctive information of picture, including road network, water system, house, text annotation information.
3. the Internet Problems map screening method according to claim 1 based on deep learning, which is characterized in that in step
In rapid three, the network parameter includes the numerical value of the weight of convolutional calculation, the size of the biasing of calculating, learning rate size.
4. the Internet Problems map screening method according to claim 1 based on deep learning, which is characterized in that in step
In rapid four, the noticeable feature refers to the contour line of picture or the dominant hue of picture.
5. the Internet Problems map screening method according to claim 1 based on deep learning, which is characterized in that described
" problem map " model in containing will " there is no Hainan on map of China ", " not having sea area on Chinese territory " characteristics of image process
The dimensional matrix data that matrix obtains after calculating.
6. the Internet Problems map screening method according to claim 1 based on deep learning, which is characterized in that in step
In rapid one, the picture sample includes the qualified map picture and " problem map " figure obtained from internet or camera shooting
Piece.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910616225.2A CN110334228A (en) | 2019-07-09 | 2019-07-09 | A kind of Internet Problems map screening method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910616225.2A CN110334228A (en) | 2019-07-09 | 2019-07-09 | A kind of Internet Problems map screening method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110334228A true CN110334228A (en) | 2019-10-15 |
Family
ID=68143938
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910616225.2A Pending CN110334228A (en) | 2019-07-09 | 2019-07-09 | A kind of Internet Problems map screening method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110334228A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113515579A (en) * | 2021-06-17 | 2021-10-19 | 自然资源部地图技术审查中心 | Boundary peak data screening method and device |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107403200A (en) * | 2017-08-10 | 2017-11-28 | 北京亚鸿世纪科技发展有限公司 | Improve the multiple imperfect picture sorting technique of image segmentation algorithm combination deep learning |
| JP2018005520A (en) * | 2016-06-30 | 2018-01-11 | クラリオン株式会社 | Object detection device and object detection method |
| CN107690660A (en) * | 2016-12-21 | 2018-02-13 | 深圳前海达闼云端智能科技有限公司 | Image-recognizing method and device |
| CN107798340A (en) * | 2017-09-29 | 2018-03-13 | 中国地质大学（武汉） | Multiple Geochemical abnormality recognition method based on the more self-encoding encoders of space constraint |
| CN109255364A (en) * | 2018-07-12 | 2019-01-22 | 杭州电子科技大学 | A kind of scene recognition method generating confrontation network based on depth convolution |
| CN109446288A (en) * | 2018-10-18 | 2019-03-08 | 重庆邮电大学 | One kind being based on the internet Spark concerning security matters map detection algorithm |
| CN109933635A (en) * | 2019-02-13 | 2019-06-25 | 腾讯大地通途（北京）科技有限公司 | A kind of method and device updating map data base |
| CN109977191A (en) * | 2019-04-01 | 2019-07-05 | 国家基础地理信息中心 | Problem map detection method, device, electronic equipment and medium |
-
2019
- 2019-07-09 CN CN201910616225.2A patent/CN110334228A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2018005520A (en) * | 2016-06-30 | 2018-01-11 | クラリオン株式会社 | Object detection device and object detection method |
| CN107690660A (en) * | 2016-12-21 | 2018-02-13 | 深圳前海达闼云端智能科技有限公司 | Image-recognizing method and device |
| CN107403200A (en) * | 2017-08-10 | 2017-11-28 | 北京亚鸿世纪科技发展有限公司 | Improve the multiple imperfect picture sorting technique of image segmentation algorithm combination deep learning |
| CN107798340A (en) * | 2017-09-29 | 2018-03-13 | 中国地质大学（武汉） | Multiple Geochemical abnormality recognition method based on the more self-encoding encoders of space constraint |
| CN109255364A (en) * | 2018-07-12 | 2019-01-22 | 杭州电子科技大学 | A kind of scene recognition method generating confrontation network based on depth convolution |
| CN109446288A (en) * | 2018-10-18 | 2019-03-08 | 重庆邮电大学 | One kind being based on the internet Spark concerning security matters map detection algorithm |
| CN109933635A (en) * | 2019-02-13 | 2019-06-25 | 腾讯大地通途（北京）科技有限公司 | A kind of method and device updating map data base |
| CN109977191A (en) * | 2019-04-01 | 2019-07-05 | 国家基础地理信息中心 | Problem map detection method, device, electronic equipment and medium |
Non-Patent Citations (2)
| Title |
|---|
| 崔腾腾 等: "一种卷积神经网络的地图图片智能识别方法", 《测绘科学》 * |
| 黄龙: "一种快速挖掘互联网问题地图图片的方法", 《测绘与空间地理信息》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113515579A (en) * | 2021-06-17 | 2021-10-19 | 自然资源部地图技术审查中心 | Boundary peak data screening method and device |
| CN113515579B (en) * | 2021-06-17 | 2024-01-19 | 自然资源部地图技术审查中心 | Method and device for screening boundary peak data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107967475A (en) | A kind of method for recognizing verification code based on window sliding and convolutional neural networks | |
| CN113011319A (en) | Multi-scale fire target identification method and system | |
| CN106951925A (en) | Data processing method, device, server and system | |
| CN111401419A (en) | Improved RetinaNet-based employee dressing specification detection method | |
| CN106446957A (en) | Haze image classification method based on random forest | |
| CN109117703A (en) | It is a kind of that cell category identification method is mixed based on fine granularity identification | |
| CN107704878B (en) | Hyperspectral database semi-automatic establishment method based on deep learning | |
| CN107818321A (en) | A kind of watermark date recognition method for vehicle annual test | |
| CN107341508B (en) | Fast food picture identification method and system | |
| CN110458082A (en) | A kind of city management case classification recognition methods | |
| CN112749663B (en) | Agricultural fruit maturity detection system based on Internet of things and CCNN model | |
| CN109800756A (en) | A kind of text detection recognition methods for the intensive text of Chinese historical document | |
| CN112115966A (en) | Dish and attribute information identification system and method based on fine-grained identification | |
| CN114140665A (en) | Dense small target detection method based on improved YOLOv5 | |
| CN114627411A (en) | Crop growth period identification method based on parallel detection under computer vision | |
| CN114120094A (en) | Water pollution identification method and system based on artificial intelligence | |
| CN109086781A (en) | A kind of cabinet lamp state identification method based on deep learning | |
| CN110334228A (en) | A kind of Internet Problems map screening method based on deep learning | |
| CN105404682B (en) | A kind of book retrieval method based on digital image content | |
| CN114387529A (en) | Black and odorous water body extraction method based on CART classification model | |
| CN112906510A (en) | Fishery resource statistical method and system | |
| CN110349119B (en) | Pavement disease detection method and device based on edge detection neural network | |
| CN112183640A (en) | Detection and classification method based on irregular object | |
| CN111414855A (en) | Telegraph pole sign target detection and identification method based on end-to-end regression model | |
| CN115760854A (en) | Deep learning-based power equipment defect detection method and device and electronic equipment |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191015 |
|
| RJ01 | Rejection of invention patent application after publication |