def forward(self, inputs):
x = inputs['image']
conv1 = self.conv_layer1_1(x)
conv2 = self.conv_layer1_2(conv1)
la1 = self.la1(conv2)
tr1 = self.tr1([la1])
st2 = self.st2(tr1)
tr2 = self.tr2(st2)
st3 = self.st3(tr2)
tr3 = self.tr3(st3)
st4 = self.st4(tr3)
if self.upsample:
# Upsampling
x0_h, x0_w = st4[0].shape[2:4]
x1 = F.upsample(st4[1], size=(x0_h, x0_w), mode='bilinear')
x2 = F.upsample(st4[2], size=(x0_h, x0_w), mode='bilinear')
x3 = F.upsample(st4[3], size=(x0_h, x0_w), mode='bilinear')
x = paddle.concat([st4[0], x1, x2, x3], 1)
return x
res = []
for i, layer in enumerate(st4):
if i == self.freeze_at:
layer.stop_gradient = True
if i in self.return_idx:
res.append(layer)
return res
def forward(self, x):
x = self.conv1(x)
x = self.bn1(x)
x = self.relu(x)
x = self.conv2(x)
x = self.bn2(x)
x = self.relu(x)
x = self.layer1(x)
x_list = []
for i in range(self.stage2_cfg['NUM_BRANCHES']):
# if self.transition1[i] is not None:
if self.transition1[i] is not equalmap:
x_list.append(self.transition1[i](x))
else:
x_list.append(x)
y_list = self.stage2(x_list)
x_list = []
for i in range(self.stage3_cfg['NUM_BRANCHES']):
# if self.transition2[i] is not None:
if self.transition2[i] is not equalmap:
if i < self.stage2_cfg['NUM_BRANCHES']:
x_list.append(self.transition2[i](y_list[i]))
else:
x_list.append(self.transition2[i](y_list[-1]))
else:
x_list.append(y_list[i])
y_list = self.stage3(x_list)
x_list = []
for i in range(self.stage4_cfg['NUM_BRANCHES']):
# if self.transition3[i] is not None:
if self.transition3[i] is not equalmap:
if i < self.stage3_cfg['NUM_BRANCHES']:
x_list.append(self.transition3[i](y_list[i]))
else:
x_list.append(self.transition3[i](y_list[-1]))
else:
x_list.append(y_list[i])
x = self.stage4(x_list)
# Upsampling
# x0_h, x0_w = x[0].size(2), x[0].size(3)
x0_h, x0_w = x[0].shape[2], x[0].shape[3]
x1 = F.upsample(x[1], size=(x0_h, x0_w), mode='bilinear')
x2 = F.upsample(x[2], size=(x0_h, x0_w), mode='bilinear')
x3 = F.upsample(x[3], size=(x0_h, x0_w), mode='bilinear')
# x = torch.cat([x[0], x1, x2, x3], 1)
x = paddle.concat([x[0], x1, x2, x3], 1)
z = {}
for head in self.heads:
z[head] = self.__getattr__(head)(x)
return [z]
def forward(self, img_id, img_shape, inputs):
outputs = []
boxes = []
scores = []
downsample = 32
feats = self.backbone(inputs)
route = None
for idx, feat in enumerate(feats[::-1]):
if idx > 0:
feat = paddle.concat(x=[route, feat], axis=1)
route, tip = self.yolo_blocks[idx](feat)
block_out = self.block_outputs[idx](tip)
outputs.append(block_out)
if idx < 2:
route = self.route_blocks[idx](route)
route = F.upsample(route, scale_factor=2)
if self.model_mode != 'train':
anchor_mask = self.anchor_masks[idx]
mask_anchors = []
for m in anchor_mask:
mask_anchors.append(self.anchors[2 * m])
mask_anchors.append(self.anchors[2 * m + 1])
b, s = paddle.fluid.layers.yolo_box(
x=block_out,
img_size=img_shape,
anchors=mask_anchors,
class_num=self.num_classes,
conf_thresh=self.valid_thresh,
downsample_ratio=downsample)
boxes.append(b)
scores.append(paddle.transpose(s, perm=[0, 2, 1]))
downsample //= 2
if self.model_mode == 'train':
return outputs
preds = [
img_id, paddle.fluid.layers.multiclass_nms(
bboxes=paddle.concat(
boxes, axis=1),
scores=paddle.concat(
scores, axis=2),
score_threshold=self.valid_thresh,
nms_top_k=self.nms_topk,
keep_top_k=self.nms_posk,
nms_threshold=self.nms_thresh,
background_label=-1)
]
if self.model_mode == 'test':
return preds
# model_mode == "eval"
return outputs + preds
def forward(self, input):
output1 = self.conv1_fpn(input[0])
output2 = self.conv2_fpn(input[1])
up2 = F.upsample(
output2, size=paddle.shape(output1)[-2:], mode='nearest')
