CN116992962A - Landform knowledge graph construction method based on self-supervision deep learning - Google Patents
Landform knowledge graph construction method based on self-supervision deep learning Download PDFInfo
- Publication number
- CN116992962A CN116992962A CN202310995603.9A CN202310995603A CN116992962A CN 116992962 A CN116992962 A CN 116992962A CN 202310995603 A CN202310995603 A CN 202310995603A CN 116992962 A CN116992962 A CN 116992962A
- Authority
- CN
- China
- Prior art keywords
- model
- self
- landform
- training
- supervision
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000010276 construction Methods 0.000 title claims abstract description 28
- 238000013135 deep learning Methods 0.000 title claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 59
- 238000004458 analytical method Methods 0.000 claims abstract description 18
- 238000011156 evaluation Methods 0.000 claims abstract description 6
- 238000000034 method Methods 0.000 claims description 17
- 239000013598 vector Substances 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000013136 deep learning model Methods 0.000 claims description 7
- 238000013461 design Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 6
- 230000000873 masking effect Effects 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000012512 characterization method Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 claims description 3
- 230000007786 learning performance Effects 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 230000005012 migration Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000000737 periodic effect Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 238000011160 research Methods 0.000 abstract description 7
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000009833 condensation Methods 0.000 description 1
- 230000005494 condensation Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
-
- 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
- G06N3/0455—Auto-encoder networks; Encoder-decoder 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
- G06N3/0895—Weakly supervised learning, e.g. semi-supervised or self-supervised learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Image Analysis (AREA)
Abstract
The application relates to the technical field of landform knowledge graph research, in particular to a landform knowledge graph construction method based on self-supervision deep learning, which comprises the steps of self-supervision pre-training model construction, evaluation of the self-supervision pre-training model, semantic analysis and knowledge graph construction.
Description
Technical Field
The application relates to the technical field of research of landform knowledge maps, in particular to a landform knowledge map construction method based on self-supervision deep learning.
Background
Landforms are one of the most important natural geographic elements, affecting the surface climate, ecological environment and the spatial distribution of natural resources. The geomorphic research is an important branch of geography, plays an important role in solving the morphology, structure, distribution change and rule of the earth surface, and provides scientific basis for human survival and development. With the development of science and technology, the research of topography is increasingly informationized and intelligent. At the end of nineties of the last century, along with the proposal of the concept of the information map of the geomorphology, students in the field of geomorphology break through key technologies such as geomorphology information extraction, classification and the like, construct the information map of morphological characteristics, geomorphology development and the like, and realize the summarization and the condensation of a series of geomorphology knowledge.
The current knowledge graph (KnowledgeGraph) becomes a research hotspot in the information field. Knowledge patterns represent knowledge in the form of a structured directed graph, represent entities and relationships between the entities in nodes and edges, and map semantics of the entities and the relationships to a low-dimensional continuous vector space through knowledge representation learning (KnowledgeGraph Embedding), so that knowledge reasoning and prediction can be realized through vector or matrix operation, and the knowledge patterns have strong semantic computing capability, are widely applied in aspects of semantic retrieval, intelligent question-answering and individual recommendation, and are also constructed in a plurality of fields of field knowledge patterns in field study.
In particular to the field of geomorphology, because of the problems of fuzzy boundary, difficult type determination and multi-scale of the landform, the landform classification is difficult, the landform classification systems are various, the different systems are difficult to unify, and the difficulty exists in constructing a landform knowledge body. The boundary of the landform is fuzzy and difficult to determine, and the current landform type division takes human cognition as a core, is not a system which faces machine calculation and takes machine understanding as a core, and is difficult to realize quantitative expression and calculation which can be understood and utilized by a computer.
Expert knowledge is needed in domestic and foreign researches in the prior art, a body layer is built from top to bottom, the automation degree is not high, the existing landform classification system is various, a unified knowledge body is difficult to build, and sufficient various landform type entities and characteristics thereof are difficult to provide according to triples extracted from texts by the body.
Therefore, in order to solve the problems, the application provides a landform knowledge graph construction method based on self-supervision deep learning, which is developed rapidly through the self-supervision deep learning in the field of deep learning. The pre-training model based on the self-supervision learning strategy trains on a large amount of data, can automatically extract general features in the data under the condition of no domain expert knowledge, forms a self-supervision deep learning model and a training strategy suitable for DEM data, achieves the aim of fully learning geomorphic semantics and spatial features, explores a geomorphic classification system under an artificial intelligence view angle, and realizes automatic construction of a computer-understandable and computable geomorphic knowledge map from bottom to top.
