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