Time per frame used by CaImAn (red, n = 1,500 timings) and variants of FIOLA to process the K53 dataset as a function of frame size and number of detected neurons. FIOLA variants include frame-by-frame processing of a full FOV (blue, n = 2,997), cropped for motion correction (orange, 25% FOV by area, n = 2,997) and batch processing on the full FOV (purple, batch size 100 frames, no crop, n = 29). Dashed lines represent framewise typical speeds for voltage and calcium imaging. FIOLA also achieved gains in speed over CaImAn for downsampled 256 x 256 datasets (Supplementary Fig. 7a-c ). Boxes represent the 25th-75th percentiles of the recorded time per frame with a line at the median, while the whiskers range from the 0. 1 to the 99. 9 percentiles. Circles represent outliers. Data for each bar are presented as the mean.
