def forward(self, input):
"""
Forward pass of the function.
"""
return (1 / self.alpha) * F.log(
(1 + F.exp(self.alpha * input)) / (1 + F.exp(self.alpha * (input - 1)))
)
def decode(self, anchors: Tensor, deltas: Tensor) -> Tensor:
if self.reg_std is not None:
deltas *= self.reg_std
if self.reg_mean is not None:
deltas += self.reg_mean
(
anchor_width,
anchor_height,
anchor_ctr_x,
anchor_ctr_y,
) = self._box_ltrb_to_cs_opr(anchors, 1)
pred_ctr_x = anchor_ctr_x + deltas[:, 0::4] * anchor_width
pred_ctr_y = anchor_ctr_y + deltas[:, 1::4] * anchor_height
pred_width = anchor_width * F.exp(deltas[:, 2::4])
pred_height = anchor_height * F.exp(deltas[:, 3::4])
pred_x1 = pred_ctr_x - 0.5 * pred_width
pred_y1 = pred_ctr_y - 0.5 * pred_height
pred_x2 = pred_ctr_x + 0.5 * pred_width
pred_y2 = pred_ctr_y + 0.5 * pred_height
pred_box = self._concat_new_axis(pred_x1, pred_y1, pred_x2, pred_y2, 2)
pred_box = pred_box.reshape(pred_box.shapeof(0), -1)
return pred_box
def bbox_transform_inv_opr(bbox, deltas):
max_delta = math.log(1000.0 / 16)
""" Transforms the learned deltas to the final bbox coordinates, the axis is 1"""
bbox_width = bbox[:, 2] - bbox[:, 0] + 1
bbox_height = bbox[:, 3] - bbox[:, 1] + 1
bbox_ctr_x = bbox[:, 0] + 0.5 * bbox_width
bbox_ctr_y = bbox[:, 1] + 0.5 * bbox_height
pred_ctr_x = bbox_ctr_x + deltas[:, 0] * bbox_width
pred_ctr_y = bbox_ctr_y + deltas[:, 1] * bbox_height
dw = deltas[:, 2]
dh = deltas[:, 3]
dw = F.minimum(dw, max_delta)
dh = F.minimum(dh, max_delta)
pred_width = bbox_width * F.exp(dw)
pred_height = bbox_height * F.exp(dh)
pred_x1 = pred_ctr_x - 0.5 * pred_width
pred_y1 = pred_ctr_y - 0.5 * pred_height
pred_x2 = pred_ctr_x + 0.5 * pred_width
pred_y2 = pred_ctr_y + 0.5 * pred_height
# pred_boxes = F.concat((pred_x1.reshape(-1, 1), pred_y1.reshape(-1, 1),
# pred_x2.reshape(-1, 1), pred_y2.reshape(-1, 1)), axis=1)
pred_boxes = F.stack([pred_x1, pred_y1, pred_x2, pred_y2], axis=1)
return pred_boxes
def smooth_grad_1st(flo, image, alpha):
img_dx, img_dy = gradient(image)
weights_x = F.exp(-F.mean(F.abs(img_dx), 1, keepdims=True) * alpha)
weights_y = F.exp(-F.mean(F.abs(img_dy), 1, keepdims=True) * alpha)
dx, dy = gradient(flo)
loss_x = weights_x * F.abs(dx) / 2.0
loss_y = weights_y * F.abs(dy) / 2.0
return F.mean(loss_x) / 2.0 + F.mean(loss_y) / 2.0
def forward(self, input):
"""
Forward pass of the function.
