def focal_loss_fixed(y_true, y_pred):
#pdb.set_trace()
'''
return -K.mean(K.mean(K.sum(y_true * K.log(y_pred + K.epsilon()), axis=-1)))
'''
if cfg.num_classes > 1:
pt = K.sum(y_true * y_pred, axis=-1)
alpha_ = K.sum(y_true * alpha, axis=-1)
else:
pt = K.where(K.equal(y_true, 1), y_pred, 1 - y_pred)
alpha_ = K.where(K.equal(y_true, 1), alpha, 1 - alpha)
return -K.mean(
K.mean(alpha_ * K.pow(1. + K.epsilon() - pt, gamma) *
K.log(pt + K.epsilon())))
def _focal(y_true, y_pred):
""" Compute the focal loss given the target tensor and the predicted tensor.
As defined in https://arxiv.org/abs/1708.02002
Args
y_true: Tensor of target data from the generator with shape (B, N, num_classes).
y_pred: Tensor of predicted data from the network with shape (B, N, num_classes).
Returns
The focal loss of y_pred w.r.t. y_true.
"""
labels = y_true[:, :, :-1]
anchor_state = y_true[:, :,
-1] # -1 for ignore, 0 for background, 1 for object
classification = y_pred
# filter out "ignore" anchors
indices = backend.where(keras.backend.not_equal(anchor_state, -1))
labels = backend.gather_nd(labels, indices)
classification = backend.gather_nd(classification, indices)
# compute the focal loss
alpha_factor = keras.backend.ones_like(labels) * alpha
alpha_factor = backend.where(keras.backend.equal(labels, 1),
alpha_factor, 1 - alpha_factor)
focal_weight = backend.where(keras.backend.equal(labels, 1),
1 - classification, classification)
focal_weight = alpha_factor * focal_weight**gamma
cls_loss = focal_weight * keras.backend.binary_crossentropy(
labels, classification)
# compute the normalizer: the number of positive anchors
normalizer = backend.where(keras.backend.equal(anchor_state, 1))
normalizer = keras.backend.cast(
keras.backend.shape(normalizer)[0], keras.backend.floatx())
normalizer = keras.backend.maximum(keras.backend.cast_to_floatx(1.0),
normalizer)
return keras.backend.sum(cls_loss) / normalizer
def mse(y_true, y_pred):
cost = K.abs(y_pred - y_true)
costs = K.where(K.is_nan(cost), T.zeros_like(cost), cost)
return K.sum(costs, axis=-1)
def mse(y_true, y_pred):
y_true = K.where(K.is_nan(y_true), y_pred, y_true)
cost = K.abs(y_pred - y_true)
return K.sum(cost, axis=-1)