Skip to content

GitLab

Commit 470a58ad authored by liyinhao's avatar liyinhao
Browse files

change names, paras

parent bdce5236
Hide whitespace changes
Inline Side-by-side
Showing with 105 additions and 286 deletions
mmdet3d/datasets/pipelines/indoor_augment.py
View file @ 470a58ad
...@@ -22,7 +22,7 @@ def _rotz(t): ...@@ -22,7 +22,7 @@ def _rotz(t):
@PIPELINES.register_module() @PIPELINES.register_module()
class IndoorFlipData(object): class IndoorFlipData(object):
"""Indoor Flip Data """Indoor Flip Data.
Flip point_cloud and groundtruth boxes. Flip point_cloud and groundtruth boxes.
...@@ -30,28 +30,27 @@ class IndoorFlipData(object): ...@@ -30,28 +30,27 @@ class IndoorFlipData(object):
seed (int): Numpy random seed. seed (int): Numpy random seed.
""" """
def __init__(self, seed=None): def __init__(self):
if seed is not None: pass
np.random.seed(seed)
def __call__(self, results): def __call__(self, results):
point_cloud = results.get('point_cloud', None) points = results.get('points', None)
gt_boxes = results.get('gt_boxes', None) gt_bboxes_3d = results.get('gt_bboxes_3d', None)
name = 'scannet' if gt_boxes.shape[1] == 6 else 'sunrgbd' name = 'scannet' if gt_bboxes_3d.shape[1] == 6 else 'sunrgbd'
if np.random.random() > 0.5: if np.random.random() > 0.5:
# Flipping along the YZ plane # Flipping along the YZ plane
point_cloud[:, 0] = -1 * point_cloud[:, 0] points[:, 0] = -1 * points[:, 0]
gt_boxes[:, 0] = -1 * gt_boxes[:, 0] gt_bboxes_3d[:, 0] = -1 * gt_bboxes_3d[:, 0]
if name == 'sunrgbd': if name == 'sunrgbd':
gt_boxes[:, 6] = np.pi - gt_boxes[:, 6] gt_bboxes_3d[:, 6] = np.pi - gt_bboxes_3d[:, 6]
results['gt_boxes'] = gt_boxes results['gt_boxes'] = gt_bboxes_3d
if name == 'scannet' and np.random.random() > 0.5: if name == 'scannet' and np.random.random() > 0.5:
# Flipping along the XZ plane # Flipping along the XZ plane
point_cloud[:, 1] = -1 * point_cloud[:, 1] points[:, 1] = -1 * points[:, 1]
gt_boxes[:, 1] = -1 * gt_boxes[:, 1] gt_bboxes_3d[:, 1] = -1 * gt_bboxes_3d[:, 1]
results['gt_boxes'] = gt_boxes results['gt_bboxes_3d'] = gt_bboxes_3d
results['point_cloud'] = point_cloud results['points'] = points
return results return results
...@@ -79,17 +78,13 @@ class IndoorGlobalRotScale(object): ...@@ -79,17 +78,13 @@ class IndoorGlobalRotScale(object):
""" """
def __init__(self, def __init__(self,
seed=None,
use_rotate=True, use_rotate=True,
use_color=False, use_color=False,
use_scale=True, use_scale=True,
use_height=True, use_height=True,
rot_range=1 / 3, rot_range=[-np.pi / 6, np.pi / 6],
scale_range=0.3, scale_range=[0.85, 1.15],
color_mean=[0.5, 0.5, 0.5]): color_mean=[0.5, 0.5, 0.5]):
if seed is not None:
np.random.seed(seed)
self.use_rotate = use_rotate self.use_rotate = use_rotate
self.use_color = use_color self.use_color = use_color
self.use_scale = use_scale self.use_scale = use_scale
...@@ -132,53 +127,54 @@ class IndoorGlobalRotScale(object): ...@@ -132,53 +127,54 @@ class IndoorGlobalRotScale(object):
