def forward(self, x):
batch_size, _, num_points = x.size()
x = get_graph_feature(x, self.k, extra_dim=True)
x = self.conv1(x)
x = self.conv2(x)
x1 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x1, self.k)
x = self.conv3(x)
x = self.conv4(x)
x2 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x2, self.k)
x = self.conv5(x)
x3 = x.max(dim=-1, keepdim=False)[0]
x = torch.cat((x1, x2, x3), dim=1)
x = self.conv6(x)
x = x.max(dim=-1, keepdim=True)[0]
x = x.repeat(1, 1, num_points)
x = torch.cat((x, x1, x2, x3), dim=1)
x = self.conv7(x)
x = self.conv8(x)
x = self.dp1(x)
x = self.conv9(x)
x = x.view(batch_size, num_points, self.num_class)
return x
def forward(self, x):
# pdb.set_trace()
batch_size = x.size()[0]
x = get_graph_feature(x, k=self.args.k)
x = self.conv1(x)
x1 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x1, k=self.args.k)
x = self.conv2(x)
x2 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x2, k=self.args.k)
x = self.conv3(x)
x3 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x3, k=self.args.k)
x = self.conv4(x)
x4 = x.max(dim=-1, keepdim=False)[0]
x = torch.cat((x1, x2, x3, x4), dim=1)
x = self.conv5(x)
x1 = F.adaptive_max_pool1d(x, 1).view(batch_size, -1)
x2 = F.adaptive_avg_pool1d(x, 1).view(batch_size, -1)
x = torch.cat((x1, x2), 1)
x = F.leaky_relu(self.bn6(self.linear1(x)), negative_slope=0.2)
x = self.dp1(x)
x = F.leaky_relu(self.bn7(self.linear2(x)), negative_slope=0.2)
x = self.dp2(x)
x = self.linear3(x)
return x
def forward(self, x):
batch_size = x.size()[0]
x = get_graph_feature(x, k=self.args.k)
x = self.conv1(x)
x1 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x1, k=self.args.k)
x = self.conv2(x)
x2 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x2, k=self.args.k)
x = self.conv3(x)
x3 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x3, k=self.args.k)
x = self.conv4(x)
x4 = x.max(dim=-1, keepdim=False)[0]
x = torch.cat((x1, x2, x3, x4), dim=1)
x = self.conv5(x)
feature = F.adaptive_max_pool1d(x, 1).view(batch_size, -1)
# x1 = F.adaptive_max_pool1d(x, 1).view(batch_size, -1)
# x2 = F.adaptive_avg_pool1d(x, 1).view(batch_size, -1)
# feature = torch.cat((x1, x2), 1)
coarse = self.folding1(feature)
coarse = coarse.view(-1, self.num_coarse, 3)
grid = self.build_grid(x.size()[0])
# grid_feat = self.tile(grid, [1, self.num_coarse, 1])
grid_feat = grid.repeat(1, self.num_coarse, 1)
point_feat = self.tile(self.expand_dims(coarse, 2),
[1, 1, self.grid_size**2, 1])
point_feat = point_feat.view([-1, self.num_fine, 3])
global_feat = self.tile(self.expand_dims(feature, 1),
[1, self.num_fine, 1])
feat = torch.cat([grid_feat, point_feat, global_feat], dim=2)
center = self.tile(self.expand_dims(coarse, 2),
[1, 1, self.grid_size**2, 1])
center = center.view([-1, self.num_fine, 3])
fine = self.folding2(feat.transpose(2, 1)).transpose(2, 1) + center
return coarse, fine
def forward(self, x, l):
B, D, N = x.size()
x0 = get_graph_feature(x, k=self.k)
t = self.transform_net(x0)
x = x.transpose(2, 1)
if D > 3:
x, feature = x.split(3, dim=2)
x = torch.bmm(x, t)
if D > 3:
x = torch.cat([x, feature], dim=2)
x = x.transpose(2, 1)
x = get_graph_feature(x, k=self.k)
x = self.conv1(x)
x = self.conv2(x)
x1 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x1, k=self.k)
x = self.conv3(x)
x = self.conv4(x)
x2 = x.max(dim=-1, keepdim=False)[0]
x = get_graph_feature(x2, k=self.k)
x = self.conv5(x)
x3 = x.max(dim=-1, keepdim=False)[0]
x = torch.cat((x1, x2, x3), dim=1)
x = self.conv6(x)
x = x.max(dim=-1, keepdim=True)[0]
l = l.view(B, -1, 1)
l = self.conv7(l)
x = torch.cat((x, l), dim=1)
x = x.repeat(1, 1, N)
x = torch.cat((x, x1, x2, x3), dim=1)
x = self.conv8(x)
x = self.dp1(x)
x = self.conv9(x)
x = self.dp2(x)
x = self.conv10(x)
x = self.conv11(x)
return x.permute(0, 2, 1).contiguous()
def forward(self, x): batch_size, _, num_points = x.size() x = get_graph_feature(x, self.k) x = self.conv1(x) x = self.conv2(x) x1 = x.max(dim=-1, keepdim=False)[0] x = get_graph_feature(x1, k=self.k) x = self.conv3(x) x = self.conv4(x) x2 = x.max(dim=-1, keepdim=False)[0] x = get_graph_feature(x2, k=self.k) x = self.conv5(x) x3 = x.max(dim=-1, keepdim=False)[0] x = torch.cat((x1, x2, x3), dim=1) x = self.conv6(x) x = x.max(dim=-1, keepdim=True)[0] x = x.repeat(1, 1, num_points) x = torch.cat((x, x1, x2, x3), dim=1) x = self.conv7(x) x = self.conv8(x) x = self.dp1(x) x = self.conv9(x) # x = F.softmax(x, dim=1) # x = F.log_softmax(x, dim=1) '''add softmax: https://towardsdatascience.com/cuda-error-device-side-assert-triggered-c6ae1c8fa4c3 https://github.com/pytorch/pytorch/issues/1204 ''' return x