Simran's edits
hi
#21/10/06 notes merging test
Some other changes
making changes second time
third time changes NEW LINE NEW NOTES
import numpy as np import pandas as pd import cv2, os, pathlib, sys, math, os.path, json, glob import imageio, png, statistics, time from ipywidgets import interact, fixed import matplotlib import matplotlib.image as mpimg import matplotlib.pyplot as mpplt import mayavi from mpl_toolkits import mplot3d from mpl_toolkits.mplot3d import Axes3D import nibabel as nib import nipype import nipype.algorithms.metrics from nipype.algorithms.metrics import Distance, Overlap, FuzzyOverlap, ErrorMap from scipy import misc, stats, ndimage from scipy.ndimage import label, measurements from scipy.spatial.distance import cdist from scipy.spatial import Voronoi, voronoi_plot_2d, ConvexHull, convex_hull_plot_2d import SimpleITK as sitk import sklearn from sklearn import preprocessing from sklearn.cluster import KMeans import skimage from skimage import color, data, exposure, filters, img_as_float, img_as_uint, io, measure from skimage.color import label2rgb, rgb2gray from skimage.feature import peak_local_max, canny, greycomatrix from skimage.filters import rank, sobel, scharr, threshold_local, threshold_otsu from skimage.filters.rank import bottomhat, mean_bilateral, minimum, percentile from skimage.morphology import disk, local_minima from skimage.util import img_as_ubyte, invert from sklearn.metrics import jaccard_score from pathlib import Path from PIL import Image from datetime import datetime from matplotlib import cm from collections import OrderedDict import re, csv from progress.bar import IncrementalBar from collections import OrderedDict import shutil
import os, os.path, pathlib, cv2 import scipy.stats import numpy as np import pandas as pd import cv2, sys, math, json, glob import os, os.path import imageio, statistics, time import matplotlib import matplotlib.image as mpimg import matplotlib.pyplot as mpplt import nibabel as nib import SimpleITK as sitk import nipype import nipype.algorithms.metrics import mayavi import SimpleITK as sitk from scipy.spatial import ConvexHull, convex_hull_plot_2d from skimage import color, data, exposure, filters, img_as_float, img_as_uint, io, measure from pathlib import Path from PIL import Image from datetime import datetime from mpl_toolkits import mplot3d from mpl_toolkits.mplot3d import Axes3D from sklearn.metrics import jaccard_score from nipype.algorithms.metrics import Distance, Overlap, FuzzyOverlap, ErrorMap from ipywidgets import interact, fixed import png import re, csv
############################################## ##############################################
def make_mosaic_inventory(): #creating a picture from smaller pictures # Set up output filename
top_dir = os.getcwd()
run_location = top_dir.split('/')[-1:][0]
top_dir = os.getcwd() #This method returns current working directory of a process
run_location = top_dir.split('/')[-1:][0] #spliting the folder??
# The extension to search for
extensions = ['.jpx', '.txt']
# First, cataloge the directory of images and their
# paths (i.e. the file, and each one's location for retrieval)
for dirpath, dirnames, files in os.walk(top_dir):
ln = len(files)
bar = IncrementalBar('Processing:', max=ln)
for name in files:
if name.lower().endswith(tuple(extensions)):
for dirpath, dirnames, files in os.walk(top_dir): #creating file name
ln = len(files) #length of file
bar = IncrementalBar('Processing:', max=ln) #processing vs progress to see process instead visual progress
for name in files:
if name.lower().endswith(tuple(extensions)): #extensions are jpx and txt
if '._' in name:
continue
if '_CompilerResults' in name:
continue
if '_AssemblyComplete' in name:
continue
item_path = os.path.join(dirpath, name)
item_path = os.path.join(dirpath, name) #creating file names
mdata = Path(name).stem
container = '_'.join(mdata.split('_')[:-1])
# Nomenclature for output
nd = {}
nd['container'] = container
nd['fpath'] = item_path
nd['fname'] = name
# Make new dataframe with output
df1 = pd.DataFrame().append(nd, ignore_index=True)
out_fname = run_location + '_InventoryMosaic.csv'
if os.path.exists(out_fname):
df1.to_csv(out_fname, mode='a', header=False, index=False)
else:
df1.to_csv(out_fname, header=True, index=False)
bar.next()
bar.finish()
df1.to_csv(out_fname, mode='a', header=False, index=False) #creating a csv file
else:
df1.to_csv(out_fname, header=True, index=False)
