dattalab-6-cam / multicamera_acquisition

Synchronized machine vision acquisition across multiple cameras using an arduino (Supports FLIR and Basler)
MIT License
4 stars 5 forks source link

Frame num #30

Closed jonahpearl closed 7 months ago

jonahpearl commented 7 months ago

Some minor fixes to improve real acquisition performance

-- change metadata col "frame_id" to "frames_received" to reflect what the counter actually counts -- prevent acq loops from closing before camera buffer is used up (still just waiting a long time, probably there is a proper way to check that no more frames are available in the buffer) -- make dropped_frame_warnings False by default, because with the new mp-based logging system they actually take up a lot of CPU, and if you're dropping lots of frames --> issuing many warnings via the logger, then it slows down the main process. -- Corollary to-do of above: currently displaying the frames causes dropped frames, at least for me. I think if we downsample before sending to the display process, that will improve thiings, so gonna change that in a separate PR. -- minor fix on the plotting funcs afterwards to avoid confusing "azure_top" and "top" videos.