CN112949645A - Deep learning-based field plot segmentation algorithm - Google Patents
Deep learning-based field plot segmentation algorithm Download PDFInfo
- Publication number
- CN112949645A CN112949645A CN202110214853.5A CN202110214853A CN112949645A CN 112949645 A CN112949645 A CN 112949645A CN 202110214853 A CN202110214853 A CN 202110214853A CN 112949645 A CN112949645 A CN 112949645A
- Authority
- CN
- China
- Prior art keywords
- land
- deep learning
- remote sensing
- data
- segmentation
- 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.)
- Withdrawn
Links
- 230000011218 segmentation Effects 0.000 title claims abstract description 54
- 238000013135 deep learning Methods 0.000 title claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 31
- 238000004364 calculation method Methods 0.000 claims abstract description 23
- 238000000034 method Methods 0.000 claims abstract description 21
- 238000003709 image segmentation Methods 0.000 claims abstract description 17
- 238000012360 testing method Methods 0.000 claims abstract description 15
- 238000004458 analytical method Methods 0.000 claims abstract description 14
- 238000000638 solvent extraction Methods 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims description 14
- 238000005070 sampling Methods 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 8
- 238000011176 pooling Methods 0.000 claims description 4
- 238000010205 computational analysis Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 description 6
- 238000007726 management method Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 230000001788 irregular Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/182—Network patterns, e.g. roads or rivers
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a field plot partitioning algorithm based on deep learning, which adopts the technical scheme that S1 data is acquired, S2 data is preprocessed, S3 training samples are manufactured, an S4 network model is generated, S5 primary calculation analysis is performed, S6 plot partitioning model training test is performed, and S7 secondary calculation analysis is performed. According to the method, a large amount of satellite remote sensing data is collected, a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application is called based on a Pythroch deep learning framework to carry out model training on the satellite remote sensing data, a generated network model is applied to land block segmentation, and effective parameters such as vector boundaries, areas and the like of all land blocks are accurately extracted on the basis, so that the use condition of each land block can be more accurately monitored.
Description
Technical Field
The invention relates to the technical field of land planning, in particular to a field plot segmentation algorithm based on deep learning.
Background
China is a traditional agricultural big country, is currently in an urbanization acceleration process, and in the evolution process of the traditional agriculture to the modern agriculture, how to reasonably and efficiently plan and utilize the land to put forward higher requirements on the capability of human beings, aiming at the utilization management of the land, the traditional method is generally used for manual investigation in person, and has the defects of high time and labor consumption cost, the technology of monitoring, managing and planning the land through satellite images is initiated along with the development of the satellite remote sensing technology, and meanwhile, the development of the artificial intelligence technology plays a role in automatically and efficiently processing the satellite images, so that the image segmentation technology in the satellite image and AI technology is combined and applied to the land monitoring management and planning to become a field land block segmentation algorithm based on deep learning.
However, for the satellite remote sensing image, due to the complicated regional cross characteristic and irregular connection, the segmentation difficulty is extremely high, the traditional satellite remote sensing image segmentation method is easily interfered by segmentation threshold values, local similar but heterogeneous information, so that the problems of low segmentation accuracy rate, multiple divisions, redistribution and the like are caused, and the traditional satellite remote sensing image segmentation method is particularly sensitive to the ground clearance and the shooting angle of a satellite.
Therefore, it is necessary to invent a field plot segmentation algorithm based on deep learning.
Disclosure of Invention
Therefore, the invention provides a field plot segmentation algorithm based on deep learning, which is characterized in that a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application is called to perform model training on satellite remote sensing data by acquiring massive satellite remote sensing data based on a Pythroch deep learning framework, a generated network model is applied to plot segmentation, and effective parameters such as vector boundaries, areas and the like of each plot are accurately extracted on the basis, so that the problems that the segmentation accuracy is not high or more than scores and re-scores and the like due to the fact that a traditional satellite remote sensing image segmentation method is easily interfered by segmentation threshold values, local similarity but different types of information are solved, and meanwhile, the traditional satellite remote sensing image segmentation method is particularly sensitive to the satellite ground clearance and the shooting angle.
