def BrowseFolder(self):
path, _ = QFileDialog.getOpenFileName(
self, "open", "/home/",
"All Files (*);; Image Files (*.png *.tif *.jpg *.ser *.dm3)")
self.imagePath_content = self.imagePath_content if not path else path
if self.imagePath_content:
self.imagePath.setText(self.imagePath_content)
file_name = os.path.basename(self.imagePath_content)
_, suffix = os.path.splitext(file_name)
# file_name = self.imagePath_content.split('/')[-1]
# suffix = '.' + file_name.split('.')[-1]
if suffix == '.ser':
from file_readers.ser_lib.serReader import serReader
ser_data = serReader(self.imagePath_content)
ser_array = np.array(ser_data['imageData'], dtype='float64')
self.imarray_original = ser_array
ser_array = (map01(ser_array) * 255).astype('uint8')
self.ori_image = Image.fromarray(ser_array, 'L')
elif suffix == '.dm3':
from file_readers import dm3_lib as dm3
data = dm3.DM3(self.imagePath_content).imagedata
self.imarray_original = np.array(data)
data = np.array(data, dtype='float64')
data = (map01(data) * 255).astype('uint8')
self.ori_image = Image.fromarray(data, mode='L')
elif suffix == '.tif':
im = Image.open(self.imagePath_content).convert('L')
self.imarray_original = np.array(im, dtype='float64')
self.ori_image = Image.fromarray(
(map01(self.imarray_original) * 255).astype('uint8'),
mode='L')
else:
self.ori_image = Image.open(
self.imagePath_content).convert('L')
self.imarray_original = np.array(self.ori_image)
self.width, self.height = self.ori_image.size
pix_image = PIL2Pixmap(self.ori_image)
pix_image.scaled(self.ori.size(), QtCore.Qt.KeepAspectRatio)
self.ori.setPixmap(pix_image)
self.ori.show()
self.ori_content = self.ori_image
def __load_model(self):
# if not self.ori_image:
# raise Exception("No image is selected.")
self.cuda = self.use_cuda.isChecked()
model_path = os.path.join(self.__curdir, self.__models[self.model_name])
recon_path = self.recon_name
self.ori_content = self.ori_image
dirname = os.path.dirname(self.imagePath_content)
dataname = self.listData.currentText()
result_d, result = load_model(dataname, self.matdata, self.matangles, self.matsize, model_path,\
recon_path, self.cuda, self.set_iter.value())
savename = os.path.join(dirname, dataname)
if not os.path.exists(savename):
os.mkdir(savename)
scio.savemat(savename + "/angles.mat", {"angles":self.matangles})
if self.model_name=="Deepfusion(3d)" or self.model_name=="Deepfusion(3dGAN)":
scio.savemat(savename + "/" + "%s_%s.mat" % (dataname,self.model_name), result_d)
else:
scio.savemat(savename + "/" + "%s_%s.mat" % (dataname,recon_path), result)
scio.savemat(savename + "/" + "%s_%s_%s.mat" % (dataname, self.model_name, recon_path), result_d)
print("Save mat successfully!")
self.model_out = result_d
self.recon_out = result
dataname = self.listData.currentText()
idx = self.img_num.value()
print(dataname, idx)
model_out = map01(self.model_out[dataname][idx, :, :])
model_out = (model_out * 255 / np.max(model_out)).astype('uint8')
model_out = Image.fromarray((model_out), mode='L')
model_out = PIL2Pixmap(model_out)
model_out.scaled(self.model_output.size(), QtCore.Qt.KeepAspectRatio)
self.model_output.setPixmap(model_out)
self.model_output.show()
recon_out = map01(self.recon_out[dataname][idx, :, :])
recon_out = (recon_out * 255 / np.max(recon_out)).astype('uint8')
recon_out = Image.fromarray((recon_out), mode='L')
recon_out = PIL2Pixmap(recon_out)
recon_out.scaled(self.model_output.size(), QtCore.Qt.KeepAspectRatio)
self.recon_output.setPixmap(recon_out)
self.recon_output.show()
def ResultShow(self):
dataname = self.listData.currentText()
idx = self.img_num.value()
# print(self.img_num.value())
self.img_num_show.setValue(self.img_num.value())
# print(dataname, idx)
model_out = map01(self.model_out[dataname][idx,:,:])
model_out = (model_out * 255 / np.max(model_out)).astype('uint8')
model_out = Image.fromarray((model_out), mode='L')
model_out = PIL2Pixmap(model_out)
model_out.scaled(self.model_output.size(), QtCore.Qt.KeepAspectRatio)
self.model_output.setPixmap(model_out)
self.model_output.show()
recon_out = map01(self.recon_out[dataname][idx,:,:])
recon_out = (recon_out * 255 / np.max(recon_out)).astype('uint8')
recon_out = Image.fromarray((recon_out), mode='L')
recon_out = PIL2Pixmap(recon_out)
recon_out.scaled(self.model_output.size(), QtCore.Qt.KeepAspectRatio)
self.recon_output.setPixmap(recon_out)
self.recon_output.show()
def __load_model(self):
if not self.ori_image:
raise Exception("No image is selected.")
self.cuda = self.use_cuda.isChecked()
model_path = os.path.join(self.__curdir,
self.__models[self.model_name])
if self.change_size.currentText() == 'Down sample by 2':
self.width, self.height = self.ori_image.size
self.ori_content = self.ori_image.resize(
(self.width // 2, self.height // 2), Image.BILINEAR)
elif self.change_size.currentText() == 'Up sample by 2':
self.width, self.height = self.ori_image.size
self.ori_content = self.ori_image.resize(
(self.width * 2, self.height * 2), Image.BICUBIC)
elif self.change_size.currentText() == 'Down sample by 3':
self.width, self.height = self.ori_image.size
self.ori_content = self.ori_image.resize(
(self.width // 3, self.height // 3), Image.BILINEAR)
elif self.change_size.currentText() == 'Up sample by 3':
self.width, self.height = self.ori_image.size
self.ori_content = self.ori_image.resize(
(self.width * 3, self.height * 3), Image.BICUBIC)
elif self.change_size.currentText() == 'Down sample by 4':
self.width, self.height = self.ori_image.size
self.ori_content = self.ori_image.resize(
(self.width // 4, self.height // 4), Image.BILINEAR)
elif self.change_size.currentText() == 'Up sample by 4':
self.width, self.height = self.ori_image.size
self.ori_content = self.ori_image.resize(
(self.width * 4, self.height * 4), Image.BICUBIC)
else:
self.ori_content = self.ori_image
pix_image = PIL2Pixmap(self.ori_content)
pix_image.scaled(self.ori.size(), QtCore.Qt.KeepAspectRatio)
self.ori.setPixmap(pix_image)
self.ori.show()
self.width, self.height = self.ori_content.size
if self.split.isChecked():
if self.height > 1024 and self.height < 2000:
blk_row = 2
else:
if self.height > 2000:
blk_row = 4
else:
blk_row = 1
if self.width > 1024 and self.width < 2000:
blk_col = 2
else:
if self.width > 2000:
blk_col = 4
else:
blk_col = 1
else:
blk_col = 1
blk_row = 1
self.result = np.zeros((self.height, self.width)) - 100
for r in range(0, blk_row):
for c in range(0, blk_col):
inner_blk, outer_blk = GetIndexRangeOfBlk(
self.height,
self.width,
blk_row,
blk_col,
r,
c,
over_lap=int(self.width * 0.01))
temp_image = self.ori_content.crop(
(outer_blk[0], outer_blk[1], outer_blk[2], outer_blk[3]))
temp_result = load_model(model_path, temp_image, self.cuda,
self.set_iter.value())
# temp_result = map01(temp_result)
self.result[outer_blk[1]:outer_blk[3],
outer_blk[0]:outer_blk[2]] = np.maximum(
temp_result,
self.result[outer_blk[1]:outer_blk[3],
outer_blk[0]:outer_blk[2]])
self.model_output_content = map01(self.result)
self.model_output_content = (
self.model_output_content * 255 /
np.max(self.model_output_content)).astype('uint8')
self.output_image = Image.fromarray((self.model_output_content),
mode='L')
pix_image = PIL2Pixmap(self.output_image)
pix_image.scaled(self.model_output.size(), QtCore.Qt.KeepAspectRatio)
self.model_output.setPixmap(pix_image)
self.model_output.show()
del temp_image
del temp_result