bar.next() #moving on to create next name to lopp through the next and ipdate the progress bar
bar.finish() #once looped through everything
############################################## ##############################################
def assembly_txt_conversion(input_file): """ Convert Assembly Data File Type """ # Convert assembly.txt to assembly.csv # The goal here is to make accessible all the data from the Assembly file core_id = Path(input_file).stem #2021/10/06 returns the basename of input_file without extension core_id = ''.join(core_id.split('')[:-1]) #2021/10/06 joining core_id.split, seperated by '_' - getting rid of the image # (which will eventually be replaced by new id) txt_file = input_file csv_file = core_id + '_AssemblyConverted.csv' #2021/10/06 create a new variable that stores the path+csv suffix with open(txt_file, 'rt') as infile, open(csv_file, 'w+') as outfile: #2021/10/06 create output csv files stripped = (line.strip() for line in infile) #2021/10/06 strip spaces in line from the .txt infile lines = (line.split(",") for line in stripped if line) #2021/10/06 replace the between strings space with ',' and store the new string as lines writer = csv.writer(outfile) writer.writerows(lines) #2021/10/06 write lines into output csv files
core_id = '_'.join(core_id.split('_')[:-1])
txt_file = input_file
csv_file = core_id + '_AssemblyConverted.csv'
with open(txt_file, 'rt') as infile, open(csv_file, 'w+') as outfile:
stripped = (line.strip() for line in infile)
lines = (line.split(",") for line in stripped if line)
writer = csv.writer(outfile)
writer.writerows(lines)
############################################## ##############################################
def reorder_assembly_data(input_data): """ Ingest the Assembly Data """ # First, get the container metadata mdata = Path(input_data).stem mdat = mdata.split('')[:-1] mdt = ''.join(mdat) # Split the original into parts (and ignore the end) # Read in the converted original file df0 = pd.read_csv(input_data, header=None) # Split into relevant parts # Constant size for end part of file data # The last four rows are the footer df_end = df0[-4:] # Constant size for top part of file data # The first 27 rows are the header df_metadata = df0[0:26] # Variable size of coordinate data part of file data # Here, defined as portion not including header AND footer (will output df of whatever size) df_data = df0[27:-4] """ Transform converted file of tile location data """ # Ideally, would split image delta values from tile data... # First, make a list of all the Tiles... after converting to dataframe for index, row in df_data.iterrows(): # Ignore the "delta" file data for now if 'Delta' in row[0]: continue else: if ' = ' in row[0]: data = row[0].split(' = ') label = data[0] value = int(data[1]) tile_var = label[:-1] coord_label = str(label[-1]) # Adjust index for naming conventions tile_no = int(tile_var[4:]) + 1 tile_num = '{:0>6d}'.format(tile_no) coord_val = '{:0>6d}'.format(value) tile_id = mdt + '_' + tile_num # Set up naming for output: nd = {} nd['tile_id'] = tile_id nd['coord_label'] = coord_label nd['coord_value'] = coord_val new_df = pd.DataFrame().append(nd, ignore_index=True) output_path = mdt + '_AssemblyModified.csv' if os.path.exists(output_path): new_df.to_csv(output_path, mode='a', header=False, index=False) else: new_df.to_csv(output_path, header=True, index=False)
############################################## ##############################################
def ingest_assembly_data(in_data): # First, get the container metadata mdata = Path(in_data).stem medat = mdata.split('')[:-1] mdt = ''.join(medat) # Read in the modified data df1 = pd.read_csv(in_data) # Make unique list of objects flist = sorted(df1['tile_id'].unique().tolist()) completed = [] # Loop through unique list of images (jpx) for item in flist: if item not in completed: completed.append(item) else: continue # Grab relevant data from original dataframe df_temp = df1.loc[df1['tile_id'] == item].copy() # And loop through to create new file for index, row in df_temp.iterrows(): if 'X' == row.coord_label: x_coord = row.coord_value x_coord = '{:0>6d}'.format(x_coord) x_val = 'x' + x_coord if 'Y' == row.coord_label: y_coord = row.coord_value y_coord = '{:0>6d}'.format(y_coord) y_val = 'y' + y_coord curr = os.getcwd() old_id = item + '.jpx' original_path = os.path.join(curr, old_id) mdata = item.split('') mdat = mdata[:-1] sample = mdat[0] series = mdat[1] z_coord = mdat[2] block = mdat[3] side = mdat[4] new_id = ''.join([sample, side, series, block, z_coord, y_val, x_val]) # Output new dataframe nd = {} nd['tile_id'] = item nd['original_path'] = original_path nd['new_global_id'] = new_id new_df = pd.DataFrame().append(nd, ignore_index=True) output_path = mdt + '_AssemblyGlobal.csv' if os.path.exists(output_path): new_df.to_csv(output_path, mode='a', header=False, index=False) else: new_df.to_csv(output_path, header=True, index=False)