return np.concatenate([new_centers, new_lengths], axis=1) return np.concatenate([new_centers, new_lengths], axis=1)
def __call__(self, results): def __call__(self, results):
point_cloud = results.get('point_cloud', None) points = results.get('points', None)
gt_boxes = results.get('gt_boxes', None) gt_bboxes_3d = results.get('gt_bboxes_3d', None)
name = 'scannet' if gt_boxes.shape[1] == 6 else 'sunrgbd' name = 'scannet' if gt_bboxes_3d.shape[1] == 6 else 'sunrgbd'
if self.use_rotate: if self.use_rotate:
rot_angle = (np.random.random() * self.rot_range * np.pi rot_angle = np.random.random() * (
) - np.pi * self.rot_range / 2 # -30 ~ +30 degree self.rot_range[1] - self.rot_range[0]) + self.rot_range[0]
rot_mat = _rotz(rot_angle) rot_mat = _rotz(rot_angle)
point_cloud[:, 0:3] = np.dot(point_cloud[:, 0:3], points[:, 0:3] = np.dot(points[:, 0:3], np.transpose(rot_mat))
np.transpose(rot_mat))
if name == 'scannet': if name == 'scannet':
gt_boxes = self._rotate_aligned_boxes(gt_boxes, rot_mat) gt_bboxes_3d = self._rotate_aligned_boxes(
gt_bboxes_3d, rot_mat)
else: else:
gt_boxes[:, 0:3] = np.dot(gt_boxes[:, 0:3], gt_bboxes_3d[:, 0:3] = np.dot(gt_bboxes_3d[:, 0:3],
np.transpose(rot_mat)) np.transpose(rot_mat))
gt_boxes[:, 6] -= rot_angle gt_bboxes_3d[:, 6] -= rot_angle
# Augment RGB color # Augment RGB color
if self.use_color: if self.use_color:
rgb_color = point_cloud[:, 3:6] + self.color_mean rgb_color = points[:, 3:6] + self.color_mean
rgb_color *= (1 + 0.4 * np.random.random(3) - 0.2 # brightness change for each channel
) # brightness change for each channel rgb_color *= (1 + 0.4 * np.random.random(3) - 0.2)
rgb_color += (0.1 * np.random.random(3) - 0.05 # color shift for each channel
) # color shift for each channel rgb_color += (0.1 * np.random.random(3) - 0.05)
# jittering on each pixel
rgb_color += np.expand_dims( rgb_color += np.expand_dims(
(0.05 * np.random.random(point_cloud.shape[0]) - 0.025), (0.05 * np.random.random(points.shape[0]) - 0.025), -1)
-1) # jittering on each pixel
rgb_color = np.clip(rgb_color, 0, 1) rgb_color = np.clip(rgb_color, 0, 1)
# randomly drop out 30% of the points' colors # randomly drop out 30% of the points' colors
rgb_color *= np.expand_dims( rgb_color *= np.expand_dims(
np.random.random(point_cloud.shape[0]) > 0.3, -1) np.random.random(points.shape[0]) > 0.3, -1)
point_cloud[:, 3:6] = rgb_color - self.color_mean points[:, 3:6] = rgb_color - self.color_mean
if self.use_scale: if self.use_scale:
# Augment point cloud scale: 0.85x-1.15x # Augment point cloud scale
scale_ratio = np.random.random( scale_ratio = np.random.random() * (
) * self.scale_range + 1 - self.scale_range / 2 self.scale_range[1] -
scale_ratio = np.expand_dims(np.tile(scale_ratio, 3), 0) self.scale_range[0]) + self.scale_range[0]
point_cloud[:, 0:3] *= scale_ratio scale_ratio = np.tile(scale_ratio, 3)[None, ...]
