/computing-density-maps

Fast computing density maps for ShanghaiTech and other datasets

Primary LanguageJupyter Notebook

Fast computing density maps for ShanghaiTech

You can choose how to compute sigma : distance to 3 nearest neighbors, distance to nearest neighbor, or fixed value. Optionally clip sigma if it is lower than threshold.

I've used pre-computed gaussian kernels and fast neighbors search to speed up density maps creation.