Shape landmark detection through computer vision

Dirk Steinke (IB)

Geometric morphometrics is the statistical analysis of shape using landmark coordinates. Those landmark coordinates can come from various sources and are usually captured manually using a digitizing tablet or by measuring an image using software. There have been a few attempts to find shape landmarks automatically through computer vision and this project will make use of such tools to build a pipeline that automatically retrieves shape landmarks as well as basic morphometric measurements from a stack of images retrieved from the Barcode of Life Datasystems (BOLD) database.

Prospective students could work on the following aspects of the project:

  • Automatic shape recognition of pre-filtered image sets (Arthopods)
  • Landmark setting (including orientation correction)
  • Basic automatic measurements (length, width of specimens)