policy = tf.keras.mixed_precision.experimental.Policy('mixed_float16')
tf.keras.mixed_precision.experimental.set_policy(policy)
#################################### Load Data #####################################
folder = '../BCDU-Net/Lung Segmentation/processed_data/'
te_data = np.load(folder + 'data_test.npy')
FOV = np.load(folder + 'FOV_te.npy')
te_mask = np.load(folder + 'mask_test.npy')
te_data = np.expand_dims(te_data, axis=3)
print('Dataset loaded')
#te_data2 = dataset_normalized(te_data)
te_data2 = te_data / 255.
model = M.BCDU_net_D3(input_size=(512, 512, 1))
model.summary()
model.load_weights('../BCDU-Net/Lung Segmentation/weight_lung')
predictions = model.predict(te_data2, batch_size=1, verbose=1)
# Post-processing
predictions = np.squeeze(predictions)
predictions = np.where(predictions > 0.5, 1, 0)
plt.imshow(np.uint8(np.squeeze(predictions[0])))
plt.savefig("test.png")
exit()
Estimated_lung = np.where((FOV - predictions) > 0.5, 1, 0)
print(Estimated_lung)
print(Estimated_lung.shape)
name_experiment = 'test'
#training settings
batch_size = 8
#################################### Load Data #####################################3
patches_imgs_train = np.load('patches_imgs_train.npy')
patches_masks_train = np.load('patches_masks_train.npy')
patches_imgs_train = np.einsum('klij->kijl', patches_imgs_train)
patches_masks_train = np.einsum('klij->kijl', patches_masks_train)
print('Patch extracted')
#model = M.unet2_segment(input_size = (64,64,1))
model = M.BCDU_net_D3(input_size=(64, 64, 1))
model.summary()
print('Training')
nb_epoch = 50
mcp_save = ModelCheckpoint('weight_lstm.hdf5',
save_best_only=True,
monitor='val_loss',
mode='min')
reduce_lr_loss = ReduceLROnPlateau(monitor='val_loss',
factor=0.1,
patience=7,
verbose=1,
epsilon=1e-4,
from sklearn.metrics import roc_auc_score
from sklearn.metrics import confusion_matrix
from sklearn.metrics import precision_recall_curve
from sklearn.metrics import jaccard_score
from sklearn.metrics import f1_score
from tensorflow import keras
import skimage
from skimage.io import imread, imsave
from skimage.transform import resize
from skimage.color import gray2rgb
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
dataset_add = "/home/users/DATASETS/MapBiomas_SAR/"
model = M.BCDU_net_D3(input_size = (32,32,10))
model.summary()
model.load_weights('weights')
dataset_mean_ind = (-15.34541179, -8.87553847)
dataset_std_ind = (1.53520544, 1.45585154)
file = open("data_split/test.txt", "r")
te_list = [line.rstrip('\n') for line in file]
file.close()
for idx in range(len(te_list)):
print(idx)
path = te_list[idx]
def testing(channels, height, width, img_dir, borderMasks_dir, path_data = "./dataset_training/"):
img = np.empty((height, width, channels))
border_mask = np.empty((height, width))
if not os.path.exists(path_data):
os.makedirs(path_data)
#with path, subdirs, file in os.walk(self.img_dir):
img = Image.open(img_dir)
b_mask = Image.open(borderMasks_dir)
#border_mask = np.reshape(border_mask,(1,height,width))
assert(np.max(border_masks)==255)
assert(np.min(border_masks)==0)
"""
img = np.transpose(img,(0,3,1,2))
assert(img.shape == (channels,height,width))
border_mask = np.reshape(border_masks,(1,height,width))
assert(border_mask.shape == (1,height,width))
"""
print ("saving train datasets")
write_hdf5(img, path_data + "img.hdf5")
write_hdf5(border_mask, path_data + "borderMask.hdf5")
#original test images (for FOV selection)
DRIVE_test_imgs_original = path_data + 'img.hdf5'
test_imgs_orig = load_hdf5(DRIVE_test_imgs_original)
full_img_height = test_imgs_orig.shape[2]
full_img_width = test_imgs_orig.shape[3]
#the border masks provided by the DRIVE
DRIVE_test_border_masks = path_data + 'borderMask.hdf5'
test_border_masks = load_hdf5(DRIVE_test_border_masks)
# dimension of the patches
patch_height = 64
patch_width = 64
#the stride in case output with average
stride_height = 5
stride_width = 5
assert (stride_height < patch_height and stride_width < patch_width)
#model name
name_experiment = 'test'
path_experiment = './' + name_experiment +'/'
#N full images to be predicted
Imgs_to_test = 2
#Grouping of the predicted images
N_visual = 1
#====== average mode ===========
average_mode = True
#============ Load the data and divide in patches
