Supermap gis 10i(2020) ai gis technology v1.0
•
0 likes•225 views
This document provides information about SuperMap Software Co., Ltd. It includes: 1. Background information on SuperMap such as its founding date and headquarters location. 2. Market share data showing SuperMap has the largest share of the GIS software market in China. 3. An overview of SuperMap's products and technologies including distributed GIS, cross-platform GIS, 3D GIS, big data GIS, and AI GIS.
Report
Share
Supermap gis 10i(2020) ai gis technology v1.0
- 1. P1 SuperMap Software Co., Ltd V10.1.0 SuperMap Research Institute Product Consulting
- 2. P2 nChinese Academy of Sciences nJune 18, 1997 nFounder: Dr. Ershun Zhong
- 3. P3
- 5. P5 SuperMap 31.6% ESRI 29.0% MapGIS 7.9% GeoStar 5.9% Skyline 5.3% GviTech 2.0% EV image 1.2% Others 17.1% Data from: CMIC 2016.09
- 6. P6
- 7. P7 Distributed GIS Cross Platform GIS T hree Dimension GIS B ig Data GIS A I GIS A I GIS (BitDC)
- 8. P8 Liberated Human from Repetitive Physical Labour Liberated Human from Repetitive Mental Labour Industrial Revolution Artificial Intelligence
- 9. P9 Function GeoAI GIS for AI AI for GIS Processing Tools Model Building Model Application Data Preparation Geospatial Machine Learning Geospatial Deep Learning Spatial Analysis for AI Spatial Visualization for AI AI Interaction AI Mapping AI Attribute Collection AI Measuring AI + AR Framework Spark MLlib TensorFlow Keras PyTorch Data File Relational data NoSQL Data Library Samples Models New
- 10. P10 File Relational Data NoSQL
- 11. P11 Domain Base Sample • Sample data • Sample characteristics organization and management Original Library • Store raw geospatial information data Model • Model file • Model parameters Data cleaning Model training
- 12. P12 • Distributed machine learning framework • Support linear regression, random forest regression and other machine learning operators • Machine learning analysis algorithms to support big data GIS • Deep learning framework supported by Google • Have a relatively complete ecosystem (mobile support, server support) • More suitable for AI product construction • A high-level neural network API • Written in Python and using TensorFlow, Theano, and CNTK as back ends • Suitable for rapid iterative development • Deep learning framework supported by Facebook • The model was evaluated better than TensorFlow in terms of ease of use and performance • There are also substantial model resources to support Technical System Analysis - Framework
- 13. P13 GeoAI 1 AI for GIS 2 GIS for AI 3
- 14. P14 1
- 15. P15 Deep Learning Geospatial Deep Learning Machine Learning Geospatial Machine Learning Statistics Spatial Statistics
- 16. P16 Study and analyze spatial data and problems based on statistical theory Because of the special properties of spatial data such as "spatial autocorrelation" and "spatial stratification heterogeneity" It is necessary to analyze the overall characteristics of space, spatial interpolation, spatial pattern and spatial regression
- 17. P17 Geospatial Distribution Characteristics • Spatial Autocorrelation • Spatial Stratified Heterogeneity Geospatial Pattern and Regression • Geospatial Point Pattern • General Model Geospatial Interpolation • Kernel Density • Inverse Distance Weight • Kriging Geographic Sampling • Spatial Random Sampling • Spatial Stratified Sampling • Sandwich Sampling • Spa • B-shade Geographical Distribution • Mean Center • Median Center • Centre Feature • Direction Distribution • Standard Distance • Linear Directional Mean • New
- 18. P18 Environmental Factor Identification: Affect Spatial Pattern of Disease Disease Incidence Watershed Zoning Soil Type DEM Watershed Zoning Dominates the Spatial Pattern of Disease Incidence
- 19. P19 A series of methods to analyze spatial data using machine learning models The spatial features of the task need to be embedded into the machine learning feature calculation process Many machine learning models are involved such as spatial clustering, classification and regression
- 20. P20 Regression • Geographic Simulation • Linear Regression • Decision Tree Regression • Geographically Weighted Regression • Generalized Linear Regression • Forest- Based Regression Clustering • Hot Spot Analysis • Spatial Density Clustering Classification • Map Matching • Logistic Regression • Gradient Boosting Classification • Decision Tree Classification • Naive Bayes Classification • Support Vector Machine Classification • Recognition Of Address Elements • Forest- Based Classification • New
- 21. P21 Explanatory Variables: • Transportation • Education • Medical Service
- 22. P22 Crime Rate Crime Type Crime Hotspot Hot Spot Analysis Support Vector Machine Classification Forest- Based Classification Crime Analysis Based on Spatial Machine Learning New
