|
|
import numpy as np |
|
|
|
|
|
|
|
|
|
|
|
class PointcloudNoise(object): |
|
|
''' Point cloud noise transformation class. |
|
|
|
|
|
It adds noise to point cloud data. |
|
|
|
|
|
Args: |
|
|
stddev (int): standard deviation |
|
|
''' |
|
|
|
|
|
def __init__(self, stddev): |
|
|
self.stddev = stddev |
|
|
|
|
|
def __call__(self, data): |
|
|
''' Calls the transformation. |
|
|
|
|
|
Args: |
|
|
data (dictionary): data dictionary |
|
|
''' |
|
|
data_out = data.copy() |
|
|
points = data[None] |
|
|
noise = self.stddev * np.random.randn(*points.shape) |
|
|
noise = noise.astype(np.float32) |
|
|
data_out[None] = points + noise |
|
|
return data_out |
|
|
|
|
|
|
|
|
class SubsamplePointcloud(object): |
|
|
''' Point cloud subsampling transformation class. |
|
|
|
|
|
It subsamples the point cloud data. |
|
|
|
|
|
Args: |
|
|
N (int): number of points to be subsampled |
|
|
''' |
|
|
|
|
|
def __init__(self, N): |
|
|
self.N = N |
|
|
|
|
|
def __call__(self, data): |
|
|
''' Calls the transformation. |
|
|
|
|
|
Args: |
|
|
data (dict): data dictionary |
|
|
''' |
|
|
data_out = data.copy() |
|
|
points = data[None] |
|
|
|
|
|
indices = np.random.randint(points.shape[0], size=self.N) |
|
|
data_out[None] = points[indices, :] |
|
|
|
|
|
if 'normals' in data.keys(): |
|
|
normals = data['normals'] |
|
|
data_out['normals'] = normals[indices, :] |
|
|
|
|
|
return data_out |
|
|
|
|
|
|
|
|
class SubsamplePoints(object): |
|
|
''' Points subsampling transformation class. |
|
|
|
|
|
It subsamples the points data. |
|
|
|
|
|
Args: |
|
|
N (int): number of points to be subsampled |
|
|
''' |
|
|
|
|
|
def __init__(self, N): |
|
|
self.N = N |
|
|
|
|
|
def __call__(self, data): |
|
|
''' Calls the transformation. |
|
|
|
|
|
Args: |
|
|
data (dictionary): data dictionary |
|
|
''' |
|
|
points = data[None] |
|
|
occ = data['occ'] |
|
|
|
|
|
data_out = data.copy() |
|
|
if isinstance(self.N, int): |
|
|
idx = np.random.randint(points.shape[0], size=self.N) |
|
|
data_out.update({ |
|
|
None: points[idx, :], |
|
|
'occ': occ[idx], |
|
|
}) |
|
|
else: |
|
|
Nt_out, Nt_in = self.N |
|
|
occ_binary = (occ >= 0.5) |
|
|
points0 = points[~occ_binary] |
|
|
points1 = points[occ_binary] |
|
|
|
|
|
idx0 = np.random.randint(points0.shape[0], size=Nt_out) |
|
|
idx1 = np.random.randint(points1.shape[0], size=Nt_in) |
|
|
|
|
|
points0 = points0[idx0, :] |
|
|
points1 = points1[idx1, :] |
|
|
points = np.concatenate([points0, points1], axis=0) |
|
|
|
|
|
occ0 = np.zeros(Nt_out, dtype=np.float32) |
|
|
occ1 = np.ones(Nt_in, dtype=np.float32) |
|
|
occ = np.concatenate([occ0, occ1], axis=0) |
|
|
|
|
|
volume = occ_binary.sum() / len(occ_binary) |
|
|
volume = volume.astype(np.float32) |
|
|
|
|
|
data_out.update({ |
|
|
None: points, |
|
|
'occ': occ, |
|
|
'volume': volume, |
|
|
}) |
|
|
return data_out |
|
|
|
|
|
|
|
|
class SubsamplePointcloudSeq(object): |
|
|