############################################## ##############################################
import numpy as np import pandas as pd import cv2, os, pathlib, sys, math, os.path, json, glob import imageio, png, statistics, time from ipywidgets import interact, fixed import matplotlib import matplotlib.image as mpimg import matplotlib.pyplot as mpplt import mayavi from mpl_toolkits import mplot3d from mpl_toolkits.mplot3d import Axes3D import nibabel as nib import nipype import nipype.algorithms.metrics from nipype.algorithms.metrics import Distance, Overlap, FuzzyOverlap, ErrorMap from scipy import misc, stats, ndimage from scipy.ndimage import label, measurements from scipy.spatial.distance import cdist from scipy.spatial import Voronoi, voronoi_plot_2d, ConvexHull, convex_hull_plot_2d import SimpleITK as sitk import sklearn from sklearn import preprocessing from sklearn.cluster import KMeans import skimage from skimage import color, data, exposure, filters, img_as_float, img_as_uint, io, measure from skimage.color import label2rgb, rgb2gray from skimage.feature import peak_local_max, canny, greycomatrix from skimage.filters import rank, sobel, scharr, threshold_local, threshold_otsu from skimage.filters.rank import bottomhat, mean_bilateral, minimum, percentile from skimage.measure import find_contours from skimage.morphology import disk, local_minima from skimage.segmentation import * from skimage.restoration import * from skimage.util import img_as_ubyte, invert from sklearn.metrics import jaccard_score from pathlib import Path from PIL import Image from datetime import datetime from matplotlib import cm from collections import OrderedDict import re, csv from progress.bar import IncrementalBar from collections import OrderedDict import shutil
############################################## ##############################################
def object_detection(image): curr_dir = os.getcwd() # For segmenting stained (black) neurons in light brown tissue -- MGN project #im_rgb = imageio.imread('W312_SeriesBDAd_Section0020_Brain_Whole_000210_xa1000xb1400ya0400yb) im_rgb = imageio.imread(image) metadata = Path(image).stem mdata = ''.join(metadata.split('')[:-1]) # Convert to HSV space im_hsv = color.convert_colorspace(arr=im_rgb, fromspace='rgb', tospace='hsv') val = im_hsv[:,:,2] # Normalize image lighting claheV = exposure.equalize_adapthist(image=val, kernel_size=10, clip_limit=0.01, nbins=100) # Aggressively smooth image to eliminate background claheV = img_as_ubyte(claheV) bil1V = rank.mean_bilateral(image=claheV, selem=disk(radius=50), s0=100, s1=100) # Set up threshold parameters to eliminate outlier minimal values mx = np.amax(bil1V) mn = np.amin(bil1V) rng = (mx-mn) thresh = mx - (rng*0.5) # print('max', mx, 'min', mn, 'range', rng, 'threshold', thresh) im = bil1V im[im < thresh] = 0 # Aggressively smooth image to eliminate background bilm = rank.mean_bilateral(image=im, selem=disk(radius=50), s0=100, s1=100) # Merge nearby local minimum values, and test array range for object (!!!) kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5,5)) seeds = img_as_ubyte(cv2.dilate(bilm, kernel, iterations=5)) mx1 = np.max(seeds) mn1 = np.min(seeds) rng1 = mx1-mn1 if rng1 > 100: # Set nomenclature out_fname = mdata + '_Object.png' out_dir = '/Users/djm/Desktop/2018_python/test_targets/' # If object is present, write contours to file kernel_m = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (10,10)) mask = img_as_ubyte(cv2.dilate(bilm, kernel_m, iterations=1)) contours, hierarchy = cv2.findContours(mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE) prediction = cv2.drawContours(mask, contours, -1, (255, 0, 0), cv2.FILLED) pred = invert(prediction) im0 = Image.fromarray(pred) im0.save(out_fname) # Now, move both files to folder containing target segmentations shutil.move(src=out_fname, dst=out_dir) shutil.move(src=image, dst=out_dir) else: # # WIP ...
