def for_dahui():
labels = read_excel()
file_path = "D:\\2018-04-24\\tiff"
file_names = scan_files(file_path, postfix=".tif")
print(len(labels), len(file_names))
for i in range(len(labels)):
os.rename(file_names[i], os.path.join(file_path, str(labels[i]) + ".tif"))
def get_crops(tif_file, image_path, new_image_path, size):
name = tif_file[-23:-4]
image_names = scan_files(image_path)
image_xy = []
for image_name in image_names:
xy_part = image_name[:-4].rsplit("_", 2)
x = int(xy_part[-2])
y = int(xy_part[-1])
image_xy.append([x, y])
"""
# when size is larger than original image size
pad = int(size/4)
slide = openslide.OpenSlide(tif_file)
i = 0
for xy in image_xy:
x = max(xy[0] - pad, 0)
y = max(xy[1] - pad, 0)
cell = slide.read_region((x, y), 0, (size, size))
cell = cell.convert("RGB")
scipy.misc.imsave(new_image_path + "/" + name + "_" + str(i).zfill(6) + ".jpg", cell)
i += 1
print("total images in " + name + ": " + str(i))
slide.close()
"""
# when size is smaller than original image size
slide = openslide.OpenSlide(tif_file)
i = 0
for xy in image_xy:
x = xy[0] + 16
y = xy[1] + 16
cell = slide.read_region((x, y), 0, (size, size))
cell = cell.convert("RGB")
scipy.misc.imsave(
new_image_path + "/" + name + "_" + str(i).zfill(6) + ".jpg", cell)
i += 1
print("total images in " + name + ": " + str(i))
slide.close()
def batch_delete_dots(file_path):
file_list = scan_files(file_path, postfix=".xml")
for file in file_list:
delete_dots(file)
parent.removeChild(annotation)
with open(file, 'w') as newfile:
DOMTree.writexml(newfile)
def batch_delete_dots(file_path):
file_list = scan_files(file_path, postfix=".xml")
for file in file_list:
delete_dots(file)
if __name__ == "__main__":
# the file path that contains the images to delete or change
file_path = "D:\\data0\\2018-05-05-AGC+VIRUS\\2018-05-05-VIRUS-shenhe\\shandiaode"
files_list = scan_files(file_path, postfix=".jpg")
# the file path that contains xml files
target_path = "D:\\"
files_dic = parse_path(files_list, len(file_path))
# delete annotations
batch_delete(files_dic, target_path)
# # change annotations
# batch_change(files_dic, target_path, "#aa0000")
# # delete single dots in xml
# file_path = "D:\\data7\\15_special"
# batch_delete_dots(file_path)
def add_postfix():
file_path = "D:\\2018-04-24\\tiff"
file_names = scan_files(file_path, postfix=".jpg")
for file_name in file_names:
os.rename(file_name, file_name[:-4] + "-label.jpg")
from random import randint
from utils.scan_files import scan_files
def cellSampling(files_list, n, size, save_path):
for file in files_list:
filename = os.path.splitext(file)[0].rsplit("\\", 1)[1]
try:
slide = openslide.OpenSlide(file)
[size_x, size_y] = slide.dimensions
for i in range(n):
x = randint(size, size_x - size)
y = randint(size, size_y - size)
cell = slide.read_region((x, y), 0, (size, size))
cell = cell.convert("RGB")
scipy.misc.imsave(
save_path + "\\" + filename + "_" + str(i).zfill(3) +
".jpg", cell)
slide.close()
except:
print(filename + " cannot be processed")
if __name__ == "__main__":
file_path = "E:\\2018-04-08-normal\\tiff"
files_list = scan_files(file_path, postfix=".tif")
save_path = "D:\\data0\\2018-04-08-normal-crops"
n = 10 # number of crops per slide
size = 4096 # size of each crop
cellSampling(files_list, n, size, save_path)
def replace():
files_list = scan_files(tif_path, postfix=".xml")
total = 0
replaced = 0
for xml_file in files_list:
# from .xml filename, get .til filename
filename = os.path.splitext(xml_file)[0]
name = filename.rsplit("\\", 1)[1]