gt_boxes[:, 0:3] *= scale_ratio points[:, 0:3] *= scale_ratio
gt_boxes[:, 3:6] *= scale_ratio gt_bboxes_3d[:, 0:3] *= scale_ratio
gt_bboxes_3d[:, 3:6] *= scale_ratio
if self.use_height: if self.use_height:
point_cloud[:, -1] *= scale_ratio[0, 0] points[:, -1] *= scale_ratio[0, 0]
results['point_cloud'] = point_cloud results['points'] = points
results['gt_boxes'] = gt_boxes results['gt_bboxes_3d'] = gt_bboxes_3d
return results return results
def __repr__(self): def __repr__(self):
......
mmdet3d/datasets/pipelines/indoor_loading.py deleted 100644 → 0
View file @ bdce5236
import os.path as osp
import numpy as np
from mmdet.datasets.registry import PIPELINES
@PIPELINES.register_module
class LoadPointsFromFile(object):
"""Load Points From File.
Load sunrgbd and scannet points from file.
Args:
name (str): scannet or sunrgbd.
use_color (bool): Whether to use color.
use_height (bool): Whether to use height.
mean_color (List[float]): Mean color of the point cloud.
"""
def __init__(self, use_color, use_height, mean_color):
self.use_color = use_color
self.use_height = use_height
self.mean_color = mean_color
def __call__(self, results):
data_path = results.get('data_path', None)
info = results.get('info', None)
name = 'scannet' if info.get('image', None) is None else 'sunrgbd'
if name == 'scannet':
scan_name = info['point_cloud']['lidar_idx']
point_cloud = self._get_lidar(scan_name, data_path)
else:
point_cloud = np.load(
osp.join(data_path, 'lidar',
'%06d.npz' % info['point_cloud']['lidar_idx']))['pc']
if not self.use_color:
if name == 'scannet':
pcl_color = point_cloud[:, 3:6]
point_cloud = point_cloud[:, 0:3]
else:
if name == 'scannet':
pcl_color = point_cloud[:, 3:6]
point_cloud = point_cloud[:, 0:6]
point_cloud[:, 3:] = (point_cloud[:, 3:] -
np.array(self.mean_color)) / 256.0
if self.use_height:
floor_height = np.percentile(point_cloud[:, 2], 0.99)
height = point_cloud[:, 2] - floor_height
point_cloud = np.concatenate(
[point_cloud, np.expand_dims(height, 1)], 1)
results['point_cloud'] = point_cloud
if name == 'scannet':
results['pcl_color'] = pcl_color
return results
def __repr__(self):
repr_str = self.__class__.__name__
repr_str += '(use_height={})'.format(self.use_height)
repr_str += '(use_color={}'.format(self.use_color)
repr_str += '(mean_color={})'.format(self.mean_color)
return repr_str
def _get_lidar(self, scan_name, data_path):
lidar_file = osp.join(data_path, scan_name + '_vert.npy')
assert osp.exists(lidar_file)
return np.load(lidar_file)
@PIPELINES.register_module
class LoadAnnotations3D(object):
"""Load Annotations3D.
Load sunrgbd and scannet annotations.
Args:
name (str): scannet or sunrgbd.
