Overview

AI GIS is the integration of AI and GIS. It includes the following features:
1) Combines GeoAI and relevant process tools.
2) Enhances the function and interactive user experience of GIS software based on AI technology, and improves the intelligence of GIS software.
3) Management, visualization and analysis of GeoAI results based on GIS.

Highlights

Improves AI GIS function of all products
  • Improves server machine learning service, and newly supports various geospatial machine learning functions, such as general change detection.
  • Supports desktop AI marking function, broken road detection model and Yolo V5 model.
  • Supports general change detection of component terminal, and supports multiple new deep learning model like SFNet.
  • Improves mobile AI+AR analysis, AI attribute acquisition, AI mapping, etc.

Improves AI GIS workflow tools
  • Supports image sample management in the data preparation stage.
  • Supports post-processing tools for image analysis reasoning results, such as polygon aggregation, building regularization, etc.
  • Enhances model evaluation ability in model application stage.

Improves geospatial sampling and statistical inference function
  • Supports simple random sampling, systematic sampling and stratified sampling.
  • Supports geospatial random sampling, geospatial stratified sampling and sandwich sampling.
  • Spports SPA model and B-Shade model.

Supports various geospatial machine learning functions
  • Cluster analysis: supports geospatial hotspot analysis, geospatial density clustering, k-means clustering, shift mean clustering, etc.
  • Classification analysis: map matching, address element identification, forest-based classification, etc.
  • Regression analysis: geosimulation, geographically weighted regression, Spatiotemporal geographical weighted regression, forest-based regression, etc.

Supports various deep learning model
  • Image analysis target detection: Cascade R-CNN, Faster R-CNN, RetinaNet.
  • Binary classification of image analysis: FPN, DeepLabv3+, U-Net, D-LinkNet, SFNet.
  • Image analysis ground-object classification: FPN, DeepLab V3+, U-Net, SFNet.
  • Image analysis scene classification: EfficientNet.
  • Image analysis object extraction: Mask R-CNN.
  • Image general change detection: DSAMNet, Siam-SFNet.

Upgrades deep learning framework
  • Upgrades TensorFlow framework from version 2.3 to 2.6 and Pytorch framework from version 1,8 to 1.10.
  • Upgrades CUDA to version 11.3, and supports RTX30.

Improves mobile AI
  • Improves AI attribute acquisition, AI mapping and AI+AR analysis.
  • Supports geo-fencing, speed limit analysis and video segmentation.