Car Detection

Orbital Insight’s car detection algorithm uses wide area object detection in Pleiades imagery to accurately identify and quantify cars. This saves analysts significant amounts of time in conducting pattern of life analyses and activity based intelligence.

Technical Description

Orbital Insight’s Car Detector operates on high resolution imagery from Pléiades Display data and produces point detections corresponding to the location of cars within a specified input image (GeoTIFF).

Supported workflows

Data platform:

• Catalog -> Pléiades Display -> Order and get the asset from storage of UP42
• In Projects workflows, Processing from Storage -> DIMAP GeoTIFF conversion -> Raster CRS Conversion (CRS: EPSG:3857) -> Raster Tiling -> Car Detection

Data blocks:

• Pléiades Display (Streaming) -> Raster Tiling -> Car Detection

  General Information	Description
  Block Type	Processing for Pléiades Car Detection
  Supported input data	Pléiades Tile (1232x1232px)
  Output data format	GeoJSON
  Performance	Algorithm qualified on a large set of data encompassing different types of landscapes worldwide by a third party, with a high level of performances. The algorithm is applicable in wide areas worldwide. Its performance may however vary according to geography and imagery features. For example, the algorithm is more reliable in urban areas. The algorithm generally performs better in North American and European areas of interest. The algorithm has known limitations when dealing with highly shadowed imagery, those containing closely packed over vehicles (0 pixels in between), and desert imagery.
  Algorithm Training Data Details	This algorithm was tested by a 3rd party using a validation set, consisting of approximately 100,000 marked cars in 50 countries spanning 20,000 images of deserts, ports and parking lots across 6 continents. The training set contains 180,000 images and spans many variable conditions including time of day, time of year, terrain, configuration of vehicles, etc.