"""
def __init__(self):
pass
def __call__(self, results):
data_path = results.get('data_path', None)
info = results.get('info', None)
if info['annos']['gt_num'] != 0:
gt_boxes = info['annos']['gt_boxes_upright_depth']
gt_classes = info['annos']['class'].reshape(-1, 1)
gt_boxes_mask = np.ones_like(gt_classes)
else:
gt_boxes = np.zeros((1, 6), dtype=np.float32)
gt_classes = np.zeros((1, 1))
gt_boxes_mask = np.zeros((1, 1))
name = 'scannet' if info.get('image', None) is None else 'sunrgbd'
if name == 'scannet':
scan_name = info['point_cloud']['lidar_idx']
instance_labels = self._get_instance_label(scan_name, data_path)
semantic_labels = self._get_semantic_label(scan_name, data_path)
results['instance_labels'] = instance_labels
results['semantic_labels'] = semantic_labels
results['gt_boxes'] = gt_boxes
results['gt_classes'] = gt_classes
results['gt_boxes_mask'] = gt_boxes_mask
return results
def __repr__(self):
repr_str = self.__class__.__name__
return repr_str
def _get_instance_label(self, scan_name, data_path):
ins_file = osp.join(data_path, scan_name + '_ins_label.npy')
assert osp.exists(ins_file)
return np.load(ins_file)
def _get_semantic_label(self, scan_name, data_path):
sem_file = osp.join(data_path, scan_name + '_sem_label.npy')
assert osp.exists(sem_file)
return np.load(sem_file)
tests/test_indoor_augment.py
View file @ 470a58ad
...@@ -5,114 +5,120 @@ from mmdet3d.datasets.pipelines.indoor_augment import (IndoorFlipData, ...@@ -5,114 +5,120 @@ from mmdet3d.datasets.pipelines.indoor_augment import (IndoorFlipData,
def test_indoor_flip_data(): def test_indoor_flip_data():
sunrgbd_indoor_flip_data = IndoorFlipData(0) np.random.seed(0)
sunrgbd_indoor_flip_data = IndoorFlipData()
sunrgbd_results = dict() sunrgbd_results = dict()
sunrgbd_results['point_cloud'] = np.array( sunrgbd_results['points'] = np.array(
[[1.02828765e+00, 3.65790772e+00, 1.97294697e-01, 1.61959505e+00], [[1.02828765e+00, 3.65790772e+00, 1.97294697e-01, 1.61959505e+00],
[-3.95979017e-01, 1.05465031e+00, -7.49204338e-01, 6.73096001e-01]]) [-3.95979017e-01, 1.05465031e+00, -7.49204338e-01, 6.73096001e-01]])
sunrgbd_results['gt_boxes'] = np.array([[ sunrgbd_results['gt_bboxes_3d'] = np.array([[
0.213684, 1.036364, -0.982323, 0.61541, 0.572574, 0.872728, 3.07028526 0.213684, 1.036364, -0.982323, 0.61541, 0.572574, 0.872728, 3.07028526
], ],
[ [
-0.449953, 1.395455, -1.027778, -0.449953, 1.395455,
1.500956, 1.637298, 0.636364, -1.027778, 1.500956,
-1.58242359 1.637298, 0.636364,
]]) -1.58242359
]])
sunrgbd_results = sunrgbd_indoor_flip_data(sunrgbd_results) sunrgbd_results = sunrgbd_indoor_flip_data(sunrgbd_results)
sunrgbd_point_cloud = sunrgbd_results.get('point_cloud', None) sunrgbd_points = sunrgbd_results.get('points', None)
sunrgbd_gt_boxes = sunrgbd_results.get('gt_boxes', None) sunrgbd_gt_bboxes_3d = sunrgbd_results.get('gt_bboxes_3d', None)
expected_sunrgbd_point_cloud = np.array( expected_sunrgbd_points = np.array(
[[-1.02828765, 3.65790772, 0.1972947, 1.61959505], [[-1.02828765, 3.65790772, 0.1972947, 1.61959505],
[0.39597902, 1.05465031, -0.74920434, 0.673096]]) [0.39597902, 1.05465031, -0.74920434, 0.673096]])
expected_sunrgbd_gt_boxes = np.array([[ expected_sunrgbd_gt_bboxes_3d = np.array([[
-0.213684, 1.036364, -0.982323, 0.61541, 0.572574, 0.872728, 0.07130739 -0.213684, 1.036364, -0.982323, 0.61541, 0.572574, 0.872728, 0.07130739
], [ ], [
0.449953, 1.395455, -1.027778, 1.500956, 1.637298, 0.636364, 4.72401624 0.449953, 1.395455, -1.027778, 1.500956, 1.637298, 0.636364, 4.72401624
]]) ]])
assert np.allclose(sunrgbd_point_cloud, expected_sunrgbd_point_cloud) assert np.allclose(sunrgbd_points, expected_sunrgbd_points)
assert np.allclose(sunrgbd_gt_boxes, expected_sunrgbd_gt_boxes) assert np.allclose(sunrgbd_gt_bboxes_3d, expected_sunrgbd_gt_bboxes_3d)
scannet_indoor_flip_data = IndoorFlipData(0) np.random.seed(0)
scannet_indoor_flip_data = IndoorFlipData()
scannet_results = dict() scannet_results = dict()
scannet_results['point_cloud'] = np.array( scannet_results['points'] = np.array(
[[1.6110241e+00, -1.6903955e-01, 5.8115810e-01, 5.9897250e-01], [[1.6110241e+00, -1.6903955e-01, 5.8115810e-01, 5.9897250e-01],
[1.3978075e+00, 4.2035791e-01, 3.8729519e-01, 4.0510958e-01]]) [1.3978075e+00, 4.2035791e-01, 3.8729519e-01, 4.0510958e-01]])
scannet_results['gt_boxes'] = np.array([[ scannet_results['gt_bboxes_3d'] = np.array([[
0.55903838, 0.48201692, 0.65688646, 0.65370704, 0.60029864, 0.5163464 0.55903838, 0.48201692, 0.65688646, 0.65370704, 0.60029864, 0.5163464
], [ ], [
-0.03226406, 1.70392646, 0.60348618, 0.65165804, 0.72084366, 0.64667457 -0.03226406, 1.70392646, 0.60348618, 0.65165804, 0.72084366, 0.64667457
]]) ]])
scannet_results = scannet_indoor_flip_data(scannet_results) scannet_results = scannet_indoor_flip_data(scannet_results)
scannet_point_cloud = scannet_results.get('point_cloud', None) scannet_points = scannet_results.get('points', None)
scannet_gt_boxes = scannet_results.get('gt_boxes', None) scannet_gt_bboxes_3d = scannet_results.get('gt_bboxes_3d', None)
expected_scannet_point_cloud = np.array( expected_scannet_points = np.array(
[[-1.6110241, 0.16903955, 0.5811581, 0.5989725], [[-1.6110241, 0.16903955, 0.5811581, 0.5989725],
[-1.3978075, -0.42035791, 0.38729519, 0.40510958]]) [-1.3978075, -0.42035791, 0.38729519, 0.40510958]])
expected_scannet_gt_boxes = np.array([[ expected_scannet_gt_bboxes_3d = np.array([[
-0.55903838, -0.48201692, 0.65688646, 0.65370704, 0.60029864, 0.5163464 -0.55903838, -0.48201692, 0.65688646, 0.65370704, 0.60029864, 0.5163464
], [ ], [
0.03226406, -1.70392646, 0.60348618, 0.65165804, 0.72084366, 0.64667457 0.03226406, -1.70392646, 0.60348618, 0.65165804, 0.72084366, 0.64667457
]]) ]])
assert np.allclose(scannet_point_cloud, expected_scannet_point_cloud) assert np.allclose(scannet_points, expected_scannet_points)
assert np.allclose(scannet_gt_boxes, expected_scannet_gt_boxes) assert np.allclose(scannet_gt_bboxes_3d, expected_scannet_gt_bboxes_3d)
def test_global_rot_scale(): def test_global_rot_scale():
sunrgbd_augment = IndoorGlobalRotScale(0, True, False, True, True) np.random.seed(0)
sunrgbd_augment = IndoorGlobalRotScale(True, False, True, True)
