PMD Compression

Simple PMD compression, the hello world of masknmf

/home/runner/work/masknmf-toolbox/masknmf-toolbox/masknmf/utils/_cuda.py:7: UserWarning: You've explicitly selected to perform computations on the cpu, performance will be significantly slower
  warn(
[25-12-05 19:24:48]: Starting compression
[25-12-05 19:24:48]: sampled from the following regions: [0]
[25-12-05 19:24:48]: We are initializing on a total of 2000 frames
[25-12-05 19:24:48]: Loading data to estimate complete spatial basis
[25-12-05 19:24:48]: skipping the pruning step for frame cutoff
[25-12-05 19:24:48]: Finding spatiotemporal roughness thresholds

  0%|          | 0/250 [00:00<?, ?it/s]
  2%|▏         | 5/250 [00:00<00:05, 47.59it/s]
  4%|▍         | 10/250 [00:00<00:04, 48.95it/s]
  6%|▌         | 15/250 [00:00<00:04, 49.31it/s]
  8%|▊         | 21/250 [00:00<00:04, 49.60it/s]
 11%|█         | 27/250 [00:00<00:04, 49.74it/s]
 13%|█▎        | 32/250 [00:00<00:04, 49.50it/s]
 15%|█▌        | 38/250 [00:00<00:04, 49.68it/s]
 17%|█▋        | 43/250 [00:00<00:04, 49.64it/s]
 19%|█▉        | 48/250 [00:00<00:04, 49.31it/s]
 21%|██        | 53/250 [00:01<00:03, 49.39it/s]
 23%|██▎       | 58/250 [00:01<00:03, 49.34it/s]
 25%|██▌       | 63/250 [00:01<00:03, 49.27it/s]
 28%|██▊       | 69/250 [00:01<00:03, 49.47it/s]
 30%|███       | 75/250 [00:01<00:03, 49.65it/s]
 32%|███▏      | 81/250 [00:01<00:03, 49.73it/s]
 34%|███▍      | 86/250 [00:01<00:03, 49.76it/s]
 36%|███▋      | 91/250 [00:01<00:03, 49.81it/s]
 38%|███▊      | 96/250 [00:01<00:03, 49.77it/s]
 40%|████      | 101/250 [00:02<00:02, 49.71it/s]
 42%|████▏     | 106/250 [00:02<00:02, 49.63it/s]
 44%|████▍     | 111/250 [00:02<00:02, 49.51it/s]
 46%|████▋     | 116/250 [00:02<00:02, 49.52it/s]
 48%|████▊     | 121/250 [00:02<00:02, 49.46it/s]
 50%|█████     | 126/250 [00:02<00:02, 49.22it/s]
 52%|█████▏    | 131/250 [00:02<00:02, 49.26it/s]
 55%|█████▍    | 137/250 [00:02<00:02, 49.68it/s]
 57%|█████▋    | 142/250 [00:02<00:02, 49.67it/s]
 59%|█████▉    | 147/250 [00:02<00:02, 49.55it/s]
 61%|██████    | 153/250 [00:03<00:01, 49.66it/s]
 63%|██████▎   | 158/250 [00:03<00:01, 49.74it/s]
 65%|██████▌   | 163/250 [00:03<00:01, 49.77it/s]
 67%|██████▋   | 168/250 [00:03<00:01, 49.70it/s]
 70%|██████▉   | 174/250 [00:03<00:01, 49.94it/s]
 72%|███████▏  | 179/250 [00:03<00:01, 49.90it/s]
 74%|███████▎  | 184/250 [00:03<00:01, 49.72it/s]
 76%|███████▌  | 189/250 [00:03<00:01, 49.72it/s]
 78%|███████▊  | 194/250 [00:03<00:01, 49.52it/s]
 80%|███████▉  | 199/250 [00:04<00:01, 49.57it/s]
 82%|████████▏ | 204/250 [00:04<00:00, 49.46it/s]
 84%|████████▍ | 210/250 [00:04<00:00, 49.79it/s]
 86%|████████▋ | 216/250 [00:04<00:00, 49.78it/s]
 89%|████████▉ | 222/250 [00:04<00:00, 49.81it/s]
 91%|█████████ | 228/250 [00:04<00:00, 49.90it/s]
 94%|█████████▎| 234/250 [00:04<00:00, 50.00it/s]
 96%|█████████▌| 240/250 [00:04<00:00, 50.09it/s]
 98%|█████████▊| 246/250 [00:04<00:00, 49.95it/s]
100%|██████████| 250/250 [00:05<00:00, 49.65it/s]
[25-12-05 19:24:53]: Running Blockwise Decompositions
[25-12-05 19:24:53]: Constructed U matrix. Rank of U is 208
[25-12-05 19:24:53]: PMD Objected constructed
/opt/hostedtoolcache/Python/3.12.12/x64/lib/python3.12/site-packages/fastplotlib/graphics/features/_base.py:18: UserWarning: casting float64 array to float32
  warn(f"casting {array.dtype} array to float32")

# test_example = true

import masknmf
import torch
import fastplotlib as fpl
from urllib.request import urlretrieve
import tifffile


urlretrieve(
    "https://github.com/flatironinstitute/CaImAn/raw/refs/heads/main/example_movies/demoMovie.tif",
    "./demo.tif"
)

# always lazy load raw data by memmaping or other methods
data = tifffile.imread("./demo.tif")

block_sizes = [32, 32]
max_components = 20

# it's recommended to use masknmf on a machine with a GPU
# device = "cuda" if torch.cuda.is_available() else "cpu"
device = "cpu"
# number of frames used to estimate the spatial basis in PMD
num_frames_for_spatial_fit = data.shape[0]

# perform PMD
pmd_result = masknmf.compression.pmd_decomposition(
    data,
    block_sizes,
    num_frames_for_spatial_fit,
    max_components=max_components,
    device=device,
    frame_batch_size=1024
)

# get the residual
pmd_residual = masknmf.PMDResidualArray(data, pmd_result)

# view the movies, note that all these array are LAZY evaluated, allowing you to view extremely large datasets!
iw = fpl.ImageWidget(
    data=[data, pmd_result, pmd_residual],
    names=["raw", "pmd", "residual"],
    figure_kwargs={"size": (1000, 340), "shape": (1, 3)},
    cmap="gnuplot2",
)

iw.show()

# use the time slider or set the frame index programmatically
iw.current_index = {"t": 1610}

# manually set vmin-vmax to emphasize noise in raw video
# you can also adjust the vmin-vmax using the histogram tool
# reset the vmin-vmax by clicking the buttons under "ImageWidget Controls"
for image in iw.managed_graphics:
    image.vmax = 3_200

# remove toolbar to reduce clutter
for subplot in iw.figure:
    subplot.toolbar = False


# ignore the remaining lines these are just for docs generation
figure = iw.figure
if __name__ == "__main__":
    print(__doc__)
    fpl.loop.run()