# Set nomenclature
out_fname = mdata + '_Blank.png'
out_dir = '/Users/djm/Desktop/2018_python/test_blanks/'
## REVISED THIS TO BE A LARGER PATCH
mask = np.zeros([400,400], dtype=np.uint32)
im0 = Image.fromarray(mask)
im0.save(out_fname)
# Now, move both files to folder containing blank segmentations
shutil.move(src=out_fname, dst=out_dir)
shutil.move(src=image, dst=out_dir)
#################################### #################################### ####################################
####################################
run_location = os.getcwd().split('/')[-1:][0] search0 = run_location + '_InventoryMosaic.csv'
df1 = pd.read_csv(search0)
container_list = sorted(df1['container'].unique().tolist()) completed = [] """Loop through the containers to get global coordinate nomenclature""" for item in container_list: if item not in completed: completed.append(item) else: continue # First, get the Assembly Data file and process it... search1 = item + 'AssemblyData.txt' df_temp = df1.loc[df1['fname'] == search1].copy() # Ignore empty frames if len(df_temp) < 1: continue # Get first file path for container mosaic data assemble_fpath = df_temp['fpath'].values[0] # Run first function assembly_txt_conversion(assemble_fpath) # Next file path reorder_fpath = item + 'AssemblyConverted.csv' # Run next function reorder_assembly_data(reorder_fpath) # Final file path ingest_fpath = item + 'AssemblyModified.csv' # Run final function ingest_assembly_data(ingest_fpath) #""" Second, read-in the transformed data for each container, and loop through images """ search2 = item + 'AssemblyGlobal.csv' df2 = pd.read_csv(search2) ln = len(df2) bar = IncrementalBar('Processing:', max=ln) # Loop through the images associated with each datafile for index, row in df2.iterrows(): # For now, when testing, simply write out the file order (do not process files) # df2 contains: 'tile_id', 'original_path', 'new_global_id' tile_id = row.tile_id global_tile_id = row.new_global_id # Set up nomenclature for subarray output mdata = row.new_global_id.split('') metadata = ''.join(mdata[:-2]) x_coord = int(mdata[-1][1:]) y_coord = int(mdata[-2][1:]) # # Read-in the file, after setting nomenclature # img0 = imageio.imread(fname) # # Get dim # xdim = img0.shape[1] # ydim = img0.shape[0] # When testing, simply set the dim xdim = 2752 ydim = 2192 # Derive limits xlim = round(xdim/400) + 1 ylim = round(ydim/400) + 1 # Derive cut-off value (i.e. penultimate subarray) xnum = xlim - 1 ynum = ylim - 1 # Loop through range of Y values # docs for fnx -- range(start, stop, interval) for x in range(0, xlim, 1): # Define cropping rectangle # where (x0 = left, x1 = right; y0 = upper, y1 = lower) # Recall: Global coordinate embedded as nomenclature must reflect x|y_coord(s) # First option if x == 0: left = 0 Global_Left = left + x_coord right = 400 for y in range(0, ylim, 1): if y == 0: upper = 0 Global_Upper = upper + y_coord lower = 400 # Set coordinate values to being larger bc of global coordinate space GlobalX = 'x' + '{:0>6d}'.format(Global_Left) GlobalY = 'y' + '{:0>6d}'.format(Global_Upper) out_fname = metadata + '' + GlobalY + '' + GlobalX # Set up naming for output nd = {} nd['xdim'] = x nd['ydim'] = y nd['global_container'] = metadata nd['global_tile_id'] = global_tile_id nd['global_patch_id'] = out_fname new_df = pd.DataFrame().append(nd, ignore_index=True) output_path = metadata + '_GlobalCoordinates.csv' if os.path.exists(output_path): new_df.to_csv(output_path, mode='a', header=False, index=False) else: new_df.to_csv(output_path, header=True, index=False)