# open .tif file
tif_file = filename + ".tif"
slide = openslide.OpenSlide(tif_file)
# open .xml file
DOMTree = xml.dom.minidom.parse(xml_file)
collection = DOMTree.documentElement
annotations = collection.getElementsByTagName("Annotation")
for annotation in annotations:
coordinates = annotation.getElementsByTagName("Coordinate")
# read (x, y) coordinates
x_coords = []
y_coords = []
for coordinate in coordinates:
x_coords.append(float(coordinate.getAttribute("X")))
y_coords.append(float(coordinate.getAttribute("Y")))
if len(x_coords) < 3:
continue
x_min = min(x_coords)
x_max = max(x_coords)
y_min = min(y_coords)
y_max = max(y_coords)
x_size = int(x_max - x_min)
y_size = int(y_max - y_min)
x_min = int(x_min)
y_min = int(y_min)
x = int((x_min + x_max) / 2 - des_image_size / 2)
y = int((y_min + y_max) / 2 - des_image_size / 2)
# marked region is small, can be replaced
if annotation.getAttribute(
"Color"
) in colors and x_size < des_image_size and y_size < des_image_size:
print("marked region size: " + str(x_size) + ", " +
str(y_size))
cell = slide.read_region((x, y), 0,
(des_image_size, des_image_size))
cell = cv2.cvtColor(np.asarray(cell), cv2.COLOR_RGBA2BGR)
cell[y_min - y:y_min - y + y_size,
x_min - x:x_min - x + x_size] = get_random_image(
x_size, y_size)
cv2.imwrite(
des_image_path1 + "\\" + name + "_" +
annotation.getAttribute("Name") +
"_replaced1_tri_0514.jpg", cell)
cell[y_min - y:y_min - y + y_size,
x_min - x:x_min - x + x_size] = get_random_image(
x_size, y_size)
cv2.imwrite(
des_image_path2 + "\\" + name + "_" +
annotation.getAttribute("Name") +
"_replaced2_tri_0514.jpg", cell)
cell[y_min - y:y_min - y + y_size,
x_min - x:x_min - x + x_size] = get_random_image(
x_size, y_size)
cv2.imwrite(
des_image_path3 + "\\" + name + "_" +
annotation.getAttribute("Name") +
"_replaced3_tri_0514.jpg", cell)
replaced += 1
total += 1
slide.close()
print("# total images: " + str(total))
print("# replaced: " + str(replaced))
# # test
# dest = cv2.imread("../res/temp/minions.jpg")
# height, width, channels = dest.shape
# print(height, width, channels)
# print(dest.dtype)
#
# src = cv2.imread("../res/temp/minions.jpg")
# dest[0:int(height/2), 0:int(width/2)] = src[int(height/2):height, int(width/2):width]
# status = cv2.imwrite("../res/temp/minions_1.jpg", dest)
# if not status:
# print("error")
src_image_path = "F:\\data0\\2018-04-08-normal-crops"
src_image_size = 4096
src_images = scan_files(src_image_path)
tif_path = "F:\\data8\\11_TRI"
des_image_path1 = "F:\\0514-data1-8-replace2normal-size256\\replace2normal_1"
des_image_path2 = "F:\\0514-data1-8-replace2normal-size256\\replace2normal_2"
des_image_path3 = "F:\\0514-data1-8-replace2normal-size256\\replace2normal_3"
des_image_size = 256
colors = {
"#000000": 0,
"#aa0000": 0,
"#aa007f": 0,
"#aa00ff": 0,
"#ff0000": 0,
"#005500": 0,
"#00557f": 0,
if annotation.getAttribute("Color") in colorCounts:
colorCounts[annotation.getAttribute("Color")] += 1
n = 0
for color, count in colorCounts.items():
if count > 0:
n += 1
if len(categoryCounts) < 10:
categoryCounts.append((file, n))
if len(categoryCounts) == 10:
sorted(categoryCounts, key=lambda x: x[1], reverse=True)
else:
i = 0
for filename, count in categoryCounts:
if count > n:
i += 1
categoryCounts.insert(i, (file, n))
del categoryCounts[-1]
for color in colorCounts:
colorCounts[color] = 0
file_path = "E:\\data7"
files_list = scan_files(file_path, postfix=".xml")
print("# files: " + str(len(files_list)))
select_most(files_list)
for filename, count in categoryCounts:
print(filename, count)