def runTest(self):
fafb = Fafb()
p = (217400, 164242, 438817) # z,y,x convention
crop_size = (16, 160, 160)
p_crop = fafb.get_crops([p], crop_size)
normalized_p_crop = fafb.normalize(p_crop)
self.assertTrue(
np.all(np.shape(normalized_p_crop) == np.shape(p_crop)))
self.assertTrue(np.min(normalized_p_crop) >= -1)
self.assertTrue(np.max(normalized_p_crop) <= 1)
def runTest(self):
fafb = Fafb()
p = (217400, 164242, 438817) # z,y,x convention
crop_size = (16, 160, 160)
p_crop = fafb.get_crops([p], crop_size)
self.assertTrue(np.max(p_crop) <= 255)
self.assertTrue(np.min(p_crop) >= 0)
self.assertTrue(np.all(crop_size == np.shape(p_crop)[1:]))
self.assertTrue(1 == np.shape(p_crop)[0])
self.assertTrue(4 == len(np.shape(p_crop)))
def runTest(self):
fafb = Fafb()
self.assertTrue(fafb.name == "FAFB")
self.assertTrue(fafb.container ==
"/nrs/saalfeld/FAFB00/v14_align_tps_20170818_dmg.n5")
self.assertTrue(fafb.dataset == "volumes/raw/s0")
self.assertTrue(np.all(fafb.voxel_size == np.array([40, 4, 4])))
def __init__(self):
dataset = Fafb()
service = Flywire(
dataset=dataset,
api_url="https://spine.janelia.org/app/transform-service",
api_dataset="flywire_v1",
)
super().__init__(dataset=dataset, service=service)
self.catmaid = FafbCatmaid()
execute=False,
expand=True)
else:
cmd = base_cmd
cmd = [c for c in cmd.split(" ") if c != '']
cmd_string = ""
for c in cmd:
cmd_string += str(c) + " "
print(cmd_string)
Popen(cmd_string, shell=True)
if __name__ == "__main__":
log_config("brain.log")
submit_jobs(
db_credentials=
"/groups/funke/home/ecksteinn/Projects/synex/synisterbrain/db_credentials.ini",
db_name="synful_synapses",
collection_name="partners",
predict_id=3,
dataset=Fafb(),
model=FafbModel(),
n_gpus=2,
n_cpus=5,
batch_size=16,
prefetch_factor=20,
queue="gpu_any",
singularity_container=None,
mount_dirs=["/nrs", "/scratch", "/groups", "/misc"])
def __init__(self):
dataset = Fafb()
service = Catmaid(
dataset=dataset,
api_url="https://neuropil.janelia.org/tracing/fafb/v14/")
super().__init__(dataset=dataset, service=service)
array_data = None
if self.data.roi.contains(roi):
array = self.data[roi]
array.materialize()
array_data = array.data.astype(np.float32)
array_data = self.dataset.normalize(array_data)
if self.transform is not None and array_data is not None:
array_data = self.transform(array_data)
return {"id": synapse_id, "data": array_data}
if __name__ == "__main__":
mongo_em = MongoIterator("/groups/funke/home/ecksteinn/Projects/synex/synister/db_credentials.ini",
"synful_synapses",
"partners",
3,
Fafb(),
400,
400,
80,
1,0,1,0)
print(Fafb().voxel_size)
i = 0
for doc in mongo_em:
print(doc)
i += 1
if i > 2: break
self.db_name,
self.collection_name,
self.predict_id,
self.dataset,
self.dx,
self.dy,
self.dz,
self.n_gpus,
self.gpu_id,
n_cpus=n_cpus,
cpu_id=cpu_id,
transform=self.transform_to_tensor)
def transform_to_tensor(self, data_array):
tensor_array = torch.tensor(data_array)
# Add channel and batch dim:
tensor_array = tensor_array.unsqueeze(0).unsqueeze(0)
return tensor_array
if __name__ == "__main__":
mongo_em = MongoEM(
"/groups/funke/home/ecksteinn/Projects/synex/synister/db_credentials.ini",
"synful_synapses", "partners", 3, Fafb(), 400, 400, 80, 2, 1)
i = 0
for doc in mongo_em:
print(doc)
i += 1
if i > 2: break