Disclosure of Invention
The application aims to fill the blank of the prior art, and provides a landform knowledge graph construction method based on self-supervision deep learning, which is rapid in development through the self-supervision deep learning in the field of deep learning. The pre-training model based on the self-supervision learning strategy trains on a large amount of data, can automatically extract general features in the data under the condition of no domain expert knowledge, forms a self-supervision deep learning model and a training strategy suitable for DEM data, achieves the aim of fully learning geomorphic semantics and spatial features, explores a geomorphic classification system under an artificial intelligence view angle, and realizes automatic construction of a computer-understandable and computable geomorphic knowledge map from bottom to top.
In order to achieve the above purpose, the application provides a landform knowledge graph construction method based on self-supervision deep learning, which comprises the steps of self-supervision pre-training model construction, evaluation of the self-supervision pre-training model, semantic analysis and knowledge graph construction;
self-supervision pre-training model construction: the method comprises the steps of perfecting a self-supervision learning strategy, constructing a training data set, designing a model structure, exploring a learning strategy and a design loss function, and performing model training;
evaluation of self-supervised pre-training model: applying the pre-training model to a downstream task, evaluating learning performance and migration performance of the model, finding problems, feeding back the problems to the pre-training model, and adjusting model parameters;
semantic analysis and knowledge graph construction: carrying out semantic analysis on the feature vector characterization obtained by pre-training to construct a feature knowledge graph;
the self-supervision pre-training model construction specifically comprises the following steps:
s1, pre-training data set:
constructing a data source of a large-scale pre-training data set by utilizing the existing global DEM products with different resolutions, randomly selecting partial areas to construct the pre-training data set, cutting data into raster pattern images with uniform sizes, constructing the pre-training data set with the total data size of 900,000 ～ 1,000,000, dividing smaller-scale data into verification sets, and taking the rest as the training set;
s2, model design:
constructing a self-supervision DEM deep learning model, designing the structures of an encoder and a decoder, selecting a mask structure of an MAE model and a basic framework based on a ViT model, comparing the performances of different ViT model structures with the performances of other convolutional neural network models, and selecting an optimal model through a plurality of groups of comparison experiments to solve the coding key technology;
changing the data input mode of the encoder, and synchronously learning the characteristics of different resolutions by adopting a method for synchronously inputting multi-resolution data in the same area; because pixels of data with different resolutions of the same picture are inconsistent, changing a pixel position coding mode of MAE, referring to an absolute distance coding mode of Scale-MAE, and designing a proper position coding mode aiming at the relativity of terrains among different pictures, so that the position information of the data with different resolutions is consistent, the problem of information leakage of absolute position coding is avoided, the periodic mode of the terrains is reflected to a certain extent, and the universality of the model to the data with different scales is improved by adjusting the decoder structure;
s3, self-supervision learning strategy:
the method comprises the steps of perfecting a self-supervision learning strategy, exploring the setting of different super parameters, randomly masking a DEM (digital elevation model) picture in a certain proportion in a mode of image masking, taking the uncovered part as the input of a pre-training model, obtaining a restored image after passing through an encoder and a decoder, calculating loss compared with an original image, and optimizing the model; different data covering modes are tried or data are covered by random shapes generated randomly, different covering proportions are tried, and the optimal proportion is explored;
s4, loss function:
the method comprises the steps of designing a reasonable loss function, introducing a training effect of a terrain factor control model to the most direct index which is the numerical difference between a recovered numerical value and a real numerical value in a self-supervision learning process, and adding the loss values of the factors into the loss function to enable the model to consider the terrain factor in optimization;
the semantic analysis and knowledge graph construction specifically comprises the following steps:
s10, the supervised learning model represents the landform features by vectors, and the landforms with similar features and adjacent spaces are combined into a landform unit by similarity calculation of the vectors, wherein one landform unit is a landform type entity;
s20, obtaining a hierarchical structure of the landform type through hierarchical clustering;
s30, obtaining the attention score of each two plots through an attention mechanism of self-supervision learning, constructing an attention score matrix, calculating the interdependence degree of the plots, adopting visual analysis to explore the interdependence relation among different landform entities, researching a landform type distribution space mode, and constructing a landform space grammar tree;
s40, forming a landform knowledge graph comprising a landform type hierarchical structure, landform entity semantic representation and a landform type spatial relationship.
Compared with the prior art, the application learns the multi-scale landform type semantics and spatial features on the global multi-resolution DEM data to form the landform feature vector representation aiming at machine calculation.
