Weighted binary cross entropy dice loss for segmentation problem
up vote
1
down vote
favorite
I am using weighted Binary cross entropy Dice loss for a segmentation problem with class imbalance (80 times more black pixels than white pixels) .
def weighted_bce_dice_loss(y_true, y_pred):
y_true = K.cast(y_true, 'float32')
y_pred = K.cast(y_pred, 'float32')
averaged_mask = K.pool2d(
y_true, pool_size=(50, 50), strides=(1, 1), padding='same', pool_mode='avg')
weight = K.ones_like(averaged_mask)
w0 = K.sum(weight)
weight = 5. * K.exp(-5. * K.abs(averaged_mask - 0.5))
w1 = K.sum(weight)
weight *= (w0 / w1)
loss = weighted_bce_loss(y_true, y_pred, weight) + dice_loss(y_true, y_pred)
return loss
Dice coeffecient increased and the loss decreased but at every epoch I am getting a black image as output (all the pixels are labelled black)
Any ideas on why this is happening?
python keras image-segmentation loss cross-entropy
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
add a comment |
up vote
1
down vote
favorite
I am using weighted Binary cross entropy Dice loss for a segmentation problem with class imbalance (80 times more black pixels than white pixels) .
def weighted_bce_dice_loss(y_true, y_pred):
y_true = K.cast(y_true, 'float32')
y_pred = K.cast(y_pred, 'float32')
averaged_mask = K.pool2d(
y_true, pool_size=(50, 50), strides=(1, 1), padding='same', pool_mode='avg')
weight = K.ones_like(averaged_mask)
w0 = K.sum(weight)
weight = 5. * K.exp(-5. * K.abs(averaged_mask - 0.5))
w1 = K.sum(weight)
weight *= (w0 / w1)
loss = weighted_bce_loss(y_true, y_pred, weight) + dice_loss(y_true, y_pred)
return loss
Dice coeffecient increased and the loss decreased but at every epoch I am getting a black image as output (all the pixels are labelled black)
Any ideas on why this is happening?
python keras image-segmentation loss cross-entropy
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
"Found 60 images belonging to 1 classes"?
– from keras import michael
Nov 20 at 5:06
It is a segmentation problem. So it is pixelwise binary classification.
– AKSHAYAA VAIDYANATHAN
Nov 20 at 6:58
What happens if you use a non-Dice weighted cross-entropy (i.e.loss = 'binary_crossentropy'andclass_weight = {0: 1 / 81, 1: 80 / 81}?
– from keras import michael
Nov 21 at 4:36
add a comment |
up vote
1
down vote
favorite
up vote
1
down vote
favorite
I am using weighted Binary cross entropy Dice loss for a segmentation problem with class imbalance (80 times more black pixels than white pixels) .
def weighted_bce_dice_loss(y_true, y_pred):
y_true = K.cast(y_true, 'float32')
y_pred = K.cast(y_pred, 'float32')
averaged_mask = K.pool2d(
y_true, pool_size=(50, 50), strides=(1, 1), padding='same', pool_mode='avg')
weight = K.ones_like(averaged_mask)
w0 = K.sum(weight)
weight = 5. * K.exp(-5. * K.abs(averaged_mask - 0.5))
w1 = K.sum(weight)
weight *= (w0 / w1)
loss = weighted_bce_loss(y_true, y_pred, weight) + dice_loss(y_true, y_pred)
return loss
Dice coeffecient increased and the loss decreased but at every epoch I am getting a black image as output (all the pixels are labelled black)
Any ideas on why this is happening?
python keras image-segmentation loss cross-entropy
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
I am using weighted Binary cross entropy Dice loss for a segmentation problem with class imbalance (80 times more black pixels than white pixels) .
def weighted_bce_dice_loss(y_true, y_pred):
y_true = K.cast(y_true, 'float32')
y_pred = K.cast(y_pred, 'float32')
averaged_mask = K.pool2d(
y_true, pool_size=(50, 50), strides=(1, 1), padding='same', pool_mode='avg')
weight = K.ones_like(averaged_mask)
w0 = K.sum(weight)
weight = 5. * K.exp(-5. * K.abs(averaged_mask - 0.5))
w1 = K.sum(weight)
weight *= (w0 / w1)
loss = weighted_bce_loss(y_true, y_pred, weight) + dice_loss(y_true, y_pred)
return loss
Dice coeffecient increased and the loss decreased but at every epoch I am getting a black image as output (all the pixels are labelled black)
Any ideas on why this is happening?
python keras image-segmentation loss cross-entropy
python keras image-segmentation loss cross-entropy
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
edited Nov 19 at 23:33
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
asked Nov 19 at 23:27
AKSHAYAA VAIDYANATHAN
3891522
3891522
"Found 60 images belonging to 1 classes"?
– from keras import michael
Nov 20 at 5:06
It is a segmentation problem. So it is pixelwise binary classification.
– AKSHAYAA VAIDYANATHAN
Nov 20 at 6:58
What happens if you use a non-Dice weighted cross-entropy (i.e.loss = 'binary_crossentropy'andclass_weight = {0: 1 / 81, 1: 80 / 81}?
– from keras import michael
Nov 21 at 4:36
add a comment |
"Found 60 images belonging to 1 classes"?
– from keras import michael
Nov 20 at 5:06
It is a segmentation problem. So it is pixelwise binary classification.
– AKSHAYAA VAIDYANATHAN
Nov 20 at 6:58
What happens if you use a non-Dice weighted cross-entropy (i.e.loss = 'binary_crossentropy'andclass_weight = {0: 1 / 81, 1: 80 / 81}?
– from keras import michael
Nov 21 at 4:36
"Found 60 images belonging to 1 classes"?
– from keras import michael
Nov 20 at 5:06
"Found 60 images belonging to 1 classes"?
– from keras import michael
Nov 20 at 5:06
It is a segmentation problem. So it is pixelwise binary classification.
– AKSHAYAA VAIDYANATHAN
Nov 20 at 6:58
It is a segmentation problem. So it is pixelwise binary classification.
– AKSHAYAA VAIDYANATHAN
Nov 20 at 6:58
What happens if you use a non-Dice weighted cross-entropy (i.e.
loss = 'binary_crossentropy' and class_weight = {0: 1 / 81, 1: 80 / 81}?– from keras import michael
Nov 21 at 4:36
What happens if you use a non-Dice weighted cross-entropy (i.e.
loss = 'binary_crossentropy' and class_weight = {0: 1 / 81, 1: 80 / 81}?– from keras import michael
Nov 21 at 4:36
add a comment |
active
oldest
votes
active
oldest
votes
active
oldest
votes
active
oldest
votes
active
oldest
votes
Thanks for contributing an answer to Stack Overflow!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
To learn more, see our tips on writing great answers.
Some of your past answers have not been well-received, and you're in danger of being blocked from answering.
Please pay close attention to the following guidance:
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
To learn more, see our tips on writing great answers.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53384110%2fweighted-binary-cross-entropy-dice-loss-for-segmentation-problem%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
"Found 60 images belonging to 1 classes"?
– from keras import michael
Nov 20 at 5:06
It is a segmentation problem. So it is pixelwise binary classification.
– AKSHAYAA VAIDYANATHAN
Nov 20 at 6:58
What happens if you use a non-Dice weighted cross-entropy (i.e.
loss = 'binary_crossentropy'andclass_weight = {0: 1 / 81, 1: 80 / 81}?– from keras import michael
Nov 21 at 4:36