Object based feature extraction of coniferous seedlings in Alberta

    Slide 1

    Slide 2
  • Slide 3
  • Slide 4
  • Slide 5
  • Slide 6
  • Slide 7
  • Slide 8
  • Slide 9
  • Slide 10


Object Oriented Extraction of Coniferous Seedlings in Alberta, Canada.

The goal of this project was to assess whether UAVs can collect efficient and accurate density estimates of seedlings on replanted forestry blocks. Two cameras were flown by Corvis Aerial Inc. to collect imagery on a forestry block managed by Weyerhaeuser Canada. The imagery was combined and segmented. Those image objects were classified using various methods to produce a workflow to count coniferous seedlings in each plot. See a video about the project below.

  • Extensive use of ARGIS for: combining RGB and NIR images, creating probability layers (Spatial Analyst), processing CART classifications
  • Multispectral UAV and satellite imagery
  • CART (Machine Learning)
  • eCognition for object oriented classification
  • Python for accuracy assessment
  • Pointcloud and 3D model editing and metric extraction
  • 3D printing and editing
  • UAV customization, sensor integration, programming and repair


Browse this 3D model of a seedling plot. Give it 30 seconds or so to load. Point cloud metrics were derived from models like these following photogrammetric processing from UAV imagery.