patches_imgs_test = None
new_height = None
new_width = None
masks_test = None
patches_masks_test = None
if average_mode == True:
patches_imgs_test, new_height, new_width, masks_test = get_data_testing_overlap(
DRIVE_test_imgs_original = DRIVE_test_imgs_original, #original
DRIVE_test_groudTruth = path_data + 'DRIVE_dataset_groundTruth_test.hdf5', #masks
Imgs_to_test = 20,
patch_height = patch_height,
patch_width = patch_width,
stride_height = stride_height,
stride_width = stride_width
)
else:
patches_imgs_test, patches_masks_test = get_data_testing(
DRIVE_test_imgs_original = DRIVE_test_imgs_original, #original
DRIVE_test_groudTruth = path_data + 'DRIVE_dataset_groundTruth_test.hdf5', #masks
Imgs_to_test = 20,
patch_height = patch_height,
patch_width = patch_width,
)
#================ Run the prediction of the patches ==================================
best_last = 'best'
patches_imgs_test = np.einsum('klij->kijl', patches_imgs_test)
model = M.BCDU_net_D3(input_size = (64,64,1))
model.summary()
model.load_weights('weight_lstm.hdf5')
predictions = model.predict(patches_imgs_test, batch_size=16, verbose=1)
predictions = np.einsum('kijl->klij', predictions)
print(patches_imgs_test.shape)
pred_patches = predictions
print ("predicted images size :")
print (predictions.shape)
#===== Convert the prediction arrays in corresponding images
#========== Elaborate and visualize the predicted images ====================
pred_imgs = None
orig_imgs = None
gtruth_masks = None
if average_mode == True:
pred_imgs = recompone_overlap(pred_patches, new_height, new_width, stride_height, stride_width)# predictions
orig_imgs = my_PreProc(test_imgs_orig[0:pred_imgs.shape[0],:,:,:]) #originals
gtruth_masks = masks_test #ground truth masks
else:
pred_imgs = recompone(pred_patches,13,12) # predictions
orig_imgs = recompone(patches_imgs_test,13,12) # originals
gtruth_masks = recompone(patches_masks_test,13,12) #masks
# apply the DRIVE masks on the repdictions #set everything outside the FOV to zero!!
print('killing border')
kill_border(pred_imgs, test_border_masks) #DRIVE MASK #only for visualization
## back to original dimensions
orig_imgs = orig_imgs[:,:,0:full_img_height,0:full_img_width]
pred_imgs = pred_imgs[:,:,0:full_img_height,0:full_img_width]
gtruth_masks = gtruth_masks[:,:,0:full_img_height,0:full_img_width]
np.save('pred_imgs',pred_imgs)
print ("Orig imgs shape: " +str(orig_imgs.shape))
print ("pred imgs shape: " +str(pred_imgs.shape))
print ("Gtruth imgs shape: " +str(gtruth_masks.shape))
np.save('resutls', pred_imgs)
np.save('origin', gtruth_masks)
assert (orig_imgs.shape[0]==pred_imgs.shape[0] and orig_imgs.shape[0]==gtruth_masks.shape[0])
N_predicted = orig_imgs.shape[0]
group = N_visual
assert (N_predicted%group==0)
#====== Evaluate the results
print ("\n\n======== Evaluate the results =======================")
#predictions only inside the FOV
y_scores, y_true = pred_only_FOV(pred_imgs,gtruth_masks, test_border_masks) #returns data only inside the FOV
print(y_scores.shape)
print ("Calculating results only inside the FOV:")
print ("y scores pixels: " +str(y_scores.shape[0]) +" (radius 270: 270*270*3.14==228906), including background around retina: " +str(pred_imgs.shape[0]*pred_imgs.shape[2]*pred_imgs.shape[3]) +" (584*565==329960)")
print ("y true pixels: " +str(y_true.shape[0]) +" (radius 270: 270*270*3.14==228906), including background around retina: " +str(gtruth_masks.shape[2]*gtruth_masks.shape[3]*gtruth_masks.shape[0])+" (584*565==329960)")
#Area under the ROC curve
fpr, tpr, thresholds = roc_curve((y_true), y_scores)
AUC_ROC = roc_auc_score(y_true, y_scores)
# test_integral = np.trapz(tpr,fpr) #trapz is numpy integration
print ("\nArea under the ROC curve: " +str(AUC_ROC))
roc_curve =plt.figure()
plt.plot(fpr,tpr,'-',label='Area Under the Curve (AUC = %0.4f)' % AUC_ROC)
plt.title('ROC curve')
plt.xlabel("FPR (False Positive Rate)")
plt.ylabel("TPR (True Positive Rate)")
plt.legend(loc="lower right")
plt.savefig(path_experiment+"ROC.png")
#Precision-recall curve
precision, recall, thresholds = precision_recall_curve(y_true, y_scores)
precision = np.fliplr([precision])[0] #so the array is increasing (you won't get negative AUC)
recall = np.fliplr([recall])[0] #so the array is increasing (you won't get negative AUC)