- 23. P23 • Based on machine learning (probability graph) model, calculate the reasonable matching of the trajectory points to be matched and restore the real trajectory Description • Trajectory Points Input Data • Hidden Markov Model Model New
- 24. P24 • Restore real trajectory based on trajectory points Hidden Markov Model More Advance Machine Learning Operators Total Length Matched Length Accuracy = Road Matching Based on Hidden Markov Model New
- 25. P25 A method to analyze spatial data based on deep learning model Through deep learning model, multidimensional fusion, correlation analysis and in-depth mining of complex spatiotemporal relations are carried out At present, deep learning models such as convolutional neural network (CNN) are widely used
- 26. P26 3D Data Analysis • Extracting Building Footprint from Oblique Photography DSM Data Image Analysis • Object Extraction • Object Detection • Scene Classification • Binary Classification • Land Use/Cover Classification Graph Analysis • Object Detection • Image Classification Spatiotemporal Prediction • Graph Time and Space Regression • New
- 27. P27 New
- 28. P28 New
- 29. P29 New Type 1 Type 2 Type 3 Using a large amount of historical electricity meter data Training AI model to identify the electricity meter type Deploying and using on mobile devices
- 30. P30 • Regression calculation for points • Convert spatiotemporal data into sequence signals on the graph Description • Point Input Data • Graph Neural Network (GNN) Model V: Nodes E: Edge V E New
- 31. P31 Prediction Value vs True Value True Value Prediction Value Prediction Accuracy
- 32. P32
- 33. P33 • Training neural network based on image and sample data • Used for instance segmentation of image data Description • Image Input Data • Mask R-CNN Model New
- 34. P34
- 35. P35
- 36. P36
- 37. P37
- 41. P41 Damaged Road Ads Messy Materials Trash Can
- 42. P42
- 43. P43
- 44. P44 Original map Color matching Extract the main color by K-Means Style picture Extract the main color by K-Means
- 45. P45
- 46. P46
- 48. P48
- 49. P49 New LOGO Fire facilities Bench Vehicle Road Lamp Flower …
- 52. P52 Vehicle Congestion (0-10) Motor Vehicle Non-motor Vehicle Motor Vehicle, Non-motor Vehicle, Pedestrian Congestion Top 5 Congested District Real Time Information District Congestion Congestion Type Video
- 53. P53 Real-Time AI Vehicle Recognizing Platform Statistics Illegal Vehicle Pedestrian Bicycle Vehicle Motorcycle Bus Truck Motor Vehicle Non-motor Vehicle Cameras
- 54. P54
- 55. P55 Function GeoAI GIS for AI AI for GIS Processing Tools Model Building Model Application Data Preparation Geospatial Machine Learning Geospatial Deep Learning Spatial Analysis for AI Spatial Visualization for AI AI Interaction AI Mapping AI Attribute Collection AI Measuring AI + AR Framework Spark MLlib TensorFlow Keras PyTorch Data File Relational data NoSQL Data Library Samples Models New
- 56. P56 SuperMap iObjects Python AI Framework SuperMap iObjects for Spark SuperMap iServer GIS Objects AI Framework GIS Terminal Cloud GIS Server SuperMap iManager Data Processing Analyst MachineLearning Service DataScience Service SuperMap iMobile/iTablet (Mobile Terminal) SuperMap iDesktopX (Desktop Terminal) TensorFlow Keras Spark MLlib PyTorch SuperMap WebApps (Web Terminal) SpatialStatistics ML.DataPrepartion ML.Training ML. Inference SuperMap iClient /iClient3D (Web Terminal) Streaming SuperMap iPortal Resource.Notebook Enhanced
- 57. P57 GIS Software Spatial Statistics Spatial Machine Learning Spatial Deep Learning Smart Attribute Collection Smart Mapping Video Target Detection and Tracking SuperMap iMobile / iTablet SuperMap iClient SuperMap iDesktopX SuperMap iServer SuperMap iObjects SuperMap iManager SuperMap iPortal
- 58. P58 5
- 59. P59 Product Parameters GPU Server Motherboard Configure According to the Situation GPU NVIDIA Tesla GPU T4 CPU 2*(intel E5-2630V4 2.2GHz 10 cores) ，intel® C612 RAM 16 *(Ecc4/Recc4 1.2v，2133~2666mhz ， Max128G), Total Max 2048GB， Standard: 128G RECC4 2666 Disk 8 * 3.5 Inch Hot-sway Bay，1*S4510 3.8t SSD SATA SATA2;SATA3; PCI-E SATA2; SATA3; PCI-ER0; R1;R10;R5，Raid Graphics Card Aspeed Ast2400 Bmc； VGA*1 Internet 82576 （Intel® I350 ）+IPMI2.0 Power Configure According to the Situation
- 60. P60 Distributed GIS Cross Platform GIS T hree Dimension GIS B ig Data GIS A I GIS A I GIS (BitDC)
- 61. P61 Cloud Edge Terminal Edge GIS Server • SuperMap iEdge Cloud GIS Server • SuperMap iServer • SuperMap iPortal • SuperMap iManager Terminal GIS for Components • SuperMap iObjects C++/Java/.NET • SuperMap iObjects Python • SuperMap iObjects for Spark • SuperMap iObjects for Blockchain • SuperMap Scene SDKs for game engines Terminal GIS for Web • SuperMap iClient JavaScript • SuperMap iClient3D for WebGL Terminal GIS for Mobile • SuperMap iMobile • SuperMap iTablet Terminal GIS for Desktop • SuperMap iDesktopX • SuperMap iDesktop New
- 62. P62