Through exploration, a self-supervision deep learning model and a training strategy suitable for DEM data are formed, and the purpose of fully learning geomorphic semantics and spatial features is achieved; through analysis and calculation of landform feature vector representation, a landform classification system under an artificial intelligence view angle is explored, and a computer understandable and computable landform knowledge map is automatically built from bottom to top.
Drawings
FIG. 1 is a schematic diagram of the general research framework and technical route of the project of the application.
FIG. 2 is a schematic diagram of the MAE model of the present application.
FIG. 3 is a schematic diagram of a knowledge graph construction technology route of the present application.
Detailed Description
The application will now be further described with reference to the accompanying drawings.
Referring to fig. 1 to 3, a method for constructing a geomorphic knowledge graph based on self-supervision deep learning includes constructing a self-supervision pre-training model, evaluating the self-supervision pre-training model, and constructing a semantic analysis and a knowledge graph;
self-supervision pre-training model construction: the method comprises the steps of perfecting a self-supervision learning strategy, constructing a training data set, designing a model structure, exploring a learning strategy and a design loss function, and performing model training;
evaluation of self-supervised pre-training model: applying the pre-training model to a downstream task, evaluating learning performance and migration performance of the model, finding problems, feeding back the problems to the pre-training model, and adjusting model parameters;
semantic analysis and knowledge graph construction: carrying out semantic analysis on the feature vector characterization obtained by pre-training to construct a feature knowledge graph;
the self-supervision pre-training model construction specifically comprises the following steps:
s1, pre-training data set:
constructing a data source of a large-scale pre-training data set by utilizing the existing global DEM products with different resolutions, randomly selecting partial areas to construct the pre-training data set, cutting data into raster pattern images with uniform sizes, constructing the pre-training data set with the total data size of 900,000 ～ 1,000,000, dividing smaller-scale data into verification sets, and taking the rest as the training set;
s2, model design:
constructing a self-supervision DEM deep learning model, designing the structures of an encoder and a decoder, selecting a mask structure of an MAE model and a basic framework based on a ViT model, comparing the performances of different ViT model structures with the performances of other convolutional neural network models, and selecting an optimal model through a plurality of groups of comparison experiments to solve the coding key technology;
changing the data input mode of the encoder, and synchronously learning the characteristics of different resolutions by adopting a method for synchronously inputting multi-resolution data in the same area; because pixels of data with different resolutions of the same picture are inconsistent, changing a pixel position coding mode of MAE, referring to an absolute distance coding mode of Scale-MAE, and designing a proper position coding mode aiming at the relativity of terrains among different pictures, so that the position information of the data with different resolutions is consistent, the problem of information leakage of absolute position coding is avoided, the periodic mode of the terrains is reflected to a certain extent, and the universality of the model to the data with different scales is improved by adjusting the decoder structure;
s3, self-supervision learning strategy:
the method comprises the steps of perfecting a self-supervision learning strategy, exploring the setting of different super parameters, randomly masking a DEM (digital elevation model) picture in a certain proportion in a mode of image masking, taking the uncovered part as the input of a pre-training model, obtaining a restored image after passing through an encoder and a decoder, calculating loss compared with an original image, and optimizing the model; different data covering modes are tried or data are covered by random shapes generated randomly, different covering proportions are tried, and the optimal proportion is explored;
s4, loss function:
the method comprises the steps of designing a reasonable loss function, introducing a training effect of a terrain factor control model to the most direct index which is the numerical difference between a recovered numerical value and a real numerical value in a self-supervision learning process, and adding the loss values of the factors into the loss function to enable the model to consider the terrain factor in optimization;
the semantic analysis and knowledge graph construction specifically comprises the following steps:
s10, the supervised learning model represents the landform features by vectors, and the landforms with similar features and adjacent spaces are combined into a landform unit by similarity calculation of the vectors, wherein one landform unit is a landform type entity;
s20, obtaining a hierarchical structure of the landform type through hierarchical clustering;
s30, obtaining the attention score of each two plots through an attention mechanism of self-supervision learning, constructing an attention score matrix, calculating the interdependence degree of the plots, adopting visual analysis to explore the interdependence relation among different landform entities, researching a landform type distribution space mode, and constructing a landform space grammar tree;
s40, forming a landform knowledge graph comprising a landform type hierarchical structure, landform entity semantic representation and a landform type spatial relationship.
The above is only a preferred embodiment of the present application, only for helping to understand the method and the core idea of the present application, and the scope of the present application is not limited to the above examples, and all technical solutions belonging to the concept of the present application belong to the scope of the present application. It should be noted that modifications and adaptations to the present application may occur to one skilled in the art without departing from the principles of the present application and are intended to be within the scope of the present application.
