Optical flow computation in parallel with TopologyOpticalFlow

This minimal example presents how to carry out a simple optical flow computation. See also the documentation of the class fluidimage.topologies.optical_flow.TopologyOpticalFlow and the work defined in the subpackage fluidimage.works.optical_flow.

from fluidimage import get_path_image_samples
from fluidimage.optical_flow import Topology

params = Topology.create_default_params()

params.series.path = get_path_image_samples() / "Karman/Images"
params.series.ind_step = 1

# params.features._print_doc()
params.features.maxCorners = 100000
params.features.qualityLevel = 0.05
params.features.blockSize = 20

# params.optical_flow._print_doc()
params.optical_flow.maxLevel = 2
params.optical_flow.winSize = (48, 48)

params.mask.strcrop = ":, 50:500"

params.saving.how = "recompute"
params.saving.postfix = "optflow_example"

topology = Topology(params, logging_level="info")

# To produce a graph of the topology
# topology.make_code_graphviz('topo.dot')

# Compute in parallel
topology.compute()

# Compute in sequential (for debugging)
# topology.compute(sequential=True)

assert len(topology.results) == 3