def __init__(self, args):
self.dataset = self.LBLS_DATASET
self.default_data_type = self.LBLS_DTYPE # added to constructors
dpWriteh5.__init__(self,args)
# reinitialize these for non-default data-type
self.EMPTY_LABEL = np.iinfo(self.data_type).max
self.fillvalue = self.EMPTY_LABEL
def __init__(self, args):
self.LIST_ARGS += ['train_offsets', 'prob_types']
dpWriteh5.__init__(self, args)
self.nprob_types = len(self.prob_types)
self.train_chunks = self.train_chunks.reshape((-1, 3))
self.ntrain_chunks = self.train_chunks.shape[0]
if len(self.train_offsets) == 0:
self.train_offsets = np.zeros_like(self.train_chunks)
else:
self.train_offsets = np.array(self.train_chunks).reshape((-1, 3))
assert (self.ntrain_chunks == self.train_offsets.shape[0])
if (self.train_size < 1).all():
self.train_size = np.array(self.chunksize)
assert ((self.size[:2] % self.test_size == 0).all())
self.ntest = self.size[:2] // self.test_size
self.nztrain = self.train_size[2] * self.ntrain_chunks
# print out all initialized variables in verbose mode
if self.dpVolumeXcorr_verbose:
print('dpVolumeXcorr, verbose mode:\n')
print(vars(self))
def __init__(self, args):
dpWriteh5.__init__(self,args)
# print out all initialized variables in verbose mode
if self.dpWarp_verbose: print('dpWarp, verbose mode:\n'); print(vars(self))
def __init__(self, args):
self.LIST_ARGS += dpCubeIter.LIST_ARGS
dpWriteh5.__init__(self, args)
# xxx - also semi-unclean, would fix along with cleaner in/out method
self.dataset_in = self.dataset
self.datasize_in = self.datasize
self.resample_dims = self.resample_dims.astype(bool)
self.nresample_dims = self.resample_dims.sum(dtype=np.uint8)
assert (self.nresample_dims > 0) # no resample dims specified
self.nslices = self.factor**self.nresample_dims
# print out all initialized variables in verbose mode
if self.dpResample_verbose:
print('dpResample, verbose mode:\n')
print(vars(self))
## xxx - probably a way to do this programatically, but easier to read as enumerated.
## this code commented out was only for downsampling by factor of 2
#if (self.resample_dims == np.array([1,0,0])).all():
# self.slices = [np.s_[::2,:,:], np.s_[1::2,:,:]]
#elif (self.resample_dims == np.array([0,1,0])).all():
# self.slices = [np.s_[:,::2,:], np.s_[:,1::2,:]]
#elif (self.resample_dims == np.array([0,0,1])).all():
# self.slices = [np.s_[:,:,::2], np.s_[:,:,1::2]]
#elif (self.resample_dims == np.array([1,1,0])).all():
# self.slices = [np.s_[::2,::2,:], np.s_[1::2,::2,:], np.s_[::2,1::2,:], np.s_[1::2,1::2,:]]
#elif (self.resample_dims == np.array([1,0,1])).all():
# self.slices = [np.s_[::2,:,::2], np.s_[1::2,:,::2], np.s_[::2,:,1::2], np.s_[1::2,:,1::2]]
#elif (self.resample_dims == np.array([0,1,1])).all():
# self.slices = [np.s_[:,::2,::2], np.s_[:,1::2,::2], np.s_[:,::2,1::2], np.s_[:,1::2,1::2]]
#elif self.resample_dims.all():
# self.slices = [np.s_[::2,::2,::2], np.s_[1::2,::2,::2], np.s_[::2,1::2,::2], np.s_[::2,::2,1::2],
# np.s_[1::2,1::2,::2], np.s_[1::2,::2,1::2], np.s_[::2,1::2,1::2], np.s_[1::2,1::2,1::2]]
#assert( len(self.slices) == self.nslices ) # sanity check
# programmatic for factor, but still not for dimensions, again didn't seem worth it, always 3d
self.slices = [None] * self.nslices
f = self.factor
if (self.resample_dims == np.array([1, 0, 0])).all():
for i in range(f):
self.slices[i] = np.s_[i::f, :, :]
elif (self.resample_dims == np.array([0, 1, 0])).all():
for i in range(f):
self.slices[i] = np.s_[:, i::f, :]
elif (self.resample_dims == np.array([0, 0, 1])).all():
for i in range(f):
self.slices[i] = np.s_[:, :, i::f]
elif (self.resample_dims == np.array([1, 1, 0])).all():
for i in range(f):
for j in range(f):
self.slices[i * f + j] = np.s_[i::f, j::f, :]
elif (self.resample_dims == np.array([1, 0, 1])).all():
for i in range(f):
for j in range(f):
self.slices[i * f + j] = np.s_[i::f, :, j::f]
elif (self.resample_dims == np.array([0, 1, 1])).all():
for i in range(f):
for j in range(f):
self.slices[i * f + j] = np.s_[:, i::f, j::f]
elif self.resample_dims.all():
ff = f * f
for i in range(f):
for j in range(f):
for k in range(f):
self.slices[i * ff + j * f + k] = np.s_[i::f, j::f,
k::f]
def __init__(self, args):
self.default_data_type = self.VOXTYPE_DTYPE
#self.data_type = self.VOXTYPE_DTYPE
self.dataset = self.VOXTYPE_DATASET
self.fillvalue = self.EMPTY_VOXTYPE
dpWriteh5.__init__(self,args)
def __init__(self, args):
#self.data_type = self.PROBS_DTYPE
self.default_data_type = self.PROBS_DTYPE
self.fillvalue = self.EMPTY_PROB
#self.dataset = self.PROBS_DATASET # don't do this, set in classmethod
dpWriteh5.__init__(self,args)