"""
if self.hard is False:
return (input >= 0).float() * swish_function(
input, False, False, None, None
) + (input < 0).float() * (F.exp(input) - 1) * F.sigmoid(input)
else:
return (input >= 0).float() * input * F.max(
self.a, F.min(self.b, (input + 1.0) / 2.0)
) + (input < 0).float() * (
F.exp(input - 1) * F.max(self.a, F.min(self.b, (input + 1.0) / 2.0))
)
def _bce_loss_with_logits(output, labels, **kwargs):
r"""
Sigmoid cross entropy with logits, see tensorflow
https://www.tensorflow.org/api_docs/python/tf/nn/sigmoid_cross_entropy_with_logits
"""
loss = F.maximum(output, 0) - output * labels + F.log(1 + F.exp(-F.abs(output)))
return loss.mean()
def softmax_loss(pred, label, ignore_label=-1):
max_pred = F.zero_grad(pred.max(axis=1, keepdims=True))
pred -= max_pred
log_prob = pred - F.log(F.exp(pred).sum(axis=1, keepdims=True))
mask = 1 - F.equal(label, ignore_label)
vlabel = label * mask
loss = -(F.indexing_one_hot(log_prob, vlabel, 1) * mask)
return loss
def softmax_loss(score, label, ignore_label=-1):
max_score = F.zero_grad(score.max(axis=1, keepdims=True))
score -= max_score
log_prob = score - F.log(F.exp(score).sum(axis=1, keepdims=True))
mask = (label != ignore_label)
vlabel = label * mask
loss = -(F.indexing_one_hot(log_prob, vlabel.astype("int32"), 1) *
mask).sum()
loss = loss / F.maximum(mask.sum(), 1)
return loss
def softmax_loss_opr(pred, label, ignore_label=-1):
max_pred = pred.max(axis=1, keepdims=True).detach()
pred -= max_pred
log_prob = pred - F.log(F.exp(pred).sum(axis=1, keepdims=True))
mask = 1 - F.equal(label, ignore_label)
vlabel = label * mask.astype(np.float32)
loss = -(F.nn.indexing_one_hot(log_prob, vlabel.astype(np.int32),
1).flatten() * mask)
return loss
def softmax_cross_entropy(pred, label, axis=1, ignore_index=255):
offset = F.zero_grad(pred.max(axis=axis, keepdims=True))
pred = pred - offset
log_prob = pred - F.log(F.exp(pred).sum(axis=axis, keepdims=True))
mask = 1 - F.equal(label, ignore_index)
vlabel = label * mask
loss = -(F.indexing_one_hot(log_prob, vlabel, axis) *
mask).sum() / F.maximum(mask.sum(), 1)
return loss
def softmax_loss(scores: Tensor,
labels: Tensor,
ignore_label: int = -1) -> Tensor:
max_scores = F.zero_grad(scores.max(axis=1, keepdims=True))
scores -= max_scores
log_prob = scores - F.log(F.exp(scores).sum(axis=1, keepdims=True))
mask = labels != ignore_label
vlabels = labels * mask
loss = -(F.indexing_one_hot(log_prob, vlabels.astype("int32"), 1) *
mask).sum()
loss = loss / F.maximum(mask.sum(), 1)
return loss
def forward(self, x):
"""
Forward pass of the function
"""
if self.alpha == 0.0:
return x
if self.alpha < 0.0:
return -F.log(1 - self.alpha * (x + self.alpha)) / self.alpha
if self.alpha > 0.0:
return (F.exp(self.alpha * x) - 1) / self.alpha + self.alpha
def encoding(self, x):
for conv in self.convs:
x = conv(x)
print('encoding', x.shape)
print('fmap', x.shape)
#code = self.gap(x)
#code = code.reshape(-1, code.shape[1])
code = x.dimshuffle(0, 2, 3, 1)
print(code.shape)
mean = self.fc_mean(code).dimshuffle(0, 3, 1, 2)
var = 1e-5 + F.exp(self.fc_var(code)).dimshuffle(0, 3, 1, 2)
return mean, var
def decode(self, anchors: Tensor, deltas: Tensor) -> Tensor:
deltas *= self.reg_std
deltas += self.reg_mean
(
anchor_width,
anchor_height,
anchor_ctr_x,
anchor_ctr_y,
) = self._box_ltrb_to_cs_opr(anchors, 1)
pred_ctr_x = anchor_ctr_x + deltas[:, 0::4] * anchor_width
pred_ctr_y = anchor_ctr_y + deltas[:, 1::4] * anchor_height