In order to achieve the above purpose, the invention provides the following technical scheme: a field plot segmentation algorithm based on deep learning comprises the following specific steps:
s1 data acquisition: obtaining a remote sensing image by using a satellite map loader;
s2 data preprocessing: acquiring a large number of land satellite remote sensing data images of a plurality of counties and cities in China in a dotting mode, and carrying out size cutting pretreatment according to a certain resolution;
s3 preparing training samples: marking the land area and the non-land area by using a marking tool to form a label file, and thus, manufacturing a training data sample set;
s4 network model: selecting a U-net semantic segmentation network for the field land parcel segmentation network model, wherein the U-net comprises a first part of feature extraction part and a second part of up-sampling part;
s5 one-time calculation analysis: calling a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application based on a Pythrch deep learning framework to perform computational analysis on satellite remote sensing data;
s6 training test of land segmentation model: training, testing and learning the training data sample set in the S3 by using a U-net network to obtain a field plot segmentation model;
s7 secondary calculation analysis: and applying the model field land block segmentation model in the S6 to an actual image test to segment the field land blocks in the satellite images, and extracting effective specific parameters of each land block on the basis.
Preferably, in the step S1, the loader acquires the remote sensing image with the resolution of 2m, and the satellite map used is not limited to the google satellite map, and may also be a high-resolution or Baidu satellite map.
Preferably, the collected land satellite remote sensing data images in the data preprocessing of step S2 are data images of counties and cities at different distances, different shooting angles and different time periods.
Preferably, in the step S2, the data pre-processing includes cutting the data image according to a size of 1024 × 1024 by the client.
Preferably, in the step S3, the plot area and the non-plot area in the training sample are labeled with different color values.
Preferably, in the step S4, the feature extraction part in the network model: mainly a convolution layer and a pooling layer; an up-sampling part: mainly an deconvolution layer and a deconvolution layer.
Preferably, in the step S5, in one calculation analysis, the client calculates that each feature extraction part is located between the corresponding upsampling layer and the corresponding upsampling layer, the feature map obtained by each convolution layer is located between the corresponding upsampling layer and the corresponding vertex layer, and meanwhile, the feature map of each layer is effectively used in the subsequent calculation, and a corresponding data set is manufactured.
Preferably, the specific parameters of the parcel in the secondary calculation analysis in step S7 include effective parameters such as vector boundary, area, etc.
The invention has the beneficial effects that:
according to the method, a large amount of satellite remote sensing data is collected, a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application is called based on a Pythrch deep learning framework to perform model training on the satellite remote sensing data, a generated network model is applied to land block segmentation, effective parameters such as vector boundaries, areas and the like of all land blocks are accurately extracted on the basis, the using condition of each land block is monitored more accurately, and a large number of data tests prove that the method can be accurately and effectively applied to field land block management and planning Unreasonable and the like, and greatly improves the utilization rate of the land.
Drawings
FIG. 1 is a diagram of a U-net network model architecture provided by the present invention;
FIG. 2 is an original satellite image map provided by the present invention;
FIG. 3 is a field plot segmentation result provided by the present invention;
fig. 4 is a vector boundary diagram of each parcel provided by the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
Embodiment 1, referring to fig. 1, the field plot partitioning algorithm based on deep learning provided by the present invention includes the following specific steps:
s1 data acquisition: obtaining a remote sensing image by using a satellite map loader;
s2 data preprocessing: acquiring a large number of land satellite remote sensing data images of a plurality of counties and cities in China in a dotting mode, and carrying out size cutting pretreatment according to a certain resolution;
s3 preparing training samples: marking the land area and the non-land area by using a marking tool to form a label file, and thus, manufacturing a training data sample set;
s4 network model: selecting a U-net semantic segmentation network for the field land parcel segmentation network model, wherein the U-net comprises a first part of feature extraction part and a second part of up-sampling part,
s5 one-time calculation analysis: calling a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application based on a Pythrch deep learning framework to perform computational analysis on satellite remote sensing data;
s6 training test of land segmentation model: training, testing and learning the training data sample set in the S3 by using a U-net network to obtain a field plot segmentation model;
s7 secondary calculation analysis: and applying the model field land block segmentation model in the S6 to an actual image test to segment the field land blocks in the satellite images, and extracting effective specific parameters of each land block on the basis.
According to the method, a large amount of satellite remote sensing data is collected, a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application is called based on a Pythrch deep learning framework to perform model training on the satellite remote sensing data, a generated network model is applied to land block segmentation, effective parameters such as vector boundaries, areas and the like of all land blocks are accurately extracted on the basis, the using condition of each land block is monitored more accurately, and a large number of data tests prove that the method can be accurately and effectively applied to field land block management and planning Unreasonable and the like, and greatly improves the utilization rate of the land.
Further, in the step S1, the loader acquires the remote sensing image with the resolution of 2m, and the adopted satellite map is not limited to the google satellite map, and can also use the high-grade satellite map, the Baidu satellite map and the like;
furthermore, the collected land satellite remote sensing data images in the step S2 are data images of counties and cities at different distances, different shooting angles and different time periods, and compared with the problems that the traditional image segmentation accuracy is not high, and the traditional image segmentation method is particularly sensitive to the satellite ground clearance and the shooting angles, when the data images are collected, the complicated regional cross characteristics and the irregular connection possibility of the counties and cities at different shooting angles and different time periods are reduced to a certain extent, so that the segmentation difficulty is greatly simplified, and the time and labor are relatively saved;
further, in the data preprocessing of step S2, the data image is cut by the client according to the size of 1024 × 1024 of resolution, and the cut data image is summarized and integrated by the client, so as to complete the training data sample set;
further, the plot area and the non-plot area in the training sample manufactured in the step S3 are labeled by using different color values, and are labeled by using different color values, so that the data can be obtained more intuitively and clearly in observation and data acquisition, and the method is simple and rapid.
According to the method, 1000 pieces of land satellite remote sensing data of 300 counties and cities in China with different distances, different shooting angles and different time periods are collected in a dotting mode, cutting is carried out according to the size of 1024 × 1024 of each land satellite remote sensing data, then a labeling tool is used for labeling a land area and a non-land area by using different color values to form a label file, and therefore a training data sample set is manufactured.
Further, in the step S4, the feature extraction part in the network model: mainly a convolution layer and a pooling layer; an up-sampling part: mainly comprises an deconvolution layer and a deconvolution layer, and when the U-net semantic segmentation network operation is used, the feature graph obtained by each convolution layer is concateenated to the corresponding upsampling layer, so that the feature graph of each layer is effectively used in the subsequent calculation;
the U-net semantic segmentation network selected by the invention comprises two parts, wherein the first part is a feature extraction part, mainly comprises a convolution layer and a pooling layer, the second part is an up-sampling part, mainly comprises an anti-convolution layer and an anti-convolution layer, and a feature map obtained by each convolution layer is collocated to the corresponding up-sampling layer, so that the feature map of each layer is effectively used in subsequent calculation, therefore, compared with other network structures such as FCN, the U-net avoids directly monitoring and low calculation in a high-level feature map, but combines the features in a low-level feature map, so that the finally obtained feature map not only contains high-level features but also contains a plurality of low-level features, the fusion of the features under different scales is realized, the U-net is selected and used, and the calculation rate of the model can be improved more quickly and effectively, and the result accuracy of the model is also improved.
Further, in the step S5, in one calculation and analysis, the client calculates that each feature extraction part is located between the corresponding upper sampling layer and the corresponding feature map, the feature map obtained by each convolution layer is located between the corresponding upper sampling layer and the corresponding feature map, meanwhile, the feature map of each layer is effectively used for subsequent calculation, and a corresponding data set is made, the training data sample set is trained and learned by using the U-net network, so as to obtain a field block segmentation model, and the field block segmentation model is applied to an actual image test, so that the field block in the satellite image can be segmented accurately in real time;
further, the specific parameters of the plots in the secondary calculation analysis in the step S7 include effective parameters such as vector boundaries and areas, the effective parameters such as vector boundaries and areas of the plots are further extracted by using a traditional image processing technology-contour extraction on the field plot segmentation model, and according to the obtained effective parameters such as vector boundaries and areas of the plots, the land planning and monitoring tasks can be better and more effectively performed or executed by governments or farmers, and the land can be planned and utilized more reasonably and efficiently.
Example 2, a preferred embodiment of the present invention will be described below.
Referring to fig. 2 to 4, a satellite image of a certain area in Sixian county in Jiangsu province is acquired through a satellite map loader, the acquired satellite image is subjected to data preprocessing, a pre-trained field land parcel segmentation network model is used for reasoning calculation, a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application is called based on a Pythroch deep learning framework to perform model training on satellite remote sensing data, the generated network model is applied to land parcel segmentation, a land parcel segmentation result is obtained, and effective parameters such as vector boundaries, areas and the like of each land parcel are further extracted through a traditional image processing technology, namely contour extraction.
