Skip to content

Latest commit

 

History

History
15 lines (11 loc) · 505 Bytes

README.md

File metadata and controls

15 lines (11 loc) · 505 Bytes

Build status

Fast 2D DBSCAN

Fast C++ implementation of DBSCAN for 2D data that follows the sklearn interface.

It's not exactly fast yet. TODO:

  • Add tests
  • Add benchmark (maybe together with tests using Catch2)
  • Use 2D grid-based hashing instead of kd-tree for neighborhood lookup
  • Use SIMD vectorization
  • port to GPU
  • build and export Python wheels
  • support CMake for building