output1 = paddle.add(output1, up2)
output1 = self.conv3_fpn(output1)
return output1, output2
def forward(self, x):
c2, c3, c4, c5 = x
in5 = self.in5_conv(c5)
in4 = self.in4_conv(c4)
in3 = self.in3_conv(c3)
in2 = self.in2_conv(c2)
out4 = in4 + F.upsample(
in5, size=in4.shape[2:4], mode="nearest", align_mode=1) # 1/16
out3 = in3 + F.upsample(
out4, size=in3.shape[2:4], mode="nearest", align_mode=1) # 1/8
out2 = in2 + F.upsample(
out3, size=in2.shape[2:4], mode="nearest", align_mode=1) # 1/4
p4 = F.upsample(out4,
size=in5.shape[2:4],
mode="nearest",
align_mode=1)
p3 = F.upsample(out3,
size=in5.shape[2:4],
mode="nearest",
align_mode=1)
p2 = F.upsample(out2,
size=in5.shape[2:4],
mode="nearest",
align_mode=1)
fuse = paddle.concat([in5, p4, p3, p2], axis=1)
fuse_conv = self.fuse_conv(fuse) * 0.005
return [c5 + fuse_conv]
def forward(self, x):
c2, c3, c4, c5 = x
in5 = self.in5_conv(c5)
in4 = self.in4_conv(c4)
in3 = self.in3_conv(c3)
in2 = self.in2_conv(c2)
out4 = in4 + F.upsample(
in5, scale_factor=2, mode="nearest", align_mode=1) # 1/16
out3 = in3 + F.upsample(
out4, scale_factor=2, mode="nearest", align_mode=1) # 1/8
out2 = in2 + F.upsample(
out3, scale_factor=2, mode="nearest", align_mode=1) # 1/4
p5 = self.p5_conv(in5)
p4 = self.p4_conv(out4)
p3 = self.p3_conv(out3)
p2 = self.p2_conv(out2)
p5 = F.upsample(p5, scale_factor=8, mode="nearest", align_mode=1)
p4 = F.upsample(p4, scale_factor=4, mode="nearest", align_mode=1)
p3 = F.upsample(p3, scale_factor=2, mode="nearest", align_mode=1)
fuse = paddle.concat([p5, p4, p3, p2], axis=1)
return fuse
def forward(self, x):
skips = []
for i in range(self.n_skip):
skips.append(getattr(self, 'ConvBlockskip' + str(i + 1))(x))
x = F.avg_pool2d(x, 2)
x = getattr(self, 'ConvBlock' + str(i + 1))(x)
x = self.ConvBlock5(x)
for i in range(self.n_skip):
x = getattr(self, 'ConvBlock' + str(i + 6))(x)
x = F.upsample(x, scale_factor=2)
x = skips[self.n_skip - i - 1] + x
return x
def forward(self, x, res_dict=None):
out = []
residual_func_idx = 0
for i in range(len(self._in_channels)):
residual = x[i]
for j in range(len(self._in_channels)):
if j > i:
xj = self.residual_func_list[residual_func_idx](x[j])
residual_func_idx += 1
xj = upsample(xj, scale_factor=2**(j - i), mode="nearest")
residual = paddle.add(x=residual, y=xj)
elif j < i:
xj = x[j]
for k in range(i - j):
xj = self.residual_func_list[residual_func_idx](xj)
residual_func_idx += 1
residual = paddle.add(x=residual, y=xj)
residual = self.relu(residual)
out.append(residual)
return out
def forward(self, input):
outs = []
residual_func_idx = 0
for i in range(self._actual_ch):
residual = input[i]
for j in range(len(self._in_channels)):
if j > i:
y = self.residual_func_list[residual_func_idx](input[j])
residual_func_idx += 1
y = F.upsample(y, scale_factor=2**(j - i), mode="nearest")
residual = paddle.add(x=residual, y=y)
elif j < i:
y = input[j]
for k in range(i - j):
y = self.residual_func_list[residual_func_idx](y)
residual_func_idx += 1
residual = paddle.add(x=residual, y=y)
residual = F.relu(residual)
outs.append(residual)
return outs
def _upsample_like(src, tar):
src = F.upsample(src, size=tar.shape[2:], mode='bilinear')
return src
def _upsample_add(self, x, y):
_, _, H, W = y.shape
return F.upsample(x, size=(H, W), mode='bilinear') + y
def _upsample(self, x, y, scale=1):
_, _, H, W = y.shape
return F.upsample(x, size=(H // scale, W // scale), mode='bilinear')
def _upsample_add(self, x, y, scale=1):
return F.upsample(x, scale_factor=scale, mode='bilinear') + y
def _upsample_like(src, tar):
src = F.upsample(src, size=paddle.shape(tar)[2:], mode='bilinear')
return src
