PMD Compression

Simple PMD compression, the hello world of masknmf

[25-10-26 02:13:47]: Starting compression
[25-10-26 02:13:47]: sampled from the following regions: [0]
[25-10-26 02:13:47]: We are initializing on a total of 2000 frames
[25-10-26 02:13:48]: Loading data to estimate complete spatial basis
[25-10-26 02:13:48]: skipping the pruning step for frame cutoff
[25-10-26 02:13:48]: Finding spatiotemporal roughness thresholds

  0%|          | 0/250 [00:00<?, ?it/s]
  2%|▏         | 5/250 [00:00<00:05, 45.36it/s]
  4%|▍         | 10/250 [00:00<00:05, 46.63it/s]
  6%|▌         | 15/250 [00:00<00:05, 46.88it/s]
  8%|▊         | 21/250 [00:00<00:04, 48.90it/s]
 11%|█         | 27/250 [00:00<00:04, 50.05it/s]
 13%|█▎        | 33/250 [00:00<00:04, 50.66it/s]
 16%|█▌        | 39/250 [00:00<00:04, 51.08it/s]
 18%|█▊        | 45/250 [00:00<00:03, 51.41it/s]
 20%|██        | 51/250 [00:01<00:03, 51.51it/s]
 23%|██▎       | 57/250 [00:01<00:03, 51.55it/s]
 25%|██▌       | 63/250 [00:01<00:03, 51.69it/s]
 28%|██▊       | 69/250 [00:01<00:03, 51.84it/s]
 30%|███       | 75/250 [00:01<00:03, 51.97it/s]
 32%|███▏      | 81/250 [00:01<00:03, 51.97it/s]
 35%|███▍      | 87/250 [00:01<00:03, 51.96it/s]
 37%|███▋      | 93/250 [00:01<00:03, 51.97it/s]
 40%|███▉      | 99/250 [00:01<00:02, 51.93it/s]
 42%|████▏     | 105/250 [00:02<00:02, 51.99it/s]
 44%|████▍     | 111/250 [00:02<00:02, 52.00it/s]
 47%|████▋     | 117/250 [00:02<00:02, 52.04it/s]
 49%|████▉     | 123/250 [00:02<00:02, 52.06it/s]
 52%|█████▏    | 129/250 [00:02<00:02, 51.99it/s]
 54%|█████▍    | 135/250 [00:02<00:02, 52.03it/s]
 56%|█████▋    | 141/250 [00:02<00:02, 52.05it/s]
 59%|█████▉    | 147/250 [00:02<00:01, 52.12it/s]
 61%|██████    | 153/250 [00:02<00:01, 52.10it/s]
 64%|██████▎   | 159/250 [00:03<00:01, 52.09it/s]
 66%|██████▌   | 165/250 [00:03<00:01, 52.06it/s]
 68%|██████▊   | 171/250 [00:03<00:01, 51.92it/s]
 71%|███████   | 177/250 [00:03<00:01, 51.93it/s]
 73%|███████▎  | 183/250 [00:03<00:01, 51.93it/s]
 76%|███████▌  | 189/250 [00:03<00:01, 51.96it/s]
 78%|███████▊  | 195/250 [00:03<00:01, 51.90it/s]
 80%|████████  | 201/250 [00:03<00:00, 51.94it/s]
 83%|████████▎ | 207/250 [00:04<00:00, 51.99it/s]
 85%|████████▌ | 213/250 [00:04<00:00, 52.02it/s]
 88%|████████▊ | 219/250 [00:04<00:00, 51.99it/s]
 90%|█████████ | 225/250 [00:04<00:00, 52.04it/s]
 92%|█████████▏| 231/250 [00:04<00:00, 52.03it/s]
 95%|█████████▍| 237/250 [00:04<00:00, 51.99it/s]
 97%|█████████▋| 243/250 [00:04<00:00, 51.99it/s]
100%|█████████▉| 249/250 [00:04<00:00, 52.00it/s]
100%|██████████| 250/250 [00:04<00:00, 51.63it/s]
[25-10-26 02:13:53]: Running Blockwise Decompositions
[25-10-26 02:13:53]: Constructed U matrix. Rank of U is 201
[25-10-26 02:13:53]: PMD Objected constructed
/opt/hostedtoolcache/Python/3.12.11/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()