CN104504389B - A kind of satellite cloudiness computational methods based on convolutional neural networks - Google Patents
A kind of satellite cloudiness computational methods based on convolutional neural networks Download PDFInfo
- Publication number
- CN104504389B CN104504389B CN201410798632.7A CN201410798632A CN104504389B CN 104504389 B CN104504389 B CN 104504389B CN 201410798632 A CN201410798632 A CN 201410798632A CN 104504389 B CN104504389 B CN 104504389B
- Authority
- CN
- China
- Prior art keywords
- cloud
- satellite
- layer
- neural networks
- convolutional neural
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
It is satellite cloudiness computational methods that the invention discloses a kind of based on convolutional neural networks, first establish the satellite cloud picture training sample for including 6000 ~ 8000 training samples, the spissatus of each 2000 ~ 3000 sample is marked out in satellite cloud picture manually, Bao Yun and clear sky cloud atlas block, in this, as the training sample of convolutional neural networks；Training sample and satellite cloud picture are carried out pretreatment again to input as the data of convolutional neural networks, convolutional neural networks detection is then carried out, each spissatus in cloud atlas, Bao Yun and clear sky region position is detected with this；Finally according to the position of spissatus in cloud atlas, Bao Yun and clear sky, its gray value is calculated separately, is calculated according to its gray value to carry out the cloud amount of satellite cloud picture.Satellite cloud picture image can be incorporated neural network by the present invention directly as the input of CNN, and by feature extraction functions, and the implicit feature to image extracts, more more convenient than the prior art and accurate, have important application value.
Description
Technical field
The present invention relates to meteorological detection field more particularly to a kind of satellite cloudiness calculating sides based on convolutional neural networks
Method.
Background technology
Cloud is one of most important factor in synoptic climate, on the one hand adjusts earth atmosphere method internal radiation balance, separately
On the one hand it plays an important role to Water Cycle, therefore, the observation of cloud plays an important roll.And the side manually to estimate is relied on for a long time
Method becomes the bottleneck of meteorological satellite automatic prediction, and the automatic identification of cloud atlas becomes urgent demand.
Detection, cloud classification based on satellite image development cloud simultaneously calculate the main side that cloud amount is the global cloud amount distribution of acquisition
Formula.Currently, satellite cloudiness computational methods mainly have ISCCP methods in the world, by ISCCP multi-threshold cloud detection method of optic, by picture
Member is divided into clear sky and has two class of cloud；There are CLAVR-1 methods, pixel is divided into clear sky, mixes and have cloud three classes；There is the side CLAVR-X
Pixel is divided into full cloud, mixed cloud, mixing clear sky and four class of situation by method；Also MODIS methods, are divided into determining cloud, can by pixel
It can four class of cloud, possible clear sky and determining clear sky；Also such as UW HIRS, NIR/VIS methods etc..Above-mentioned cloud amount computational methods can
To be generally divided into two classes：First, based on thering is the ratio between cloud pixel and total pixel to calculate cloud amount in region；Another kind is to be based on
Pixel amount of radiation/reflectivity calculates equivalent cloud amount.First kind method is easy to operate, but cannot analyze sub-pixed mapping cloud amount, often leads
Cause result of calculation higher；Second class method solves the problems, such as sub-pixed mapping cloud amount to a certain degree, but for multi layer cloud and ground surface type
Change violent situation to be less applicable in.No matter any computational methods, accuracy both depends on the precision of cloud detection result.
Mainly there are threshold method and neural network, wherein the accuracy of identification quilt of neural network to cloud detection research both at home and abroad at present
It generally believes and is higher than other graders.Although neural-network classification method has unique advantage in numerous methods,
It there are problems that.Traditional neural network uses the Gradient learning method (BP) of error feedback, relatively slow with pace of learning,
Iterations are excessive, solve the shortcomings of being easy to be absorbed in local minimum, these disadvantages have seriously affected neural network in cloud classification
Application.