pred_width = anchor_width * F.exp(deltas[:, 2::4])
pred_height = anchor_height * F.exp(deltas[:, 3::4])
pred_x1 = pred_ctr_x - 0.5 * pred_width
pred_y1 = pred_ctr_y - 0.5 * pred_height
pred_x2 = pred_ctr_x + 0.5 * pred_width
pred_y2 = pred_ctr_y + 0.5 * pred_height
pred_box = F.stack([pred_x1, pred_y1, pred_x2, pred_y2], axis=2)
pred_box = pred_box.reshape(pred_box.shape[0], -1)
return pred_box
def decode_outputs(self, outputs):
grids = []
strides = []
for (hsize, wsize), stride in zip(self.hw, self.strides):
xv, yv = meshgrid(F.arange(hsize), F.arange(wsize))
grid = F.stack((xv, yv), 2).reshape(1, -1, 2)
grids.append(grid)
shape = grid.shape[:2]
strides.append(F.full((*shape, 1), stride))
grids = F.concat(grids, axis=1)
strides = F.concat(strides, axis=1)
outputs[..., :2] = (outputs[..., :2] + grids) * strides
outputs[..., 2:4] = F.exp(outputs[..., 2:4]) * strides
return outputs
def get_output_and_grid(self, output, k, stride, dtype):
grid = self.grids[k]
batch_size = output.shape[0]
n_ch = 5 + self.num_classes
hsize, wsize = output.shape[-2:]
if grid.shape[2:4] != output.shape[2:4]:
yv, xv = meshgrid([F.arange(hsize), F.arange(wsize)])
grid = F.stack((xv, yv), 2).reshape(1, 1, hsize, wsize,
2).type(dtype)
self.grids[k] = grid
output = output.view(batch_size, self.n_anchors, n_ch, hsize, wsize)
output = (output.permute(0, 1, 3, 4,
2).reshape(batch_size,
self.n_anchors * hsize * wsize,
-1))
grid = grid.view(1, -1, 2)
output[..., :2] = (output[..., :2] + grid) * stride
output[..., 2:4] = F.exp(output[..., 2:4]) * stride
return output, grid
def forward(self, a):
# add
if self.mode == "add":
x = a + mge.tensor(np.float32(10))
y = a + mge.tensor(self.data1)
z = x + y
# sub
elif self.mode == "sub":
x = a - mge.tensor(np.float32(10))
y = a - mge.tensor(self.data1)
z = x - y
# mul
elif self.mode == "mul":
x = a * mge.tensor(np.float32(10))
y = mge.tensor(self.data1) * a
z = x * y
# div
elif self.mode == "div":
y = mge.tensor(self.data1) / a
x = a / mge.tensor(np.float32(2))
z = y / x
# cycle_div
elif self.mode == "cycle_div":
z = a / mge.tensor(self.data1)
# abs
elif self.mode == "abs":
z = F.abs(a)
# exp
elif self.mode == "exp":
z = F.exp(a)
# log
elif self.mode == "log":
z = F.log(a)
else:
raise NotImplementedError('no such elemwise mode "%s"' % self.mode)
return z
def forward(self, x):
x = x * self.a
# will result into a copy of output in grad
x = F.exp(x)
return x
def test_elemementwise():
a = Tensor(1.0)
assert F.exp(a).ndim == 0
assert (a + a).ndim == 0
assert (a + 1).ndim == 0
def g(x):
return log(exp(x))
def f(x):
return exp(x)
def g(x, y):
return log(exp(x) + exp(y))
def f(x):
return log(exp(x))
def forward(self, features, label=None, mask=None):
"""
if label and mask both None, the loss will degenerate to
SimSLR unsupervised loss.
Reference:
"A Simple Framework for Contrastive Learning of Visual Representations"<https://arxiv.org/pdf/2002.05709.pdf>
"Supervised Contrastive Learning"<https://arxiv.org/abs/2004.11362>
Args:
features(tensor): The embedding feature. shape=[bs, n_views, ...]
label(tensor): The label of images, shape=[bs]
mask(tensor): contrastive mask of shape [bsz, bsz], mask_{i,j}=1 if sample j
has the same class as sample i. Can be asymmetric.
return:
loss
"""
if len(features.shape) < 3:
raise ValueError("Features need have 3 dimensions at least")
bs, num_view = features.shape[:2]
#if dimension > 3, change the shape of the features to [bs, num_view, ...]
