def __init__(self, num_anchors, num_classes, in_channels_list, focal_init=False, lite=False, large_kernel=False):
super().__init__()
self.num_classes = num_classes
num_anchors = tuplify(num_anchors, len(in_channels_list))
kernel_size = 5 if (lite and large_kernel) else 3
self.loc_heads = nn.ModuleList([
nn.Sequential(
Norm("default", c),
Conv2d(c, n * 4, kernel_size=kernel_size,
depthwise_separable=lite, mid_norm_layer='default')
)
for c, n in zip(in_channels_list, num_anchors)
])
self.cls_heads = nn.ModuleList([
nn.Sequential(
Norm("default", c),
Conv2d(c, n * num_classes, kernel_size=kernel_size,
depthwise_separable=lite, mid_norm_layer='default')
)
for c, n in zip(in_channels_list, num_anchors)
])
if focal_init:
for p in self.cls_heads:
get_last_conv(p).bias.data.fill_(inverse_sigmoid(0.01))
def __init__(self, num_anchors, num_classes=2, in_channels=245, f_channels=256):
super().__init__()
self.num_classes = num_classes
self.conv = DWConv2d(
in_channels, f_channels, kernel_size=5,
norm='default', act='default')
self.loc_conv = Conv2d(
f_channels, num_anchors * 4, kernel_size=1)
self.cls_conv = Conv2d(
f_channels, num_anchors * num_classes, kernel_size=1)
bias_init_constant(self.cls_conv, inverse_sigmoid(0.01))
def __init__(self, num_anchors, num_classes, in_channels_list, f_channels=256, expand_ratio=4, num_layers=4):
super().__init__()
self.expand_ratio = expand_ratio
self.projects = nn.ModuleList([
MBConv(c, c, f_channels, kernel_size=5)
for c in in_channels_list
])
self.num_classes = num_classes
self.loc_head = self._make_head(
num_layers, f_channels, num_anchors * 4)
self.cls_head = self._make_head(
num_layers, f_channels, num_anchors * num_classes)
bias_init_constant(self.cls_head[-1][1], inverse_sigmoid(0.01))
def __init__(self, num_anchors, num_classes, f_channels=256, num_layers=4, lite=False):
super().__init__()
self.num_classes = num_classes
self.loc_head = _make_head(
f_channels, num_layers, num_anchors * 4, lite=lite)
self.cls_head = _make_head(
f_channels, num_layers, num_anchors * num_classes, lite=lite)
weight_init_normal(self.loc_head, 0, 0.01)
weight_init_normal(self.cls_head, 0, 0.01)
fc = self.cls_head[-1]
if lite:
fc = fc[-1]
bias_init_constant(fc, inverse_sigmoid(0.01))
def __init__(self, num_anchors, num_classes, in_channels_list, focal_init=False):
super().__init__()
self.num_classes = num_classes
num_anchors = tuplify(num_anchors, len(in_channels_list))
self.loc_heads = nn.ModuleList([
nn.Sequential(
Norm("default", c),
Conv2d(c, n * 4, kernel_size=1)
)
for c, n in zip(in_channels_list, num_anchors)
])
self.cls_heads = nn.ModuleList([
nn.Sequential(
Norm("default", c),
Conv2d(c, n * num_classes, kernel_size=1)
)
for c, n in zip(in_channels_list, num_anchors)
])
if focal_init:
for p in self.cls_heads:
get_last_conv(p).bias.data.fill_(inverse_sigmoid(0.01))