Deep learning forms more abstract high-rise expression attribute classification or feature by combining low-level feature, to find number
According to distributed nature indicate.It can be insufficient existing for effective solution existing method.Grader is core in cloud classification model
The validity of the heart, model directly affects cloud atlas intellectual analysis result.Since convolutional neural networks have adaptive, self study and non-
Linear approximation ability so that it during realizing cloud classification advantageously than some other algorithm.
Invention content
Goal of the invention：The present invention detects the various disadvantages of grader, the not high technology of cloud amount accuracy of detection for current cloud atlas
Deficiency provides a kind of satellite cloudiness computational methods based on convolutional neural networks by lot of experiments.
Technical solution：To achieve the above object, the technical solution that the present invention takes is：
Steps are as follows for a kind of cloud amount computational methods based on convolutional neural networks：
(1) the satellite cloud picture training sample for including 6000~8000 samples is established, is marked out in satellite cloud picture manually each
Spissatus, Bao Yun and the clear sky cloud atlas block of 2000~3000 samples, in this, as the training sample of convolutional neural networks,
(2) training sample and satellite cloud picture are carried out pretreatment to input as the data of convolutional neural networks, is then carried out
Convolutional neural networks detect, and each spissatus in cloud atlas, Bao Yun and clear sky region position are detected with this；
(3) according to the position of spissatus, Bao Yun and clear sky in cloud atlas after detection, its gray value is calculated separately, according to its gray scale
The cloud amount that value carrys out satellite cloud picture calculates.
Preferably, above-described a kind of satellite cloudiness computational methods based on convolutional neural networks, the step
Suddenly (1) specifically includes：
1) the HJ-1A/1B satellite cloud picture data needed for being downloaded in Chinese Resources satellite hub；
2) collector is utilized to acquire spissatus, Bao Yun and the clear sky cloud of 39*39 pixels on HJ-1A/1B satellite cloud pictures respectively
Each 2000 pieces of block, is uniformly scaled 32*32 pixels.
Preferably, the above-described satellite cloudiness computational methods based on convolutional neural networks, to step (2) institute
It is this data modes of 32*32*X that the pretreatment needed, which is by whole picture satellite cloud picture format conversion, and wherein X is quantity, is then trained
Sample is inputted together as convolutional neural networks；
The convolutional neural networks include 7 layers, and the 1st layer is input layer, and the 2nd layer is volume base, also referred to as Feature Mapping
Layer is extracted the different characteristic of picture by the convolution kernel of multiple 5*5, includes the Feature Mapping figure of 12 28*28, under the 3rd layer is
Sample level, also referred to as feature extraction layer are made of, each neuron of characteristic pattern and the 1st characteristic pattern that 12 sizes are 14*14
The fields 2*2 of layer are connected, and the 4th layer of convolutional layer being made of the characteristic pattern of 16 10*10, the 5th layer is down-sampling as the 3rd layer
Layer includes the characteristic pattern of 16 5*5, and the 6th layer is full articulamentum, shares 400 tie points, last layer is output layer, there is 3
Node respectively represents spissatus, Bao Yun and clear sky.
The calculation of wherein convolutional layer is as follows：
Wherein l is the number of plies of network, and K is convolution kernel, MjIndicate the set of input maps.
The calculation of down-sampling layer is as follows：
Wherein, down () indicates that down-sampling function, β and b correspond to the characteristic pattern each exported respectively.
Satellite cloud picture after detection is showed with redgreenblue and gray-scale map respectively；
The spissatus lowest gray value of red area, the highest of the thin cloud of green area are found out respectively according to the color diagram detected
The average gray value of gray value and blue region clear sky calculates cloud amount according to cloud amount calculation formula.
Advantageous effect：Compared to the prior art the present invention has the advantages that：
The present invention is screened by many experiments, designs a kind of satellite cloudiness computational methods based on convolutional neural networks, is made
It can be satellite cloud picture image directly as the input of CNN, and by feature extraction work(for the convolutional neural networks of deep learning
Neural network can be incorporated, the implicit feature to image extracts, and more convenient and accurate compared to manually extracting, weights are total
Enjoying reduces the training parameter of network, can reduce the complexity of neural network, adapts to the demand of present big data quantity.