The application integrally solves the problems that the prior art has multiple landform classification systems, is difficult to unify among different systems, has high dependence degree and low automation degree, is difficult to construct a unified knowledge body and the like when the knowledge map is constructed in the field of landform science, and achieves the aim of fully learning landform semantics and spatial features by a self-supervision deep learning model and training strategy suitable for DEM data. Through analysis and calculation of landform feature vector representation, a landform classification system under an artificial intelligence view angle is explored, and a computer understandable and computable landform knowledge map is automatically built from bottom to top.
Claims (1)
1. The geomorphic knowledge graph construction method based on self-supervision deep learning is characterized by comprising the steps of self-supervision pre-training model construction, evaluation of the self-supervision pre-training model, semantic analysis and knowledge graph construction;
the self-supervision pre-training model is constructed: the method comprises the steps of perfecting a self-supervision learning strategy, constructing a training data set, designing a model structure, exploring a learning strategy and a design loss function, and performing model training; evaluation of the self-supervised pre-training model: applying the pre-training model to a downstream task, evaluating learning performance and migration performance of the model, finding problems, feeding back the problems to the pre-training model, and adjusting model parameters;
the semantic analysis and knowledge graph construction: carrying out semantic analysis on the feature vector characterization obtained by pre-training to construct a feature knowledge graph;
the self-supervision pre-training model construction specifically comprises the following steps:
s1, pre-training data set:
constructing a data source of a large-scale pre-training data set by utilizing the existing global DEM products with different resolutions, randomly selecting partial areas to construct the pre-training data set, cutting data into raster pattern images with uniform sizes, constructing the pre-training data set with the total data size of 900,000 ～ 1,000,000, dividing smaller-scale data into verification sets, and taking the rest as the training set;
s2, model design:
constructing a self-supervision DEM deep learning model, designing the structures of an encoder and a decoder, selecting a mask structure of an MAE model and a basic framework based on a ViT model, comparing the performances of different ViT model structures with the performances of other convolutional neural network models, and selecting an optimal model through a plurality of groups of comparison experiments to solve the coding key technology;
changing the data input mode of the encoder, and synchronously learning the characteristics of different resolutions by adopting a method for synchronously inputting multi-resolution data in the same area; because pixels of data with different resolutions of the same picture are inconsistent, changing a pixel position coding mode of MAE, referring to an absolute distance coding mode of Scale-MAE, and designing a proper position coding mode aiming at the relativity of terrains among different pictures, so that the position information of the data with different resolutions is consistent, the problem of information leakage of absolute position coding is avoided, the periodic mode of the terrains is reflected to a certain extent, and the universality of the model to the data with different scales is improved by adjusting the decoder structure; s3, self-supervision learning strategy:
the method comprises the steps of perfecting a self-supervision learning strategy, exploring the setting of different super parameters, randomly masking a DEM (digital elevation model) picture in a certain proportion in a mode of image masking, taking the uncovered part as the input of a pre-training model, obtaining a restored image after passing through an encoder and a decoder, calculating loss compared with an original image, and optimizing the model; different data covering modes are tried or data are covered by random shapes generated randomly, different covering proportions are tried, and the optimal proportion is explored; s4, loss function:
the method comprises the steps of designing a reasonable loss function, introducing a training effect of a terrain factor control model to the most direct index which is the numerical difference between a recovered numerical value and a real numerical value in a self-supervision learning process, and adding the loss values of the factors into the loss function to enable the model to consider the terrain factor in optimization; the semantic analysis and knowledge graph construction specifically comprises the following steps:
s10, the supervised learning model represents the landform features by vectors, and the landforms with similar features and adjacent spaces are combined into a landform unit by similarity calculation of the vectors, wherein one landform unit is a landform type entity;
s20, obtaining a hierarchical structure of the landform type through hierarchical clustering;
s30, obtaining the attention score of each two plots through an attention mechanism of self-supervision learning, constructing an attention score matrix, calculating the interdependence degree of the plots, adopting visual analysis to explore the interdependence relation among different landform entities, researching a landform type distribution space mode, and constructing a landform space grammar tree;
