def test_padded(self, octree: torch.Tensor):
# Configuration
depth = self.octree_creator._depth
channels = 6 if self.octree_creator._include_color else 3
num_outputs = 5
# Pad octree to a specific, fixed length
print(f'Original size: {octree.shape}')
padded_size = 1000000
octree_padded = torch.nn.ConstantPad1d(
(0, padded_size - octree.shape[0]), 0)(octree)
# Create batch from the octree and move it to VRAM (it has to be in VRAM for the next step)
octree_batch = ocnn.octree_batch([octree_padded]).cuda()
# Extract features from the octree
data = ocnn.octree_property(octree_batch, 'feature', depth).cuda()
assert data.size(1) == channels
# Test simple convolution
conv1 = ocnn.OctreeConv(depth, channels, num_outputs)
conv1.cuda()
out1 = conv1(data, octree_batch)
# Test fast convolution
conv2 = ocnn.OctreeConvFast(depth, channels, num_outputs)
conv2.cuda()
out2 = conv2(data, octree_batch)
def __init__(self, depth, channel_in, channel_out, kernel_size=[3], stride=1):
super(OctreeConvFastRelu, self).__init__()
self.conv = ocnn.OctreeConvFast(depth,
channel_in,
channel_out,
kernel_size,
stride)
self.relu = torch.nn.ReLU(inplace=True)
def __init__(self, depth, channel_in, channel_out, kernel_size=[3], stride=1, bn_eps=0.00001, bn_momentum=0.01):
super(OctreeConvFastBnRelu, self).__init__()
self.conv = ocnn.OctreeConvFast(depth,
channel_in,
channel_out,
kernel_size,
stride)
self.bn = torch.nn.BatchNorm2d(channel_out,
bn_eps,
bn_momentum)
self.relu = torch.nn.ReLU(inplace=True)
def forward_and_backward(self, kernel_size, stride):
depth = 4
channel = 3
height = 152
num_outputs = 5
octree = ocnn.octree_batch(
ocnn.octree_samples(['octree_1', 'octree_2']))
data = np.random.uniform(-1.0, 1.0,
[1, channel, height, 1]).astype('float32')
# forward
conv1 = ocnn.OctreeConv(depth, channel, num_outputs, kernel_size,
stride)
conv2 = ocnn.OctreeConvFast(depth, channel, num_outputs, kernel_size,
stride)
# use the same initialization
with torch.no_grad():
conv2.weights.data = conv1.weights.data
# forward
octree = octree.to('cuda')
conv1.to('cuda')
data1 = torch.from_numpy(data).to('cuda').requires_grad_()
out1 = conv1(data1, octree)
conv2.to('cuda')
data2 = torch.from_numpy(data).to('cuda').requires_grad_()
out2 = conv2(data2, octree)
# backward
pesudo_grad = torch.rand(out1.shape,
dtype=out1.dtype,
device=out1.device)
out1.backward(pesudo_grad)
out2.backward(pesudo_grad)
# test
self.assertTrue(
np.array_equal(out1.cpu().detach().numpy(),
out2.cpu().detach().numpy()))
self.assertTrue(
np.allclose(data1.grad.cpu().numpy(),
data2.grad.cpu().numpy(),
atol=1e-06))
self.assertTrue(
np.allclose(conv1.weights.grad.cpu().numpy(),
conv2.weights.grad.cpu().numpy(),
atol=1e-06))
def test_simple(self, octree: torch.Tensor):
# Configuration
depth = self.octree_creator._depth
channels = 6 if self.octree_creator._include_color else 3
num_outputs = 5
# Create batch from the octree and move it to VRAM (it has to be in VRAM for the next step)
octree_batch = ocnn.octree_batch([octree]).cuda()
# Extract features from the octree
data = ocnn.octree_property(octree_batch, 'feature', depth).cuda()
assert data.size(1) == channels
# Test simple convolution
conv1 = ocnn.OctreeConv(depth, channels, num_outputs)
conv1.cuda()
out1 = conv1(data, octree_batch)
# Test fast convolution
conv2 = ocnn.OctreeConvFast(depth, channels, num_outputs)
conv2.cuda()
out2 = conv2(data, octree_batch)
def forward_and_backward(self, kernel_size, stride):
depth = 4
channel = 3
height = 152
num_outputs = 5
octree = ocnn.octree_batch(ocnn.octree_samples(['octree_1', 'octree_2']))
data = np.random.uniform(-1.0, 1.0, [1, channel, height, 1]).astype('float32')
# forward
conv1 = ocnn.OctreeConv(depth, channel, num_outputs, kernel_size, stride)
conv2 = ocnn.OctreeConvFast(depth, channel, num_outputs, kernel_size, stride)
conv3 = ocnn.OctreeConv(depth, channel, num_outputs, kernel_size, stride, True)
conv4 = ocnn.OctreeConv(depth, channel, num_outputs, kernel_size, stride)
# use the same initialization
with torch.no_grad():
conv2.weights.data.copy_(conv1.weights.data)
conv3.weights.data.copy_(conv1.weights.data)
conv4.weights.data.copy_(conv1.weights.data)
# forward - compare OctreeConv and OctreeConvFast
octree = octree.cuda()
conv1.cuda()
data1 = torch.from_numpy(data).cuda().requires_grad_()
out1 = conv1(data1, octree)
conv2.cuda()
data2 = torch.from_numpy(data).cuda().requires_grad_()
out2 = conv2(data2, octree)
# forward - compare OctreeConv with nempty = True and False
conv3.cuda()
mask3 = ocnn.octree_property(octree, 'child', depth) >= 0
data3 = torch.from_numpy(data).cuda().requires_grad_()
tmp3 = data3[:, :, mask3]
out3 = conv3(tmp3, octree)
conv4.cuda()
depth_out = depth if stride == 1 else depth - 1
mask4 = ocnn.octree_property(octree, 'child', depth_out) >= 0
data4 = torch.from_numpy(data).cuda().requires_grad_()
tmp4 = data4 * mask3.unsqueeze(-1).float()
tmp4 = conv4(tmp4, octree)
out4 = tmp4[:, :, mask4]
# backward
pesudo_grad1 = torch.rand(out1.shape, dtype=out1.dtype, device=out1.device)
out1.backward(pesudo_grad1)
out2.backward(pesudo_grad1)
pesudo_grad2 = torch.rand(out3.shape, dtype=out3.dtype, device=out3.device)
out3.backward(pesudo_grad2)
out4.backward(pesudo_grad2)
# test
self.assertTrue(np.array_equal(out1.cpu().detach().numpy(),
out2.cpu().detach().numpy()))
self.assertTrue(np.allclose(data1.grad.cpu().numpy(),
data2.grad.cpu().numpy(),
atol=1e-06))
self.assertTrue(np.allclose(conv1.weights.grad.cpu().numpy(),
conv2.weights.grad.cpu().numpy(),
atol=1e-06))
self.assertTrue(np.allclose(out3.cpu().detach().numpy(),
out4.cpu().detach().numpy(),
atol=1e-06))
self.assertTrue(np.allclose(data3.grad.cpu().numpy(),
data4.grad.cpu().numpy(),
atol=1e-06))
self.assertTrue(np.allclose(conv3.weights.grad.cpu().numpy(),
conv4.weights.grad.cpu().numpy(),
atol=1e-06))