def __init__(self):
super(PointNet2ClsSsg, self).__init__()
self.sa1 = PointNetSetAbstraction(npoint=512,
radius=0.2,
nsample=32,
in_channel=3,
mlp=[64, 64, 128],
group_all=False)
self.sa2 = PointNetSetAbstraction(npoint=128,
radius=0.4,
nsample=64,
in_channel=128 + 3,
mlp=[128, 128, 256],
group_all=False)
self.sa3 = PointNetSetAbstraction(npoint=None,
radius=None,
nsample=None,
in_channel=256 + 3,
mlp=[256, 512, 1024],
group_all=True)
self.fc1 = nn.Linear(1024, 512)
self.bn1 = nn.BatchNorm1d(512)
self.drop1 = nn.Dropout(0.4)
self.fc2 = nn.Linear(512, 256)
self.bn2 = nn.BatchNorm1d(256)
self.drop2 = nn.Dropout(0.4)
self.fc3 = nn.Linear(256, 40)
def __init__(self, feature_size=128, normal_channel=True):
super(get_model, self).__init__()
in_channel = 6 if normal_channel else 3
self.normal_channel = normal_channel
self.sa1 = PointNetSetAbstraction(npoint=512,
radius=0.2,
nsample=32,
in_channel=in_channel,
mlp=[64, 64, 128],
group_all=False)
self.sa2 = PointNetSetAbstraction(npoint=128,
radius=0.4,
nsample=64,
in_channel=128 + 3,
mlp=[128, 128, 256],
group_all=False)
self.sa3 = PointNetSetAbstraction(npoint=None,
radius=None,
nsample=None,
in_channel=256 + 3,
mlp=[256, 512, 1024],
group_all=True)
self.projection_head = nn.Sequential(nn.Linear(1024, 512),
nn.BatchNorm1d(512), nn.ReLU(),
nn.Dropout(0.4),
nn.Linear(512, 256),
nn.BatchNorm1d(256), nn.ReLU(),
nn.Dropout(0.4),
nn.Linear(256, feature_size))
def __init__(self, num_class, normal_channel=True):
super(get_model, self).__init__()
in_channel = 6 if normal_channel else 3
self.normal_channel = normal_channel
self.sa1 = PointNetSetAbstraction(npoint=512,
radius=0.2,
nsample=32,
in_channel=in_channel,
mlp=[64, 64, 128],
group_all=False)
self.sa2 = PointNetSetAbstraction(npoint=128,
radius=0.4,
nsample=64,
in_channel=128 + 3,
mlp=[128, 128, 256],
group_all=False)
self.sa3 = PointNetSetAbstraction(npoint=None,
radius=None,
nsample=None,
in_channel=256 + 3,
mlp=[256, 512, 1024],
group_all=True)
self.fc1 = nn.Linear(1024, 512)
self.bn1 = nn.BatchNorm1d(512)
self.drop1 = nn.Dropout(0.4)
self.fc2 = nn.Linear(512, 256)
self.bn2 = nn.BatchNorm1d(256)
self.drop2 = nn.Dropout(0.4)
self.fc3 = nn.Linear(256, num_class)
def __init__(self):
super(FeatureExtractor_1, self).__init__()
self.in_channel = 3
self.sa1 = PointNetSetAbstraction(npoint=1024,
radius=0.2,
nsample=32,
in_channel=self.in_channel,
mlp=[64, 64, 128],
group_all=False)
self.sa2 = PointNetSetAbstraction(npoint=512,
radius=0.2,
nsample=32,
in_channel=128 + self.in_channel,
mlp=[128, 128, 256],
group_all=False)
self.sa3 = PointNetSetAbstraction(npoint=128,
radius=0.2,
nsample=32,
in_channel=256 + self.in_channel,
mlp=[256, 256, 512],
group_all=False)
self.sa4 = PointNetSetAbstraction(npoint=1,
radius=0.2,
nsample=32,
in_channel=512 + self.in_channel,
mlp=[512, 512, 1920],
group_all=False)
def __init__(self, out_dim, normal_channel=False):
super(PointReg_dcp, self).__init__()
in_channel = 3 if normal_channel else 0
self.normal_channel = normal_channel
self.sa1 = PointNetSetAbstractionMsg(
512, [0.1, 0.2, 0.4], [16, 32, 128], in_channel,