s40, forming a landform knowledge graph comprising a landform type hierarchical structure, landform entity semantic representation and a landform type spatial relationship.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310995603.9A CN116992962B (en) | 2023-08-09 | 2023-08-09 | Landform knowledge graph construction method based on self-supervision deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310995603.9A CN116992962B (en) | 2023-08-09 | 2023-08-09 | Landform knowledge graph construction method based on self-supervision deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116992962A true CN116992962A (en) | 2023-11-03 |
| CN116992962B CN116992962B (en) | 2024-02-06 |
Family
ID=88529937
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310995603.9A Active CN116992962B (en) | 2023-08-09 | 2023-08-09 | Landform knowledge graph construction method based on self-supervision deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116992962B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117574844A (en) * | 2023-11-23 | 2024-02-20 | 华南理工大学 | Self-supervision learning DTCO process parameter performance specification feedback method |
| CN117656082A (en) * | 2024-01-29 | 2024-03-08 | 青岛创新奇智科技集团股份有限公司 | Industrial robot control method and device based on multi-mode large model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210192372A1 (en) * | 2019-12-19 | 2021-06-24 | Electronics And Telecommunications Research Institute | Multi-layered knowledge base system and processing method thereof |
| CN113590799A (en) * | 2021-08-16 | 2021-11-02 | 东南大学 | Weak supervision knowledge graph question-answering method based on multi-view reasoning |
-
2023
- 2023-08-09 CN CN202310995603.9A patent/CN116992962B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210192372A1 (en) * | 2019-12-19 | 2021-06-24 | Electronics And Telecommunications Research Institute | Multi-layered knowledge base system and processing method thereof |
| CN113590799A (en) * | 2021-08-16 | 2021-11-02 | 东南大学 | Weak supervision knowledge graph question-answering method based on multi-view reasoning |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117574844A (en) * | 2023-11-23 | 2024-02-20 | 华南理工大学 | Self-supervision learning DTCO process parameter performance specification feedback method |
| CN117656082A (en) * | 2024-01-29 | 2024-03-08 | 青岛创新奇智科技集团股份有限公司 | Industrial robot control method and device based on multi-mode large model |
| CN117656082B (en) * | 2024-01-29 | 2024-05-14 | 青岛创新奇智科技集团股份有限公司 | Industrial robot control method and device based on multi-mode large model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116992962B (en) | 2024-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116992962B (en) | Landform knowledge graph construction method based on self-supervision deep learning | |
| Zhu et al. | Adaptive deep sparse semantic modeling framework for high spatial resolution image scene classification | |
| Ozturk et al. | Comparison of fully convolutional networks (FCN) and U-Net for road segmentation from high resolution imageries | |
| Gold | Problems with handling spatial data—the Voronoi approach | |
| Robertson et al. | An integrated environmental analytics system (IDEAS) based on a DGGS | |
| CN111259936B (en) | Image semantic segmentation method and system based on single pixel annotation | |
| CN106920243A (en) | The ceramic material part method for sequence image segmentation of improved full convolutional neural networks | |
| Lett et al. | Comparison of a cellular automata network and an individual-based model for the simulation of forest dynamics | |
| Mantovan et al. | The computerization of archaeology: Survey on artificial intelligence techniques | |
| Rodríguez-Ortega | Image processing and computer vision in the field of art history | |
| Cui et al. | WetlandNet: Semantic segmentation for remote sensing images of coastal wetlands via improved UNet with deconvolution | |
| Wan et al. | [Retracted] Research on Urban Renewal Public Space Design Based on Convolutional Neural Network Model | |
| Zhang et al. | A machine learning paradigm for Studying Pictorial realism: are constable's clouds more real than his contemporaries? | |
| Guo et al. | Semantic image segmentation based on SegNetWithCRFs | |
| Zhou et al. | Microlandform classification method for grid DEMs based on support vector machine | |
| Li et al. | A new algorithm of vehicle license plate location based on convolutional neural network | |
| Pourmohammadi et al. | Predicting developed land expansion using deep convolutional neural networks | |
| Li | Application of computer image technology in 3D painting based on cloud computing | |
| Ai | Some thoughts on deep learning empowering cartography | |
| Liu | Multidimensional nonlinear landscape planning based on parameter feature extraction and multimedia technology | |
| Eremeev et al. | Algorithms for topological analysis of spatial data | |
| Liu et al. | A lightweight convolutional neural network based on dense connection for open-pit coal mine service identification using the edge-cloud architecture | |
| Tenzer et al. | The Geospatial Generalization Problem: When Mobility Isn't Mobile | |
| Surya Prakash et al. | Multi-class Segmentation of Trash in Coastal Areas Using Encoder-Decoder Architecture | |
| Patil et al. | Road Network Extraction Using Multi-path Cascade Convolution Neural Network from Remote Sensing Images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |