def __init__(self, c1, c2, k=3, s=1, g=1, e=1.0, shortcut=False):
# ch_in, ch_out, kernel, stride, groups, expansion, shortcut
super(CrossConv, self).__init__()
c_ = int(c2 * e) # hidden channels
self.cv1 = Conv(c1, c_, (1, k), (1, s))
self.cv2 = Conv(c_, c2, (k, 1), (s, 1), g=g)
self.add = shortcut and c1 == c2
def __init__(self,
c1,
c2,
k=1,
s=1,
g=1,
act=True): # ch_in, ch_out, kernel, stride, groups
super(GhostConv, self).__init__()
c_ = c2 // 2 # hidden channels
self.cv1 = Conv(c1, c_, k, s, g, act)
self.cv2 = Conv(c_, c_, 5, 1, c_, act)
def __init__(self,
c1,
c2,
n=1,
shortcut=True,
g=1,
e=0.5): # ch_in, ch_out, number, shortcut, groups, expansion
super(C3, self).__init__()
c_ = int(c2 * e) # hidden channels
self.cv1 = Conv(c1, c_, 1, 1)
self.cv2 = nn.Conv2d(c1, c_, 1, 1, bias=False)
self.cv3 = nn.Conv2d(c_, c_, 1, 1, bias=False)
self.cv4 = Conv(2 * c_, c2, 1, 1)
self.bn = nn.BatchNorm2d(2 * c_) # applied to cat(cv2, cv3)
self.act = nn.LeakyReLU(0.1, inplace=True)
self.m = nn.Sequential(
*[CrossConv(c_, c_, 3, 1, g, 1.0, shortcut) for _ in range(n)])
def __init__(self, c1, c2, k, s):
super(GhostBottleneck, self).__init__()
c_ = c2 // 2
self.conv = nn.Sequential(
GhostConv(c1, c_, 1, 1), # pw
DWConv(c_, c_, k, s, act=False) if s == 2 else nn.Identity(), # dw
GhostConv(c_, c2, 1, 1, act=False)) # pw-linear
self.shortcut = nn.Sequential(DWConv(
c1, c1, k, s, act=False), Conv(
c1, c2, 1, 1, act=False)) if s == 2 else nn.Identity()
def __init__(self,mask=1,ch=()):
super(DenseMask, self).__init__()
self.proj1 = Conv(ch[0]//2, 1,k=3)
self.proj2 = nn.ConvTranspose2d(ch[1], 1, 4, stride=2,
padding=1, output_padding=0,
groups=1, bias=False)
self.proj3 = nn.ConvTranspose2d(ch[2], 1, 8, stride=4,
padding=2, output_padding=0,
groups=1, bias=False)
self.sigmoid = nn.Sigmoid()
def __init__(self,id_embedding=256,ch=()):
super(SAAN, self).__init__()
self.proj1 = nn.Sequential(Conv(ch[0]//2, 256,k=3),
SAAN_Attention(k_size=3, ch=256, s_state=True, c_state=False))
self.proj2 = nn.Sequential( Conv(ch[1], 256,k=3),
nn.ConvTranspose2d(256, 256, 4, stride=2,
padding=1, output_padding=0,
groups=256, bias=False),
SAAN_Attention(k_size=3, ch=256, s_state=True, c_state=False))
self.proj3 = nn.Sequential(Conv(ch[2], 256,k=3),
nn.ConvTranspose2d(256, 256, 8, stride=4,
padding=2, output_padding=0,
groups=256, bias=False),
SAAN_Attention(k_size=3, ch=256, s_state=True, c_state=False))
self.node = nn.Sequential(SAAN_Attention(k_size=3, ch=256*3, s_state=False, c_state=True),
Conv(256 * 3, 256,k=3),
nn.Conv2d(256, id_embedding,
kernel_size=1, stride=1,
padding=0, bias=True)
)
def __init__(self,k_size = 3,ch = 256, s_state = False, c_state = False):
super(SAAN_Attention, self).__init__()
self.avg_pool = nn.AdaptiveAvgPool2d(1)
self.max_pool = nn.AdaptiveMaxPool2d(1)
self.sigmoid = nn.Sigmoid()
#self.conv1 = Conv(ch, ch,k=1)
self.s_state = s_state
self.c_state = c_state
if c_state:
self.c_attention = nn.Sequential(nn.Conv1d(1, 1, kernel_size=k_size, padding=(k_size - 1) // 2, bias=False),
nn.LayerNorm([1, ch]),
nn.LeakyReLU(0.3, inplace=True),
nn.Linear(ch, ch, bias=False))
if s_state:
self.conv_s = nn.Sequential(Conv(ch, ch // 4, k=1))
self.s_attention = nn.Conv2d(2, 1, 7, padding=3, bias=False)