def train_seg_mem(self, model, epoch, batch):
maxImageIndex = 100
print(self)
try:
xs = self.validation_data[0]
ys = self.validation_data[1]
except:
xs = K.eval(self.model.inputs[0])
ys = self.model.targets[0]
ys_pred = self.model.predict(xs)
image_size = self.flag.image_size
#print("Shape xs: " + str(xs.shape) )
#print("Shape ys: " + str(ys.shape) )
#print("Shape pred ys: " + str(ys_pred.shape) )
# Save the predicted output prob, image, and original mask
for imageIndex in range(min(maxImageIndex, xs.shape[0])):
x = xs[imageIndex, ...]
y = ys[imageIndex, ...]
y_pred = ys_pred[imageIndex, ...]
#if multi-class y then merge them by giving different values: e.g. cx = 2, cb = 1, bg = 0
if y.shape[2] > 1:
y = mergeLabelImage(y).astype(np.float32)
y_pred = mergeLabelImage(y_pred).astype(np.float32)
y /= y.max()
y_pred /= y_pred.max()
#print( "x min: " + str(x.min()) +", x max: " + str(x.max()) )
#print( "y min: " + str(y.min()) +", y max: " + str(y.max()) )
#print( "y_pred min: " + str(y_pred.min()) +", y_pred max: " + str(y_pred.max()) )
imageName = str(imageIndex)
imgMask = (y_pred * 255).astype(np.uint8)
imgMaskOri = (y * 255).astype(np.uint8)
#print( "y min: " + str(y.min()) +", y max: " + str(y.max()) )
img = (x).astype(np.uint8)
imgColor = cv2.cvtColor(x, cv2.COLOR_GRAY2BGR)
imgMaskColor = cv2.applyColorMap(imgMask, cv2.COLORMAP_JET)
imgMaskOriColor = cv2.applyColorMap(imgMaskOri, cv2.COLORMAP_JET)
output_path_combined = os.path.join(
self.flag.output_images_dir,
'combined_index-%04d_batch-%04d_epoch-%04d.png' %
(imageIndex, batch, epoch))
image_combined = np.concatenate(
(imgColor, imgMaskColor, imgMaskOriColor), axis=1)
cv2.imwrite(output_path_combined, image_combined)
writeImagePathToCsv(self.flag.images_csv_path, imageIndex, batch,
epoch, output_path_combined)
def saveLabelImage( mask, output_path ):
y = mergeLabelImage(mask).astype( np.float32 )
y = y + 0.9
y /= y.max()
maskOri = (y*255).astype( np.uint8 )
maskOriColor = cv2.applyColorMap( maskOri, cv2.COLORMAP_HSV )
maskColor = np.flip(maskOriColor, 2)
#maskColor = maskOriColor
cv2.imwrite( output_path, maskColor )
def makeOverlayMaskDifference( img, mask_ref, mask, output_path ):
y = mergeLabelImage(mask).astype( np.float32 )
y_ref = mergeLabelImage(mask_ref).astype( np.float32 )
diff = np.zeros( y.shape )
diff[ y != y_ref ] = 1
diff /= diff.max()
diff = (diff*255).astype( np.uint8 )
diffColor = cv2.applyColorMap( diff, cv2.COLORMAP_JET )
scaleIntensity = 500
img = img.astype( np.float32 )
img /= img.max()
img = np.fmin( 255.0 * np.ones(img.shape, np.float32 ), scaleIntensity * img )
img = img.astype( np.uint8 )
diffColor = np.flip( diffColor, 2)
imgColor = cv2.cvtColor( img, cv2.COLOR_GRAY2BGR )
imgOverlay = cv2.addWeighted(imgColor, 0.9, diffColor, 0.4, 0.0)
cv2.imwrite( output_path, imgOverlay )
def makeOverlay( img, mask, output_path ):
y = mergeLabelImage(mask).astype( np.float32 )
y = y + 0.9
y /= y.max()
maskOri = (y*255).astype( np.uint8 )
maskOriColor = cv2.applyColorMap( maskOri, cv2.COLORMAP_HSV )
scaleIntensity = 500
img = img.astype( np.float32 )
img /= img.max()
img = np.fmin( 255.0 * np.ones(img.shape, np.float32 ), scaleIntensity * img )
img = img.astype( np.uint8 )
maskColor = np.flip(maskOriColor, 2)
imgColor = cv2.cvtColor( img, cv2.COLOR_GRAY2BGR )
imgOverlay = cv2.addWeighted(imgColor, 0.9, maskColor, 0.4, 0.0)
cv2.imwrite( output_path, imgOverlay )
def showData(flag):
img_rows = flag.image_size
img_cols = flag.image_size
print('-' * 30)
print('Loading train data...')
print('-' * 30)
imgs_train, imgs_test, imgs_mask_train_all, imgs_mask_test_all, imgs_id_test, imgs_id_train = loadDataParams(
flag)
images = imgs_train
images = preprocess(images, img_rows, img_cols)
masks_all = imgs_mask_train_all
print('-' * 30)
print('Fuse masks to single multi-label image')
print('-' * 30)
regionList = flag.region_list
nClasses = len(regionList) + 1
s = masks_all[regionList[0]].shape
nImages = s[0]
masks = np.zeros((nImages, img_rows, img_cols, nClasses))
masks[:, :, :, 0] = np.ones((nImages, img_rows, img_cols))
for regionIndex in range(len(regionList)):
# put the mask layer of the region index to 1
masks_region = masks_all[regionList[regionIndex]]
masks_region = preprocess(masks_region, img_rows, img_cols)
temp = masks_region[:, :, :, 0]
masks[:, :, :, regionIndex + 1] = temp
# and the background layer to 0, for every region pixel
masks[:, :, :, 0] = (1 - temp) * masks[:, :, :, 0]
print('-' * 30)
print('Save masks as images')
print('-' * 30)
scaleIntensity = 1
for imageIndex in range(nImages):
fileName = 'mask_sample_%d.tif' % (imageIndex)
filePath = os.path.join(flag.output_masks_feed_dir, fileName)
maskData = np.transpose(masks[imageIndex, ...], (2, 0, 1))
maskData = np.expand_dims(maskData, axis=3)
writeData(filePath, maskData)
y = masks[imageIndex, ...]
y = mergeLabelImage(y).astype(np.float32)
y /= y.max()
x = images[imageIndex, ...].astype(np.float32)
x = np.fmin(255.0 * np.ones(x.shape, np.float32), scaleIntensity * x)
x = x.astype(np.uint8)
x = np.squeeze(x)
print(x.shape)
#x /= x.max()
#img = (x*255).astype( np.uint8 )
imgColor = cv2.cvtColor(x, cv2.COLOR_GRAY2BGR)
imgMaskOri = (y * 255).astype(np.uint8)
imgMaskOriColor = cv2.applyColorMap(imgMaskOri, cv2.COLORMAP_JET)
imgOverlay = cv2.addWeighted(imgColor, 0.9, imgMaskOriColor, 0.4, 0.0)
output_path_mask_ori = os.path.join(
flag.output_images_dir, 'mask_index-%04d.png' % (imageIndex))
output_path_overlay = os.path.join(
flag.output_plots_dir, 'overlay_index-%04d.png' % (imageIndex))
output_path_image = os.path.join(flag.output_plots_dir,
'image_index-%04d.png' % (imageIndex))
cv2.imwrite(output_path_image, imgColor)
cv2.imwrite(output_path_mask_ori, imgMaskOriColor)
cv2.imwrite(output_path_overlay, imgOverlay)
print('-' * 30)
print('Save (preprocessed) images as images')
print('-' * 30)
for imageIndex in range(nImages):
fileName = 'image_sample_%d.tif' % (imageIndex)
filePath = os.path.join(flag.output_images_feed_dir, fileName)
imageData = scaleIntensity * np.transpose(images[imageIndex, ...],
(2, 0, 1))
imageData = np.expand_dims(imageData, axis=3)
writeData(filePath, imageData)