The detection of satellite cloud picture medium cloud is the premise of interpretation of satellite image, for the deficiency of threshold method cloud detection, the present invention
Using the global and local feature of cloud, based on convolutional neural networks into the semantic feature study racked and cloud classification, in cloud detection
On the basis of improve Spatial coherence method based on reflectivity and calculate total amount of cloud, improve and improve cloud classification algorithm and cloud amount calculates, be
Solid theoretical foundation is established in comprehensive automatic detection of satellite cloud picture.
Description of the drawings
Fig. 1 is a kind of satellite cloudiness computational methods schematic diagram based on convolutional neural networks of the present invention；
Fig. 2 is convolutional neural networks structural schematic diagram of the present invention；
Fig. 3 is convolutional neural networks detects schematic diagram of the present invention；
Fig. 4 is convolution sum down-sampling schematic diagram during convolutional neural networks of the present invention；
Specific implementation mode
The technical solution further illustrated the present invention below in conjunction with the accompanying drawings；
Shown in institute Fig. 1, a kind of satellite cloudiness computational methods based on convolutional neural networks, including sample acquisition and processing,
Convolutional neural networks are trained, and satellite cloud detection and cloud amount calculate four-stage, and the sample acquisition and processing include following step
Suddenly：
(1) acquisition of satellite cloud picture sample comes from China Satecom's landsat HJ-1A/1B Satellite CCDs channel；
(2) 6000 sample 39*39 pixel cloud masses are acquired by collector from satellite cloud picture, spissatus, Bao Yun and clear sky are each
2000 pieces；
(3) 6000 samples are uniformly scaled 32*32 pixels, in addition whole picture satellite cloud picture format conversion is 32*32*X numbers
According to format, wherein X is quantity；
Convolutional neural networks training of the present invention includes the following steps：
(1) using 4200 samples in 6000 cloud mass samples as training sample, spissatus, Bao Yun and each 1400 pieces of clear sky, test
Sample is 1800 pieces, spissatus.Bao Yun and each 600 pieces of clear sky；
(2) training is the structure of convolutional neural networks as shown in Fig. 2, by constantly training and test cloud mass sample, constantly
Parameter in adjustment convolutional neural networks determines parameter on the basis of cloud mass verification and measurement ratio, is spread for the detection of whole picture satellite cloud picture
Pad；
Satellite cloud picture detection of the present invention includes the following steps：
(1) it is inputted pretreated whole picture satellite cloud picture as the data of convolutional neural networks, carries out cloud detection, wherein
The process of cloud mass detection is as shown in Figure 3；
(2) convolutional neural networks include 7 layers, and the 1st layer is input layer, and the 2nd layer is to roll up base, also referred to as Feature Mapping layer,
The different characteristic that picture is extracted by the convolution kernel of multiple 5*5 includes the Feature Mapping figure of 12 28*28, and the 3rd layer is down-sampling
Layer, also referred to as feature extraction layer are made of the characteristic pattern that 12 sizes are 14*14, each neuron of characteristic pattern and the 1st layer
The fields 2*2 are connected, and the 4th layer of convolutional layer being made of the characteristic pattern of 16 10*10, the 5th layer is down-sampling layer as the 3rd layer,
Include the characteristic pattern of 16 5*5, the 6th layer is full articulamentum, shares 400 tie points, last layer is output layer, there are 3 sections
Point respectively represents spissatus, Bao Yun and clear sky.
(3) as shown in figure 4, volume base is Feature Mapping layer, sub-sampling layer is characterized the learning process of convolutional neural networks
Extract layer, with a trainable filter fxThen plus a bigoted b it deconvolutes the global and local feature of satellite cloud picture,x,
Obtain volume base Cx.The form of calculation of convolutional layer is as follows：
Wherein l is the number of plies of network, and K is convolution kernel, MjIndicate the set of input maps.