[[32, 32, 64], [64, 64, 128], [64, 96, 128]])
self.sa2 = PointNetSetAbstractionMsg(
128, [0.2, 0.4, 0.8], [32, 64, 128], 320,
[[64, 64, 128], [128, 128, 256], [128, 128, 256]])
self.sa3 = PointNetSetAbstraction(None, None, None, 640 + 3,
[256, 256, 512], True)
self.sa1_x = PointNetSetAbstractionMsg(
512, [0.1, 0.2, 0.4], [16, 32, 128], -1,
[[32, 32, 64], [64, 64, 128], [64, 96, 128]])
self.sa2_x = PointNetSetAbstractionMsg(
128, [0.2, 0.4, 0.8], [32, 64, 128], 319,
[[64, 64, 128], [128, 128, 256], [128, 128, 256]])
self.sa3_x = PointNetSetAbstraction(None, None, None, 640 + 2,
[256, 256, 512], True)
self.fc1 = nn.Linear(1024, 512)
self.bn1 = nn.BatchNorm1d(512)
self.drop1 = nn.Dropout(0.4)
self.fc2 = nn.Linear(1024, 512)
self.bn2 = nn.BatchNorm1d(512)
self.drop2 = nn.Dropout(0.5)
self.fc3 = nn.Linear(512, 256)
self.bn3 = nn.BatchNorm1d(256)
self.drop3 = nn.Dropout(0.5)
self.fc4 = nn.Linear(256, out_dim)
def __init__(self, num_classes):
super(get_model, self).__init__()
self.sa1 = PointNetSetAbstraction(1024, 0.1, 32, 9 + 3, [32, 32, 64], False)
self.sa2 = PointNetSetAbstraction(256, 0.2, 32, 64 + 3, [64, 64, 128], False)
self.sa3 = PointNetSetAbstraction(64, 0.4, 32, 128 + 3, [128, 128, 256], False)
self.sa4 = PointNetSetAbstraction(16, 0.8, 32, 256 + 3, [256, 256, 512], False)
self.fp4 = PointNetFeaturePropagation(768, [256, 256]) # note: 768=512+256 equals mlp[-1] from last sa(i.e. sa4) + sa3's output channel(256)
self.fp3 = PointNetFeaturePropagation(384, [256, 256])
self.fp2 = PointNetFeaturePropagation(320, [256, 128])
self.fp1 = PointNetFeaturePropagation(128, [128, 128, 128])
self.conv1 = nn.Conv1d(128, 128, 1)
self.bn1 = nn.BatchNorm1d(128)
self.drop1 = nn.Dropout(0.5)
self.conv2 = nn.Conv1d(128, num_classes, 1)
def __init__(
self,
num_class,
normal_channel=True,
):
# super(get_model,self) 首先找到 get_model 的父类（就是类 nn.Module），然后把类 get_model 的对象转换为类 nn.Module 的对象
super(get_model, self).__init__()
# 如果增加法向量信息，则 in_channel = 3
in_channel = 3 if normal_channel else 0
self.normal_channel = normal_channel
# 有三层特征提取
self.sa1 = PointNetSetAbstractionMsg(
512, [0.1, 0.2, 0.4], [16, 32, 128], in_channel,
[[32, 32, 64], [64, 64, 128], [64, 96, 128]])
self.sa2 = PointNetSetAbstractionMsg(
128, [0.2, 0.4, 0.8], [32, 64, 128], 320,
[[64, 64, 128], [128, 128, 256], [128, 128, 256]])
self.sa3 = PointNetSetAbstraction(None, None, None, 640 + 3,
[256, 512, 1024], True)
# 有三个全连接层，最后输出指定维数的特征向量
self.fc1 = nn.Linear(1024, 512)
self.bn1 = nn.BatchNorm1d(512)
self.drop1 = nn.Dropout(0.4)
self.fc2 = nn.Linear(512, 256)
self.bn2 = nn.BatchNorm1d(256)
self.drop2 = nn.Dropout(0.5)
self.fc3 = nn.Linear(256, num_class)
def __init__(self, cfg):
super().__init__()
n_points, k, n_c, d_points, dropout = cfg.num_point, cfg.model.nneighbor, cfg.num_class, cfg.input_dim, cfg.model.dropout
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
# initial MLP
self.fc1 = nn.Linear(d_points, 32)
self.fc2 = nn.Linear(32, 32)
# point transformers
self.pt1 = TransformerBlock(32, cfg.model.transformer_dim, k, dropout,
device)
