nkdnnlr/Event-Based-Camera-Simultaneous-Mosaicing-and-Tracking

Find bottleneck and thus make code run faster

Closed this issue · 1 comments

Time every function and see which ones use the most time. Check if those functions can be rewritten to gain performance.

For

  • events total: 2501
  • batch size: 300 events
  • particles: 200

ncalls tottime percall cumtime percall filename:lineno(function)
image

as expected, measurement update uses most resources, scaling with batch size.