import numpy as np
from progressbar import *
import scipy.ndimage.filters as filters
import theano
from pylearn2.space import CompositeSpace,VectorSpace
from pylearn2.models.autoencoder import *
from pylearn2.models.mlp import MLP
import tables
import timeit
from bcfind import extract_patch
import os
import warnings
from bcfind import timer
deconvolver_timer = timer.Timer('Semantic Deconvolution analysis')
class VPrint(object):
def __init__(self,verbose=True):
self.verbose = verbose
def __call__(self, *args, **kwargs):
if self.verbose:
argstring = " ".join([str(arg) for arg in args])
if 'end' in kwargs:
print(argstring,end=kwargs['end'])
else:
print(argstring)
sys.stdout.flush()
"""
Functions for image thresholding
"""
from __future__ import print_function
import math
from bcfind import timer
multi_kapur_timer = timer.Timer('Multi Kapur')
try:
xrange # Python 2
except NameError:
xrange = range # Python 3
def _info(p):
if p > 0:
return -p * math.log(p)
else:
return 0
def kapur(histogram):
"""Computes binarization threshold using the maximum entropy approach
Parameters
----------
histogram : array-like
Image histogram
Returns
-------
import numpy as np
from PIL import Image
from PIL import ImageDraw
import cPickle as pickle
from scipy.spatial import cKDTree
from bcfind import timer
from bcfind.log import tee
from bcfind.utils import mkdir_p, which
from bcfind import log
SHARE_DIR = os.path.dirname(log.__file__)+'/share'
hi2rgb = pickle.load(open(SHARE_DIR+'/hi2rgb.pickle', 'rb'))
save_vaa3d_timer = timer.Timer('Save Vaa3D')
save_markers_timer = timer.Timer('Save markers')
def m_load_markers(filename, from_vaa3d=False):
data = pd.read_csv(filename, skipinitialspace=True, na_filter=False)
if '#x' in data.keys(): # fix some Vaa3d garbage
data.rename(columns={'#x': 'x'}, inplace=True)
if '##x' in data.keys(): # fix some Vaa3d garbage
data.rename(columns={'##x': 'x'}, inplace=True)
C = []
for i in data.index:
row = data.ix[i]
c = Center(0, 0, 0)
for k in row.keys():
setattr(c,k,row[k])
if from_vaa3d:
from torchsummary import summary
import torch.nn as nn
from bcfind import timer
forward_time_teacher = timer.Timer('Forward Time Teacher')
class FC_teacher_max_p(nn.Module):
def __init__(self, n_filters, k_conv=3, k_t_conv = 2, input_channels=1):
super(FC_teacher_max_p, self).__init__()
self.input_channels = input_channels
self.n_filters = n_filters
self.k_conv = k_conv
self.k_t_conv = k_t_conv
self.padding = k_conv // 2
# non riduce le dimensioni spaziali, aumenta soltanto il numero di filtri
self.conv1 = nn.Conv3d(input_channels,
n_filters, self.k_conv, padding=self.padding)
self.conv2 = nn.Conv3d(self.n_filters,
self.n_filters*2, self.k_conv,
padding=self.padding)
self.conv3 = nn.Conv3d(self.n_filters*2,
self.n_filters*4, self.k_conv,
padding=self.padding)