# WIP ......
#### Having successfully sketched out the subarrays to create and file to save as a record,
#### next step is to chunk the image itself into a deep array from a wide array (i.e. image)...
# First, load each array
t_arr = np.load(row.opath)
t2_arr = np.load(npy_outname)
# Then aggregate the arrays [y,x,z] with dstack (along dim [2] as tuple)
t_agg = np.dstack([t_arr, t2_arr])
# Begin conversion to NIFTI, rotate and then extract as image with spacing
vol0 = t_agg.T
vol1 = sitk.GetImageFromArray(vol0)
vol1.SetSpacing([1.0,1.0,40.0])
# Nomenclature
nifti_out = some_name_eh + '.nii.gz'
out_npath = os.path.join(cdir, nifti_out)
sitk.WriteImage(vol1, nifti_out)
# # Create subarray 1
# subarray = img0[upper:lower, left:right]
# Run object detection & move algorithm
# object_detection(subarray)
# # Alternative way, to write file first, and then run seg
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# Run object detection & move algorithm
# object_detection(out_fname)
elif (y > 0 and y < ynum):
upper = y*400
Global_Upper = upper + y_coord
lower = upper + 400
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 2
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
elif y == ynum:
upper = 1792
Global_Upper = upper + y_coord
lower = 2192
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 3
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
else:
continue
# Second option
elif (x > 0 and x < xnum):
left = x*400
Global_Left = left + x_coord
right = left + 400
for y in range(0, ylim, 1):
if y == 0:
upper = 0
Global_Upper = upper + y_coord
lower = 400
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 4
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
elif (y > 0 and y < ynum):
upper = y*400
Global_Upper = upper + y_coord
lower = upper + 400
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 5
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
elif y == ynum:
upper = 1792
Global_Upper = upper + y_coord
lower = 2192
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 6
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
else:
continue
# Third option
elif x == xnum:
left = 2352
Global_Left = left + x_coord
right = 2752
for y in range(0, ylim, 1):
if y == 0:
upper = 0
Global_Upper = upper + y_coord
lower = 400
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 7
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
elif (y > 0 and y < ynum):
upper = y*400
Global_Upper = upper + y_coord
lower = upper + 400
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 8
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
elif y == ynum:
upper = 1792
Global_Upper = upper + y_coord
lower = 2192
# Set coordinate values to being larger bc of global coordinate space
GlobalX = 'x' + '{:0>6d}'.format(Global_Left)
GlobalY = 'y' + '{:0>6d}'.format(Global_Upper)
out_fname = metadata + '_' + GlobalY + '_' + GlobalX
# Set up naming for output
nd = {}
nd['xdim'] = x
nd['ydim'] = y
nd['global_container'] = metadata
nd['global_tile_id'] = global_tile_id
nd['global_patch_id'] = out_fname
new_df = pd.DataFrame().append(nd, ignore_index=True)
output_path = metadata + '_GlobalCoordinates.csv'
if os.path.exists(output_path):
new_df.to_csv(output_path, mode='a', header=False, index=False)
else:
new_df.to_csv(output_path, header=True, index=False)
# # Create subarray 9
# subarray = img0[upper:lower, left:right]
# im1 = Image.fromarray(subarray)
# im1.save(out_fname)
# # Run object detection & move algorithm
# object_detection(out_fname)
else:
continue
else:
continue
bar.next()
bar.finish()