''' Point cloud sequence subsampling transformation class. |
|
|
|
|
|
It subsamples the point cloud sequence data. |
|
|
|
|
|
Args: |
|
|
N (int): number of points to be subsampled |
|
|
connected_samples (bool): whether to obtain connected samples |
|
|
random (bool): whether to sub-sample randomly |
|
|
''' |
|
|
|
|
|
def __init__(self, N, connected_samples=False, random=True): |
|
|
self.N = N |
|
|
self.connected_samples = connected_samples |
|
|
self.random = random |
|
|
|
|
|
def __call__(self, data): |
|
|
''' Calls the transformation. |
|
|
|
|
|
Args: |
|
|
data (dictionary): data dictionary |
|
|
''' |
|
|
data_out = data.copy() |
|
|
points = data[None] |
|
|
n_steps, T, dim = points.shape |
|
|
N_max = min(self.N, T) |
|
|
if self.connected_samples or not self.random: |
|
|
indices = (np.random.randint(T, size=self.N) if self.random else |
|
|
np.arange(N_max)) |
|
|
data_out[None] = points[:, indices, :] |
|
|
else: |
|
|
indices = np.random.randint(T, size=(n_steps, self.N)) |
|
|
data_out[None] = \ |
|
|
points[np.arange(n_steps).reshape(-1, 1), indices, :] |
|
|
return data_out |
|
|
|
|
|
|
|
|
class SubsamplePointsSeq(object): |
|
|
''' Points sequence subsampling transformation class. |
|
|
|
|
|
It subsamples the points sequence data. |
|
|
|
|
|
Args: |
|
|
N (int): number of points to be subsampled |
|
|
connected_samples (bool): whether to obtain connected samples |
|
|
random (bool): whether to sub-sample randomly |
|
|
''' |
|
|
|
|
|
def __init__(self, N, connected_samples=False, random=True, spatial_completion=False): |
|
|
self.N = N |
|
|
self.connected_samples = connected_samples |
|
|
self.random = random |
|
|
self.spatial_completion = spatial_completion |
|
|
|
|
|
def __call__(self, data): |
|
|
''' Calls the transformation. |
|
|
|
|
|
Args: |
|
|
data (dictionary): data dictionary |
|
|
''' |
|
|
points = data[None] |
|
|
occ = data['occ'] |
|
|
data_out = data.copy() |
|
|
|
|
|
if not self.spatial_completion: |
|
|
n_steps, T, dim = points.shape |
|
|
N_max = min(self.N, T) |
|
|
|
|
|
if self.connected_samples or not self.random: |
|
|
indices = (np.random.randint(T, size=self.N) if self.random |
|
|
else np.arange(N_max)) |
|
|
|
|
|
data_out.update({ |
|
|
None: points[:, indices], |
|
|
'occ': occ[:, indices], |
|
|
}) |
|
|
else: |
|
|
indices = np.random.randint(T, size=(n_steps, self.N)) |
|
|
help_arr = np.arange(n_steps).reshape(-1, 1) |
|
|
data_out.update({ |
|
|
None: points[help_arr, indices, :], |
|
|
'occ': occ[help_arr, indices] |
|
|
}) |
|
|
|
|
|
else: |
|
|
all_pts = [] |
|
|
all_occ = [] |
|
|
for pts, o_value in zip(points, occ): |
|
|
N_pts, dim = pts.shape |
|
|
indices = np.random.randint(N_pts, size=self.N) |
|
|
all_pts.append(pts[indices, :]) |
|
|
all_occ.append(o_value[indices]) |
|
|
data_out.update({ |
|
|
None: np.stack(all_pts), |
|
|
'occ': np.stack(all_occ) |
|
|
}) |
|
|
|
|
|
return data_out |
|
|
|