if len(features.shape) > 3:
features = features.reshape(bs, num_view, -1)
#label and mask cannot provided at the same time
if (label is not None) and (mask is not None):
raise ValueError("label and mask cannot provided at the same time")
elif (label is None) and (mask is None):
mask = F.eye(bs, dtype="float32")
elif label is not None:
label = label.reshape(-1, 1)
if label.shape[0] != bs:
raise RuntimeError(
"Num of labels does not match num of features")
mask = F.equal(label, label.T)
else:
mask = mask.astype("float32")
contrast_count = features.shape[1]
features = F.split(features, features.shape[1], axis=1)
contrast_feature = F.squeeze(F.concat(features, axis=0), axis=1)
if self.contrast_mode == "one":
anchor_feature = features[:, 0]
anchor_count = 1
elif self.contrast_mode == "all":
anchor_feature = contrast_feature
anchor_count = contrast_count
else:
raise ValueError("Unknown mode:{}".format(self.contrast_mode))
#compute logits
anchor_dot_contrast = F.div(
F.matmul(anchor_feature, contrast_feature.T), self.temperate)
#for numerical stability
logits_max = F.max(anchor_dot_contrast, axis=-1, keepdims=True)
logits = anchor_dot_contrast - logits_max
#tile mask
an1, con = mask.shape[:2]
nums = anchor_count * contrast_count
# mask-out self-contrast cases
mask = F.stack([mask] * nums).reshape(an1 * anchor_count,
con * contrast_count)
logits_mask = F.scatter(
F.ones_like(mask), 1,
F.arange(0, int(bs * anchor_count), dtype="int32").reshape(-1, 1),
F.zeros(int(bs * anchor_count), dtype="int32").reshape(-1, 1))
mask = mask * logits_mask
#compute log_prob
exp_logits = F.exp(logits) * logits_mask
log_prob = logits - F.log(F.sum(exp_logits, axis=1,
keepdims=True)) #equation 2
#mean
mean_log_prob_pos = F.sum(mask * log_prob, axis=1) / F.sum(mask,
axis=1)
#loss
loss = -(self.temperate / self.base_temperate) * mean_log_prob_pos
loss = F.mean(loss.reshape(anchor_count, bs))
return loss
def forward(self, input):
"""
Forward pass of the function.
"""
return F.pow((1 + F.exp(-self.beta * input)), -self.alpha)
def softplus(x: Tensor) -> Tensor:
return F.log(1 + F.exp(-F.abs(x))) + F.relu(x)
def forward(self, a):
# add
if self.mode == "add":
x = a + mge.tensor(np.float32(10))
y = a + mge.tensor(self.data1)
z = x + y
# sub
elif self.mode == "sub":
x = a - mge.tensor(np.float32(10))
y = a - mge.tensor(self.data1)
z = x - y
# mul
elif self.mode == "mul":
x = a * mge.tensor(np.float32(10))
y = mge.tensor(self.data1) * a
z = x * y
# div
elif self.mode == "max":
x = a + mge.tensor(self.data)
y = a + mge.tensor(self.data2)
z = F.maximum(x, y)
elif self.mode == "min":
x = a + mge.tensor(self.data)
y = a + mge.tensor(self.data2)
z = F.minimum(x, y)
elif self.mode == "pow":
z = a**2
elif self.mode == "ceil":
z = F.ceil(a)
elif self.mode == "floor":
z = F.floor(a)
elif self.mode == "div":
y = mge.tensor(self.data1) / a
x = a / mge.tensor(np.float32(2))
z = y / x
# cycle_div
elif self.mode == "cycle_div":
z = a / mge.tensor(self.data1)
# abs
elif self.mode == "abs":
z = F.abs(a)
# exp
elif self.mode == "exp":
z = F.exp(a)
# log
elif self.mode == "log":
z = F.log(a)
elif self.mode == "fuse_add_relu":
y = a + mge.tensor(self.data2)
z = F.relu(y)
elif self.mode == "fuse_mul_add3":
y = a * mge.tensor(self.data1)
z = y + mge.tensor(self.data2)
elif self.mode == "fuse_add_sigmoid":
y = a + mge.tensor(self.data2)
z = F.sigmoid(y)
else:
raise NotImplementedError('no such elemwise mode "%s"' % self.mode)
return z
def f(x):
return F.exp(x)
def get_losses(self, anchors, pred_logits, pred_offsets, gt_boxes,
im_info):
# pylint: disable=too-many-statements
def positive_bag_loss(logits, axis=1):
weight = 1.0 / (1.0 - logits)
weight /= weight.sum(axis=axis, keepdims=True)
bag_prob = (weight * logits).sum(axis=1)
return -layers.safelog(bag_prob)
def negative_bag_loss(logits, gamma):
return (logits**gamma) * (-layers.safelog(1.0 - logits))
pred_scores = F.sigmoid(pred_logits)
box_prob_list = []
positive_losses = []
clamp_eps = 1e-7
bucket_size = self.cfg.bucket_size
for bid in range(im_info.shape[0]):
boxes_info = gt_boxes[bid, :im_info[bid, 4].astype("int32")]
# id 0 is used for background classes, so -1 first
labels = boxes_info[:, 4].astype("int32") - 1
pred_box = self.box_coder.decode(anchors,
pred_offsets[bid]).detach()
overlaps = layers.get_iou(boxes_info[:, :4], pred_box).detach()
thresh1 = self.cfg.box_iou_threshold
thresh2 = F.clip(overlaps.max(axis=1, keepdims=True),
lower=thresh1 + clamp_eps,
upper=1.0)
gt_pred_prob = F.clip((overlaps - thresh1) / (thresh2 - thresh1),
lower=0,
upper=1.0)
image_boxes_prob = F.zeros(pred_logits.shape[1:]).detach()
# guarantee that nonzero_idx is not empty
if gt_pred_prob.max() > clamp_eps:
_, nonzero_idx = F.cond_take(gt_pred_prob != 0, gt_pred_prob)
# since nonzeros is only 1 dim, use num_anchor to get real indices
num_anchors = gt_pred_prob.shape[1]
anchors_idx = nonzero_idx % num_anchors
gt_idx = nonzero_idx // num_anchors
image_boxes_prob[anchors_idx,
labels[gt_idx]] = gt_pred_prob[gt_idx,
anchors_idx]
box_prob_list.append(image_boxes_prob)
# construct bags for objects
match_quality_matrix = layers.get_iou(boxes_info[:, :4],
anchors).detach()
num_gt = match_quality_matrix.shape[0]
_, matched_idx = F.topk(
match_quality_matrix,
k=bucket_size,
descending=True,
no_sort=True,
)
matched_idx = matched_idx.detach()
matched_idx_flatten = matched_idx.reshape(-1)
gather_idx = labels.reshape(-1, 1)
gather_idx = F.broadcast_to(gather_idx, (num_gt, bucket_size))
gather_src = pred_scores[bid, matched_idx_flatten]
gather_src = gather_src.reshape(num_gt, bucket_size, -1)
matched_score = F.indexing_one_hot(gather_src, gather_idx, axis=2)
topk_anchors = anchors[matched_idx_flatten]
boxes_broad_cast = F.broadcast_to(
F.expand_dims(boxes_info[:, :4], axis=1),
(num_gt, bucket_size, 4)).reshape(-1, 4)
matched_offsets = self.box_coder.encode(topk_anchors,
boxes_broad_cast)
reg_loss = layers.smooth_l1_loss(
pred_offsets[bid, matched_idx_flatten],
matched_offsets,
beta=self.cfg.smooth_l1_beta).sum(
axis=-1) * self.cfg.reg_loss_weight
matched_reg_scores = F.exp(-reg_loss)
positive_losses.append(
positive_bag_loss(matched_score *
matched_reg_scores.reshape(-1, bucket_size),
axis=1))
num_foreground = im_info[:, 4].sum()
pos_loss = F.concat(positive_losses).sum() / F.maximum(
1.0, num_foreground)
box_probs = F.stack(box_prob_list, axis=0)
neg_loss = negative_bag_loss(
pred_scores *
(1 - box_probs), self.cfg.focal_loss_gamma).sum() / F.maximum(
1.0, num_foreground * bucket_size)
alpha = self.cfg.focal_loss_alpha
pos_loss = pos_loss * alpha
neg_loss = neg_loss * (1 - alpha)
loss_dict = {
"total_loss": pos_loss + neg_loss,
"pos_loss": pos_loss,
"neg_loss": neg_loss,
}
return loss_dict
def forward(self, x):
y = 1 / (1 + F.exp(-x))
self.save_for_backward(y)
return y