Sub-sampling procedures include summing per field, then passing through Wx+1Weighting, then biasing hold bx+1, then pass through one
sigmoid
Activation primitive generates the Feature Mapping layer S of a diminutionx+1, calculation formula is：
Wherein, down () indicates that down-sampling function, β and b correspond to the characteristic pattern each exported respectively.
(4) result of cloud detection is distinguished with red green blue, and wherein red represents spissatus, and green represents Bao Yun, blue
Represent clear sky.
Testing result cloud amount calculating of the present invention includes the following steps：
(1) in order to solve the problems, such as part cloud desk, a kind of improved " space correlation based on reflectivity of this research and utilization
Method " calculates total amount of cloud.The basic principle of " Spatial coherence method " is based on to single pixel amount of radiation and clear sky and spissatus feelings
The detection of amount of radiation under condition, obtains the total amount of cloud of single pixel, and calculation formula is as follows：
I=(1-Ac)Iclr+AcIcld
In formula, I is the amount of radiation that pixel receives；AcIndicate pixel total amount of cloud；Iclr、IcldThe highest ash of clear sky is indicated respectively
Angle value, spissatus lowest gray value amount of radiation.Conversion can obtain as a result, and total amount of cloud is：
(2) cloud amount calculates, and between the cloud amount of each pixel is 0 to 1,1 represents spissatus, and 0 represents clear sky.The present invention utilizes
200 figures randomly selected are tested, and are tested in such a way that expert marks.The cloud amount of this method calculates rate of accuracy reached and arrives
84.3%.And the accuracy rate that conventional threshold values method obtains under kindred circumstances is 75.4%.Illustrating that this method is in practical applications can
Capable, accuracy rate is higher than the prior art, has important application value.
The above is only some embodiments of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (2)
1. based on convolutional neural networks it is satellite cloudiness computational methods a kind of, which is characterized in that it includes the following steps：
(1) the satellite cloud picture training sample for including 6000~8000 training samples is established, is marked out in satellite cloud picture manually each
Spissatus, Bao Yun and the clear sky cloud atlas block, in this, as the training sample of convolutional neural networks of 2000~3000 samples；
(2) training sample and satellite cloud picture are carried out pretreatment to input as the data of convolutional neural networks, then carries out convolution
Neural network detects, and each spissatus in cloud atlas, Bao Yun and clear sky region position are detected with this；
(3) according to the position of spissatus, Bao Yun and clear sky in cloud atlas after detection, calculate separately its gray value, according to its gray value come
The cloud amount for carrying out satellite cloud picture calculates；
The step (1) specifically includes：
1) the HJ-1A/1B satellite cloud picture data needed for being downloaded in Chinese Resources satellite hub；
2) collector is utilized to acquire the spissatus of 39*39 pixels on HJ-1A/1B satellite cloud pictures respectively, Bao Yun and clear sky cloud mass are each
2000 pieces, uniformly it is scaled 32*32 pixels；
The step (2) specifically includes：
1) it is that 32*32 pixel forms and training sample are defeated together as convolutional neural networks by whole picture satellite cloud picture format conversion
Enter source；
2) convolutional neural networks include 7 layers, and the 1st layer is input layer, and the 2nd layer is volume base, is extracted by the convolution kernel of multiple 5*5
The different characteristic of picture, includes the Feature Mapping figure of 12 28*28, and the 3rd layer is down-sampling layer, be by 12 sizes is 14*14
Characteristic pattern composition, each neuron of characteristic pattern is connected with the 1st layer of the fields 2*2, the 4th layer of characteristic pattern group by 16 10*10
At convolutional layer, the 5th layer is down-sampling layer as the 3rd layer, includes the characteristic pattern of 16 5*5, and the 6th layer is full articulamentum, altogether
It is output layer to have 400 tie points, last layer, has 3 nodes, respectively represents spissatus, Bao Yun and clear sky；
3) satellite cloud picture after detection is showed with redgreenblue and gray-scale map respectively；
Cloud amount computational methods described in step (3) are：
The basic principle of " Spatial coherence method " be based on to single pixel amount of radiation and clear sky and it is spissatus in the case of amount of radiation
Detection, obtains the total amount of cloud of single pixel, calculation formula is as follows：
I=(1-Ac)Iclr+AcIcld
In formula, I is the amount of radiation that pixel receives；AcIndicate pixel total amount of cloud；Iclr、IcldThe highest ash of clear sky/blue is indicated respectively
Angle value, spissatus/red lowest gray value amount of radiation；Total amount of cloud calculation formula is：
I is the amount of radiation that pixel receives；AcIndicate pixel total amount of cloud；Iclr、IcldRespectively indicate clear sky/blue highest gray value,
Spissatus/red lowest gray value amount of radiation.