The visualization result shows that the method can be accurately and effectively applied to field plot management and planning. Meanwhile, the segmentation accuracy of the image is 93.57% by testing the segmentation result of the image in a farmland satellite in Jiangsu region, compared with the traditional satellite remote sensing image segmentation method, the accuracy is greatly improved, the inapplicability of the satellite remote sensing image segmentation effect of different distances and shooting angles is effectively solved by learning based on massive large data, the method is commercially applied to government or peasant household land planning and monitoring tasks at present, the method is popularized and applied to Mongolia, Liaoning, Anhui, Jiangsu and other places in China, the problems of irregular and unreasonable land use and the like can be effectively monitored, and the land utilization rate is greatly improved.
The using process of the invention is as follows: when the invention is used, a satellite map loader is used for obtaining a satellite image of an area to be used, partial land satellite remote sensing data of the area with different distances, different shooting angles and different time periods are collected in a dotting mode, cutting is carried out according to the size of each piece of resolution of 1024 x 1024, then a marking tool is used for marking a land area and a non-land area with different color values to form a label file, so that a data sample set of the area of a training part is manufactured, a client is used for extracting each feature extraction part of the data sample set to a corresponding upper sampling layer, a feature map obtained by each convolution layer is concatered to the corresponding upper sampling layer, meanwhile, the feature map of each layer is effectively used for subsequent calculation, a corresponding data set is manufactured, and the training and learning are carried out on the training data sample set by using a U-net network, the field plot segmentation model is obtained and applied to an actual satellite remote sensing image of an area to be measured for testing, all field plots of the local area in a satellite image can be segmented accurately in real time through a client, and effective parameters such as vector boundaries, areas and the like of each plot are further extracted by using a traditional image processing technology-contour extraction on the basis of applying the generated network model to plot segmentation, so that the use condition of each plot can be monitored more accurately, the problems of irregular and unreasonable land use and the like can be effectively monitored, and the utilization rate of the land is greatly improved.
The above description is only a preferred embodiment of the present invention, and any person skilled in the art may modify the present invention or modify it into an equivalent technical solution by using the technical solution described above. Therefore, any simple modifications or equivalent substitutions made in accordance with the technical solution of the present invention are within the scope of the claims of the present invention.
Claims (8)
1. A field plot segmentation algorithm based on deep learning is characterized in that: the method comprises the following specific steps:
s1 data acquisition: obtaining a remote sensing image by using a satellite map loader;
s2 data preprocessing: acquiring a large number of land satellite remote sensing data images of a plurality of counties and cities in China in a dotting mode, and carrying out size cutting pretreatment according to a certain resolution;
s3 preparing training samples: marking the land area and the non-land area by using a marking tool to form a label file, and thus, manufacturing a training data sample set;
s4 network model: selecting a U-net semantic segmentation network for the field land parcel segmentation network model, wherein the U-net comprises a first part of feature extraction part and a second part of up-sampling part;
s5 one-time calculation analysis: calling a U-shaped structure semantic segmentation network (U-net) suitable for remote sensing image segmentation application based on a Pythrch deep learning framework to perform computational analysis on satellite remote sensing data;
s6 training test of land segmentation model: training, testing and learning the training data sample set in the S3 by using a U-net network to obtain a field plot segmentation model;
s7 secondary calculation analysis: and applying the model field land block segmentation model in the S6 to an actual image test to segment the field land blocks in the satellite images, and extracting effective specific parameters of each land block on the basis.
2. The deep learning based field plot partitioning algorithm as claimed in claim 1, wherein: in the step S1, the loader acquires the remote sensing image with the resolution of 2m, and the adopted satellite map is not limited to the google satellite map, and can also use the high-grade satellite map, the Baidu satellite map and the like.
3. The deep learning based field plot partitioning algorithm as claimed in claim 1, wherein: and in the step S2, the collected land satellite remote sensing data images are data images of counties and cities at different distances, different shooting angles and different time periods.
4. The deep learning based field plot partitioning algorithm as claimed in claim 1, wherein: in the step S2, the data preprocessing is performed by cutting the data image according to the size of 1024 × 1024 resolution.
5. The deep learning based field plot partitioning algorithm as claimed in claim 1, wherein: the step S3 is to make the training sample in which the parcel region and the non-parcel region are labeled with different color values.
6. The deep learning based field plot partitioning algorithm as claimed in claim 1, wherein: the step S4 includes a feature extraction part in the network model: mainly a convolution layer and a pooling layer; an up-sampling part: mainly an deconvolution layer and a deconvolution layer.
