Industry processing algorithms:

Roads detection

Description
Formation of a halftone image that determines the location of road network elements (roads and freeways) according to aerospace surveys of high spatial resolution based on the neural network classification algorithm. The main advantage of using neural networks is their noise resistance and high computational speed due to parallel processing of data

Input data
A raster image in GeoTIFF format

Output data
a raster image in GeoTIFF format

Data Requirements
Spatial resolution of initial data – not less than 1 m

Applied fields

  • Road network mapping
  • Create a road map and track their changes over time. Updating of cartographic information and identification of roads not marked on the map or destroyed. Preparing data for creating and updating a vector map of roads.
  • Online monitoring
  • Tracking the process of building roads and assessing the change in their length
  • Spatial analysis
  • Calculation of geometric and spatial characteristics, such as the length and density of the road network. Search for objects (buildings, plots of green spaces, etc.) located at a certain distance from the elements of the road network

Source data
Channels: RED, GREEN, BLUE

Forestry

Description
Classification of forest types: deciduous and coniferous. The classification is based on the analysis of spectral characteristics

Settings

Input data
Raster image in GeoTIFF format

 

Output data
Raster images in GeoTIFF format

Source data
Satellite systems Sentinel-2. Channels: RED, GREEN, BLUE, NIR

Fields contours

Description
Detection of the contours of objects from the data of space surveys using neural networks, the main advantage of which is noise resistance and high computational speed due to parallel processing of data

Input data
Raster image in GeoTIFF format

Output data
Raster image in GeoTIFF format
Vector layer in GEOJSON format

Data Requirements
Spatial resolution of the original data – at least 30 cm
Satellite images of Sentinel-2, RapidEye are used to detect contours of objects in the territories of Ukraine, California and Great Britain.

Applied fields

  • Mapping of crops
  • Classification of crops of agricultural crops and assessment of their condition at the level of individual fields. Creating a plan for growing crops in future vegetation periods
  • Operational agricultural monitoring
  • Identification of changes in crops of crops in different seasons of vegetation. Tracking the process of cultural development during one growing season
  • Spatial analysis
  • Calculation of geometric characteristics of polygonal sites (for example, agricultural land), such as area, perimeter, compactness coefficient, etc. Estimation of the density of sections

Source data
Satellite Systems Sentinel-2

Temperature map

Description
The construction of rasters of near-surface temperature (degrees Celsius, degrees Kelvin) according to Landsat 8, Landsat 7

Input data
Raster image in GeoTIFF format

Output data
Raster images in GeoTIFF format

Source data
Satellite system Landsat 8, Landsat 7. Channels: TIR1, TIR2

Applied fields
GIS analysis, environmental monitoring, ocean monitoring, agromonitoring

Buildings detection

Description
Formation of a halftone image that determines the location of buildings and structures based on aerospace survey data of high spatial resolution based on the neural network classification algorithm. The main advantage of using neural networks is their noise resistance and high computational speed due to parallel processing of data

Input data
Raster image in GeoTIFF format


Output data
Raster image in GeoTIFF format


Data Requirements
Spatial resolution – at least 1 m

Applied fields

  • Urban Planning and Management
  • Creating a map of the location of urban infrastructure. Identify the location of individual objects and outline their outlines. Rapid monitoring of urban development changes. Calculation of the characteristics of urbanization, such as the percentage of area occupied by buildings and structures
  • Online monitoring
  • Comparison of the locations of buildings and structures for different images. Continuous monitoring of urban environment changes, for example, monitoring of construction sites. Tracking the process of urban growth. Detection of new elements of urban development. Assessment of destruction due to natural and man-made disasters (earthquakes, hurricanes, floods, etc.)
  • Spatial analysis
  • Estimation of urban density. Search for buildings located at a certain distance from other elements of the urban environment (roads, parks, other structures, etc.). For example, construction of service areas for banks, shops, etc.

Source data
Channels: RED, GREEN, BLUE

Vehicle detection

Description
Detection of transport units with an area of ​​more than 10 m² according to aerospace survey data. Forming a halftone image indicating the location of vehicles using the neural network classification algorithm

Input data

Raster image in GeoTIFF format

Output data
Raster image in GeoTIFF format
Vector layer in GEOJSON format


Data Requirements
Spatial resolution – at least 30 cm
For detecting vehicles, a set of WorldView-3 satellite images with a spatial resolution of 30 cm is used. The detection accuracy is 80-92% and depends on the quality of a particular image

Applied fields

  • Monitoring of traffic flows
  • Counting the number of transport units and tracking their changes over time. Estimation of traffic density. Tracking the location of cars. Determination of congestion of main transport interchanges
  • Detection of military equipment
  • Identifying units of military equipment and tracking their location in space and time. Counting the number of units of military equipment. Creating a map of moving military equipment

Source data
Satellite system WorldView-3

Manufactured change detection

Description

The algorithm is designed to reveal the land surface changes, using multi-temporal satellite imagery. That is the discovery of new buildings&constructions, traffic arteries; the detection of changes in the infrastructure of megacities. A raster of vegetation index values ​​is used to mask vegetation cover areas that are not involved in processing.

Input data
Raster images in GeoTIFF format

Input data
Raster image in GeoTIFF format
Vector layer in GEOJSON format

Output data
Vector file in GEOJSON format


Source data
Satellite systems Landsat 8, Sentinel-2, RapidEye, PlanetScope, WorldView-2, WorldView-3, Google Aerial NRGB.
Channels: RED, GREEN, BLUE, NIR

Applied fields

Infrastructure development monitoring