2. it is according to claim 1 it is a kind of based on convolutional neural networks be satellite cloudiness computational methods, which is characterized in that
The satellite cloud picture training sample for including 6000 training samples is established in step (1).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410798632.7A CN104504389B (en) | 2014-12-18 | 2014-12-18 | A kind of satellite cloudiness computational methods based on convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410798632.7A CN104504389B (en) | 2014-12-18 | 2014-12-18 | A kind of satellite cloudiness computational methods based on convolutional neural networks |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104504389A CN104504389A (en) | 2015-04-08 |
| CN104504389B true CN104504389B (en) | 2018-11-06 |
Family
ID=52945785
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410798632.7A Active CN104504389B (en) | 2014-12-18 | 2014-12-18 | A kind of satellite cloudiness computational methods based on convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104504389B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107247955A (en) * | 2016-07-29 | 2017-10-13 | 竹间智能科技（上海）有限公司 | Accessory recognition methods and device |
| CN106776842B (en) * | 2016-11-28 | 2021-03-19 | 腾讯科技（上海）有限公司 | Multimedia data detection method and device |
| CN107529647B (en) * | 2016-12-02 | 2021-02-09 | 南京信息工程大学 | Cloud picture cloud amount calculation method based on multilayer unsupervised sparse learning network |
| CN106920231A (en) * | 2017-02-17 | 2017-07-04 | 北京理工大学 | A kind of remote sensing image clouds appraisal procedure based on full-colour image statistical nature |
| CN107730469A (en) * | 2017-10-17 | 2018-02-23 | 长沙全度影像科技有限公司 | A kind of three unzoned lens image recovery methods based on convolutional neural networks CNN |
| CN107977758A (en) * | 2018-01-04 | 2018-05-01 | 郑州云海信息技术有限公司 | A kind of rainfall intensity forecasting procedure and relevant apparatus |
| CN108846474A (en) * | 2018-05-18 | 2018-11-20 | 南京信息工程大学 | The satellite cloud picture cloud amount calculation method of convolutional neural networks is intensively connected based on multidimensional |
| CN108765035A (en) * | 2018-06-19 | 2018-11-06 | 北京奇艺世纪科技有限公司 | A kind of advertising image feature extracting method, device and electronic equipment |
| CN110533063A (en) * | 2019-07-17 | 2019-12-03 | 赛德雷特(珠海)航天科技有限公司 | A kind of cloud amount calculation method and device based on satellite image and GMDH neural network |
| CN112348058B (en) * | 2020-10-20 | 2022-10-11 | 华东交通大学 | Satellite cloud picture classification method based on CNN-LSTM network and computer readable storage medium |
| CN112926789B (en) * | 2021-03-17 | 2024-05-14 | 阳光慧碳科技有限公司 | Satellite cloud image prediction method, prediction device and readable storage medium |
| CN113361414B (en) * | 2021-06-08 | 2022-09-02 | 天津大学 | Remote sensing image cloud amount calculation method based on composite neural network |
| CN113435539A (en) * | 2021-07-21 | 2021-09-24 | 天津大学 | Cloud amount grade classification method for Antarctic night sky cloud picture based on deep learning |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002156466A (en) * | 2000-11-16 | 2002-05-31 | Mitsubishi Electric Corp | System and method for calculation of total cloud amount |
| JP3760443B2 (en) * | 2003-04-07 | 2006-03-29 | 気象庁長官 | Cloud cover judgment device |
| CN103235954A (en) * | 2013-04-23 | 2013-08-07 | 南京信息工程大学 | Improved AdaBoost algorithm-based foundation cloud picture identification method |
| CN103472501B (en) * | 2013-09-06 | 2016-05-25 | 中国气象科学研究院 | Cloud detection and all-sky total amount of cloud detection method and system |
-
2014