7. The deep learning based field plot partitioning algorithm as claimed in claim 1, wherein: in the step S5, in one calculation and analysis, the client calculates that each feature extraction part is located between the corresponding upsampling layer and the corresponding upsampling layer, the feature map obtained by each convolutional layer is located between the corresponding upsampling layer and the corresponding upsampling layer, and meanwhile, the feature map of each layer is effectively used in the subsequent calculation, and a corresponding data set is manufactured.
8. The deep learning based field plot partitioning algorithm as claimed in claim 1, wherein: the specific parameters of the parcel in the secondary calculation analysis in the step S7 include effective parameters such as vector boundary, area, and the like.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110214853.5A CN112949645A (en) | 2021-02-25 | 2021-02-25 | Deep learning-based field plot segmentation algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110214853.5A CN112949645A (en) | 2021-02-25 | 2021-02-25 | Deep learning-based field plot segmentation algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112949645A true CN112949645A (en) | 2021-06-11 |
Family
ID=76246397
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110214853.5A Withdrawn CN112949645A (en) | 2021-02-25 | 2021-02-25 | Deep learning-based field plot segmentation algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112949645A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114677246A (en) * | 2022-03-17 | 2022-06-28 | 广州市城市规划勘测设计研究院 | School student recruitment unit dividing method, device, equipment and medium |
| CN116957883A (en) * | 2023-07-25 | 2023-10-27 | 南京智绘星图信息科技有限公司 | Construction land use control method based on data analysis |
-
2021
- 2021-02-25 CN CN202110214853.5A patent/CN112949645A/en not_active Withdrawn
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114677246A (en) * | 2022-03-17 | 2022-06-28 | 广州市城市规划勘测设计研究院 | School student recruitment unit dividing method, device, equipment and medium |
| CN116957883A (en) * | 2023-07-25 | 2023-10-27 | 南京智绘星图信息科技有限公司 | Construction land use control method based on data analysis |
| CN116957883B (en) * | 2023-07-25 | 2024-04-05 | 南京智绘星图信息科技有限公司 | Construction land use control method based on data analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11521379B1 (en) | Method for flood disaster monitoring and disaster analysis based on vision transformer | |
| CN107016405B (en) | A kind of pest image classification method based on classification prediction convolutional neural networks | |
| CN109493320B (en) | Remote sensing image road extraction method and system based on deep learning, storage medium and electronic equipment | |
| Yijing et al. | Fig fruit recognition method based on YOLO v4 deep learning | |
| CN111401419A (en) | Improved RetinaNet-based employee dressing specification detection method | |
| CN112949645A (en) | Deep learning-based field plot segmentation algorithm | |
| CN112749627A (en) | Method and device for dynamically monitoring tobacco based on multi-source remote sensing image | |
| CN113191374B (en) | PolSAR image ridge line extraction method based on pyramid attention network | |
| CN113392748B (en) | Remote sensing image cultivated land information extraction method based on convolutional neural network | |
| CN111104850B (en) | Remote sensing image building automatic extraction method and system based on residual error network | |
| CN109325451B (en) | High-spatial-resolution cultivated land block full-automatic extraction method based on deep learning | |
| CN113205039A (en) | Power equipment fault image identification and disaster investigation system and method based on multiple DCNNs | |
| CN114841961B (en) | Wheat scab detection method based on image enhancement and improved YOLOv5 | |
| CN112446350A (en) | Improved method for detecting cotton in YOLOv3 complex cotton field background | |
| CN116883718A (en) | Sugarcane seedling missing detection positioning method based on improved YOLOV7 | |
| CN112329697B (en) | Improved YOLOv 3-based on-tree fruit identification method | |
| CN116343111A (en) | Abandoned land monitoring method and system | |
| CN116310802A (en) | Method and device for monitoring change of residence based on multi-scale fusion model | |
| CN116416532A (en) | Disease strain diagnosis and strain shortage detection method, system, equipment and storage medium | |
| CN116206210A (en) | NAS-Swin-based remote sensing image agricultural greenhouse extraction method | |
| CN113780175B (en) | Remote sensing identification method for typhoon and storm landslide in high vegetation coverage area | |
| CN110084125B (en) | Agricultural insurance investigation technical method based on deep learning | |
| CN112580504B (en) | Tree species classification counting method and device based on high-resolution satellite remote sensing image | |
| CN115346212A (en) | Apple flower growth state detection method based on improved YOLOv5 | |
| CN115115954A (en) | Intelligent identification method for pine nematode disease area color-changing standing trees based on unmanned aerial vehicle remote sensing |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210611 |