self.td1 = PointNetSetAbstraction(n_points // 4,
0,
k,
35, [64, 64],
group_all=False,
knn=True)
self.pt2 = TransformerBlock(64, cfg.model.transformer_dim, k, dropout,
device)
self.td2 = PointNetSetAbstraction(n_points // 16,
0,
k,
67, [128, 128],
group_all=False,
knn=True)
self.pt3 = TransformerBlock(128, cfg.model.transformer_dim, k, dropout,
device)
self.td3 = PointNetSetAbstraction(n_points // 64,
0,
k,
131, [256, 256],
group_all=False,
knn=True)
self.pt4 = TransformerBlock(256, cfg.model.transformer_dim, k, dropout,
device)
self.td4 = PointNetSetAbstraction(n_points // 256,
0,
k,
259, [512, 512],
group_all=False,
knn=True)
self.pt5 = TransformerBlock(512, cfg.model.transformer_dim, k, dropout,
device)
# final MLP
self.fc3 = nn.Linear(512, 256)
self.fc4 = nn.Linear(256, 64)
self.fc5 = nn.Linear(64, n_c)
def __init__(self, num_classes):
super(PointNet2SemSeg, self).__init__()
self.sa1 = PointNetSetAbstraction(1024, 0.1, 32, 6 + 3, [32, 32, 64],
False)
self.sa2 = PointNetSetAbstraction(256, 0.2, 32, 64 + 3, [64, 64, 128],
False)
self.sa3 = PointNetSetAbstraction(64, 0.4, 32, 128 + 3,
[128, 128, 256], False)
self.sa4 = PointNetSetAbstraction(16, 0.8, 32, 256 + 3,
[256, 256, 512], False)
self.fp4 = PointNetFeaturePropagation(768, [256, 256])
self.fp3 = PointNetFeaturePropagation(384, [256, 256])
self.fp2 = PointNetFeaturePropagation(320, [256, 128])
self.fp1 = PointNetFeaturePropagation(128, [128, 128, 128])
self.conv1 = nn.Conv1d(128, 128, 1)
self.bn1 = nn.BatchNorm1d(128)
self.drop1 = nn.Dropout(0.5)
self.conv2 = nn.Conv1d(128, num_classes, 1)
def __init__(self, k, nneighbor, channels) -> None:
super().__init__()
self.sa = PointNetSetAbstraction(k,
0,
nneighbor,
channels[0],
channels[1:],
group_all=False,
knn=True)
def __init__(self, num_class, normal_channel=True):
super(get_model, self).__init__()
in_channel = 3 if normal_channel else 0
self.normal_channel = normal_channel
self.sa1 = PointNetSetAbstractionMsg(512, [0.1, 0.2, 0.4], [16, 32, 128], in_channel,[[32, 32, 64], [64, 64, 128], [64, 96, 128]])
self.sa2 = PointNetSetAbstractionMsg(128, [0.2, 0.4, 0.8], [32, 64, 128], 320,[[64, 64, 128], [128, 128, 256], [128, 128, 256]])
self.sa3 = PointNetSetAbstraction(None, None, None, 640 + 3, [256, 512, 1024], True)
self.fc1 = nn.Linear(1024, 512)
self.bn1 = nn.BatchNorm1d(512)
self.drop1 = nn.Dropout(0.4)
self.fc2 = nn.Linear(512, 256)
self.bn2 = nn.BatchNorm1d(256)
self.drop2 = nn.Dropout(0.5)
self.fc3 = nn.Linear(256, num_class)
def __init__(self):
super(PointNet2ClsMsg, self).__init__()
self.sa1 = PointNetSetAbstractionMsg(
512, [0.1, 0.2, 0.4], [16, 32, 128], 0,
[[32, 32, 64], [64, 64, 128], [64, 96, 128]])
self.sa2 = PointNetSetAbstractionMsg(
128, [0.2, 0.4, 0.8], [32, 64, 128], 320,
[[64, 64, 128], [128, 128, 256], [128, 128, 256]])
self.sa3 = PointNetSetAbstraction(None, None, None, 640 + 3,
[256, 512, 1024], True)
self.fc1 = nn.Linear(1024, 512)
self.bn1 = nn.BatchNorm1d(512)
self.drop1 = nn.Dropout(0.4)
self.fc2 = nn.Linear(512, 256)
self.bn2 = nn.BatchNorm1d(256)
self.drop2 = nn.Dropout(0.4)
self.fc3 = nn.Linear(256, 40)