def __init__(self, input_dim=128, point_num=50, bn=True):
super(AntipodalPredictBatch, self).__init__()
self.input_dim = input_dim
# self.conv1 = nn.Conv2d(input_dim, 512, 1)
# self.max_pool = nn.MaxPool2d((1, point_num), stride=(1, 1))
# self.mean_pool = nn.AvgPool2d((1, point_num), stride=(1, 1))
self.weighted_mean = pt_utils.Conv2d(input_dim, input_dim, kernel_size=(1, point_num), padding=0, bn=bn)
self.conv1 = pt_utils.Conv1d(input_dim, 512, 1, bn=bn)
self.score = nn.Sequential(pt_utils.Conv1d(512, 256, 1, bn=bn), pt_utils.Conv1d(256, 1, 1, activation=None))
self.sigmoid = nn.Sigmoid()
def __init__(self, input_dim=128, point_num=50, bn=True, tanh=False):
super(PosePredictBatch, self).__init__()
self.input_dim = input_dim
# self.conv1 = nn.Conv2d(input_dim, 512, 1)
self.max_pool = nn.MaxPool2d((1, point_num), stride=(1, 1))
self.mean_pool = nn.AvgPool2d((1, point_num), stride=(1, 1))
self.weighted_mean = pt_utils.Conv2d(input_dim, input_dim, kernel_size=(1, point_num), padding=0, bn=bn)
self.conv1 = pt_utils.Conv1d(input_dim, 512, 1, bn=bn)
self.score = nn.Sequential(pt_utils.Conv1d(512, 256, 1, bn=bn), pt_utils.Conv1d(256, 1, 1, activation=None))
self.offset = nn.Sequential(pt_utils.Conv1d(512, 256, 1, bn=bn), pt_utils.Conv1d(256, 3, 1, activation=None))
self.angle = nn.Sequential(pt_utils.Conv1d(512, 256, 1, bn=bn), pt_utils.Conv1d(256, 1, 1, activation=None))
self.sigmoid = nn.Sigmoid()
self.use_tanh = tanh
self.tanh = nn.Tanh()
def __init__(self, in_dim=3, out_dim=128, bn=True):
super(GridFeat, self).__init__()
self.out_dim = out_dim
self.layers = nn.Sequential(pt_utils.Conv1d(in_dim, 64, 1, bn=bn), pt_utils.Conv1d(64, out_dim, 1, bn=bn))
def __init__(self, input_channels=0, use_xyz=True, bn=True):
super(PointNetFeat, self).__init__()
self.SA_modules = nn.ModuleList()
c_in = input_channels
# print(c_in)
self.SA_modules.append(
PointnetSAModuleMSG(
npoint=1024,
radii=[0.05, 0.1],
nsamples=[16, 32],
mlps=[[c_in, 16, 16, 32], [c_in, 32, 32, 64]],
use_xyz=True,
bn=bn
)
)
c_out_0 = 32 + 64
c_in = c_out_0
self.SA_modules.append(
PointnetSAModuleMSG(
npoint=256,
radii=[0.1, 0.2],
nsamples=[16, 32],
mlps=[[c_in, 64, 64, 128], [c_in, 64, 96, 128]],
use_xyz=use_xyz,
bn=bn
)
)
c_out_1 = 128 + 128
c_in = c_out_1
self.SA_modules.append(
PointnetSAModuleMSG(
npoint=64,
radii=[0.2, 0.4],
nsamples=[16, 32],
mlps=[[c_in, 128, 196, 256], [c_in, 128, 196, 256]],
use_xyz=use_xyz,
bn=bn
)
)
c_out_2 = 256 + 256
c_in = c_out_2
self.SA_modules.append(
PointnetSAModuleMSG(
npoint=16,
radii=[0.4, 0.8],
nsamples=[16, 32],
mlps=[[c_in, 256, 256, 512], [c_in, 256, 384, 512]],
use_xyz=use_xyz,
bn=bn
)
)
c_out_3 = 512 + 512
self.FP_modules = nn.ModuleList()
self.FP_modules.append(
PointnetFPModule(mlp=[256 + input_channels, 128, 128], bn=bn)
)
self.FP_modules.append(PointnetFPModule(mlp=[512 + c_out_0, 256, 256], bn=bn))
self.FP_modules.append(PointnetFPModule(mlp=[512 + c_out_1, 512, 512], bn=bn))
self.FP_modules.append(
PointnetFPModule(mlp=[c_out_3 + c_out_2, 512, 512], bn=bn)
)
self.FC_layer = nn.Sequential(
pt_utils.Conv1d(128, 128, bn=bn)
)
self.sigmoid = nn.Sigmoid()