- 2014-12-18 CN CN201410798632.7A patent/CN104504389B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104504389A (en) | 2015-04-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104504389B (en) | A kind of satellite cloudiness computational methods based on convolutional neural networks | |
| CN108596108B (en) | Aerial remote sensing image change detection method based on triple semantic relation learning | |
| CN108334847A (en) | A kind of face identification method based on deep learning under real scene | |
| CN107578390A (en) | A kind of method and device that image white balance correction is carried out using neutral net | |
| CN106462771A (en) | 3D image significance detection method | |
| CN110210608A (en) | The enhancement method of low-illumination image merged based on attention mechanism and multi-level features | |
| CN104850836A (en) | Automatic insect image identification method based on depth convolutional neural network | |
| CN107122701A (en) | A kind of traffic route sign based on saliency and deep learning | |
| CN107038416A (en) | A kind of pedestrian detection method based on bianry image modified HOG features | |
| CN107220643A (en) | The Traffic Sign Recognition System of deep learning model based on neurological network | |
| CN109376676A (en) | Highway engineering site operation personnel safety method for early warning based on unmanned aerial vehicle platform | |
| CN109557030A (en) | A kind of water quality element inversion method based on unmanned aerial vehicle remote sensing | |
| CN109377480A (en) | Arable land use change detection method based on deep learning | |
| CN112819096B (en) | Construction method of fossil image classification model based on composite convolutional neural network | |
| CN106372630A (en) | Face direction detection method based on deep learning | |
| CN114943893B (en) | Feature enhancement method for land coverage classification | |
| CN109753996A (en) | Hyperspectral image classification method based on D light quantisation depth network | |
| CN114973011A (en) | High-resolution remote sensing image building extraction method based on deep learning | |
| CN109754440A (en) | A kind of shadow region detection method based on full convolutional network and average drifting | |
| CN105405138A (en) | Water surface target tracking method based on saliency detection | |
| CN109711256A (en) | A kind of low latitude complex background unmanned plane target detection method | |
| CN108460794A (en) | A kind of infrared well-marked target detection method of binocular solid and system | |
| CN112418049A (en) | Water body change detection method based on high-resolution remote sensing image | |
| CN103295238B (en) | Video real-time location method based on ROI motion detection on Android platform | |
| CN116206214A (en) | Automatic landslide recognition method, system, equipment and medium based on lightweight convolutional neural network and double attention |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230410 Address after: Room 125, 1st Floor, Building 20, No. 1188, Huyi Road, Jiading District, Shanghai, 201802 Patentee after: Write easy network technology (Shanghai) Co.,Ltd. Address before: 210044 No. 219, Ning six road, Nanjing, Jiangsu Patentee before: Nanjing University of Information Science and Technology |
|
| TR01 | Transfer of patent right | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20150408 Assignee: Shanghai Suiyu Enterprise Management Consulting Partnership (L.P.) Assignor: Write easy network technology (Shanghai) Co.,Ltd. Contract record no.: X2023980042559 Denomination of invention: A Method for Computing Satellite Cloud Amount Based on Convolutional Neural Networks Granted publication date: 20181106 License type: Common License Record date: 20230923 |
|
| EE01 | Entry into force of recordation of patent licensing contract |