AUC_prec_rec = np.trapz(precision,recall)
print ("\nArea under Precision-Recall curve: " +str(AUC_prec_rec))
prec_rec_curve = plt.figure()
plt.plot(recall,precision,'-',label='Area Under the Curve (AUC = %0.4f)' % AUC_prec_rec)
plt.title('Precision - Recall curve')
plt.xlabel("Recall")
plt.ylabel("Precision")
plt.legend(loc="lower right")
plt.savefig(path_experiment+"Precision_recall.png")
#Confusion matrix
threshold_confusion = 0.5
print ("\nConfusion matrix: Custom threshold (for positive) of " +str(threshold_confusion))
y_pred = np.empty((y_scores.shape[0]))
for i in range(y_scores.shape[0]):
if y_scores[i]>=threshold_confusion:
y_pred[i]=1
else:
y_pred[i]=0
confusion = confusion_matrix(y_true, y_pred)
print (confusion)
accuracy = 0
if float(np.sum(confusion))!=0:
accuracy = float(confusion[0,0]+confusion[1,1])/float(np.sum(confusion))
print ("Global Accuracy: " +str(accuracy))
specificity = 0
if float(confusion[0,0]+confusion[0,1])!=0:
specificity = float(confusion[0,0])/float(confusion[0,0]+confusion[0,1])
print ("Specificity: " +str(specificity))
sensitivity = 0
if float(confusion[1,1]+confusion[1,0])!=0:
sensitivity = float(confusion[1,1])/float(confusion[1,1]+confusion[1,0])
print ("Sensitivity: " +str(sensitivity))
precision = 0
if float(confusion[1,1]+confusion[0,1])!=0:
precision = float(confusion[1,1])/float(confusion[1,1]+confusion[0,1])
print ("Precision: " +str(precision))
#Jaccard similarity index
jaccard_index = jaccard_similarity_score(y_true, y_pred, normalize=True)
print ("\nJaccard similarity score: " +str(jaccard_index))
#F1 score
F1_score = f1_score(y_true, y_pred, labels=None, average='binary', sample_weight=None)
print ("\nF1 score (F-measure): " +str(F1_score))
#Save the results
file_perf = open(path_experiment+'performances.txt', 'w')
file_perf.write("Area under the ROC curve: "+str(AUC_ROC)
+ "\nArea under Precision-Recall curve: " +str(AUC_prec_rec)
+ "\nJaccard similarity score: " +str(jaccard_index)
+ "\nF1 score (F-measure): " +str(F1_score)
+"\n\nConfusion matrix:"
+str(confusion)
+"\nACCURACY: " +str(accuracy)
+"\nSENSITIVITY: " +str(sensitivity)
+"\nSPECIFICITY: " +str(specificity)
+"\nPRECISION: " +str(precision)
)
file_perf.close()
# Visualize
fig,ax = plt.subplots(10,3,figsize=[15,15])
for idx in range(10):
ax[idx, 0].imshow(np.uint8(np.squeeze((orig_imgs[idx]))))
ax[idx, 1].imshow(np.squeeze(gtruth_masks[idx]), cmap='gray')
ax[idx, 2].imshow(np.squeeze(pred_imgs[idx]), cmap='gray')
plt.savefig(path_experiment+'sample_results.png')
patch_h = 128
patch_w = 128
stride_h = 5
stride_w = 5
Batch_size = 8
Estimated_masks = []
True_masks = []
## Get the dataset for test
Dataset_add = './DIBCO/'
Te_d, Te_m = UT.get_test_data(Dataset_add, 2016)
## Model
model = M.BCDU_net_D3(input_size = (128, 128,1))
model.load_weights('weight_text.hdf5')
y_scores = []
y_true = []
## Mask estimation using batches
for idx in range(len(Te_d)):
IMG = Te_d[idx]
MSK = Te_m[idx]
patches , new_h, new_w = UT.extract_ordered_overlap(IMG, patch_h, patch_w, stride_h, stride_w)
patches = np.expand_dims(patches, axis = 3)
predictions = model.predict(patches, batch_size= Batch_size, verbose=1)
estimated = UT.recompone_overlap(predictions[:,:,:,0], new_h, new_w, stride_h, stride_w)
estimated = np.where(estimated >= 0.5, 1, 0)
MSK = MSK[0: new_h, 0:new_w]
imgs_normalized[i] = (
(imgs_normalized[i] - np.min(imgs_normalized[i])) /
(np.max(imgs_normalized[i]) - np.min(imgs_normalized[i]))) * 255
return imgs_normalized
#################################### Load Data #####################################
te_data = np.load('data_test.npy')
te_mask = np.load('mask_test.npy')
te_mask = np.expand_dims(te_mask, axis=3)
print('ISIC18 Dataset loaded')
te_data2 = dataset_normalized(te_data)
model = M.BCDU_net_D3(input_size=(256, 256, 3))
model.summary()
model.load_weights('weight_isic18')
predictions = model.predict(te_data, batch_size=8, verbose=1)
y_scores = predictions.reshape(
predictions.shape[0] * predictions.shape[1] * predictions.shape[2] *
predictions.shape[3], 1)
print(y_scores.shape)
y_true = te_mask.reshape(
te_mask.shape[0] * te_mask.shape[1] * te_mask.shape[2] * te_mask.shape[3],
1)
y_scores = np.where(y_scores > 0.5, 1, 0)
y_true = np.where(y_true > 0.5, 1, 0)