sunrgbd_results = dict() sunrgbd_results = dict()
sunrgbd_results['point_cloud'] = np.array( sunrgbd_results['points'] = np.array(
[[1.02828765e+00, 3.65790772e+00, 1.97294697e-01, 1.61959505e+00], [[1.02828765e+00, 3.65790772e+00, 1.97294697e-01, 1.61959505e+00],
[-3.95979017e-01, 1.05465031e+00, -7.49204338e-01, 6.73096001e-01]]) [-3.95979017e-01, 1.05465031e+00, -7.49204338e-01, 6.73096001e-01]])
sunrgbd_results['gt_boxes'] = np.array([[ sunrgbd_results['gt_bboxes_3d'] = np.array([[
0.213684, 1.036364, -0.982323, 0.61541, 0.572574, 0.872728, 3.07028526 0.213684, 1.036364, -0.982323, 0.61541, 0.572574, 0.872728, 3.07028526
], ],
[ [
-0.449953, 1.395455, -1.027778, -0.449953, 1.395455,
1.500956, 1.637298, 0.636364, -1.027778, 1.500956,
-1.58242359 1.637298, 0.636364,
]]) -1.58242359
]])
sunrgbd_results = sunrgbd_augment(sunrgbd_results) sunrgbd_results = sunrgbd_augment(sunrgbd_results)
sunrgbd_point_cloud = sunrgbd_results.get('point_cloud', None) sunrgbd_points = sunrgbd_results.get('points', None)
sunrgbd_gt_boxes = sunrgbd_results.get('gt_boxes', None) sunrgbd_gt_bboxes_3d = sunrgbd_results.get('gt_bboxes_3d', None)
expected_sunrgbd_point_cloud = np.array( expected_sunrgbd_points = np.array(
[[0.89427376, 3.94489646, 0.21003141, 1.72415094], [[0.89427376, 3.94489646, 0.21003141, 1.72415094],
[-0.47835783, 1.09972989, -0.79757058, 0.71654893]]) [-0.47835783, 1.09972989, -0.79757058, 0.71654893]])
expected_sunrgbd_gt_boxes = np.array([[ expected_sunrgbd_gt_bboxes_3d = np.array([[
0.17080999, 1.11345031, -1.04573864, 0.65513891, 0.60953755, 0.17080999, 1.11345031, -1.04573864, 0.65513891, 0.60953755,
0.92906854, 3.01916788 0.92906854, 3.01916788
], ],
[ [
-0.55427876, 1.45912611, -0.55427876, 1.45912611,
-1.09412807, 1.59785293, -1.09412807, 1.59785293,
1.74299674, 0.67744563, 1.74299674, 0.67744563,
-1.63354097 -1.63354097
]]) ]])
assert np.allclose(sunrgbd_point_cloud, expected_sunrgbd_point_cloud) assert np.allclose(sunrgbd_points, expected_sunrgbd_points)
assert np.allclose(sunrgbd_gt_boxes, expected_sunrgbd_gt_boxes) assert np.allclose(sunrgbd_gt_bboxes_3d, expected_sunrgbd_gt_bboxes_3d)
np.random.seed(0)
scannet_augment = IndoorGlobalRotScale( scannet_augment = IndoorGlobalRotScale(
0, True, False, False, True, rot_range=1 / 18) True, False, False, True, rot_range=[-np.pi * 1 / 36, np.pi * 1 / 36])
scannet_results = dict() scannet_results = dict()
scannet_results['point_cloud'] = np.array( scannet_results['points'] = np.array(
[[1.6110241e+00, -1.6903955e-01, 5.8115810e-01, 5.9897250e-01], [[1.6110241e+00, -1.6903955e-01, 5.8115810e-01, 5.9897250e-01],
[1.3978075e+00, 4.2035791e-01, 3.8729519e-01, 4.0510958e-01]]) [1.3978075e+00, 4.2035791e-01, 3.8729519e-01, 4.0510958e-01]])
scannet_results['gt_boxes'] = np.array([[ scannet_results['gt_bboxes_3d'] = np.array([[
0.55903838, 0.48201692, 0.65688646, 0.65370704, 0.60029864, 0.5163464 0.55903838, 0.48201692, 0.65688646, 0.65370704, 0.60029864, 0.5163464
], [ ], [
-0.03226406, 1.70392646, 0.60348618, 0.65165804, 0.72084366, 0.64667457 -0.03226406, 1.70392646, 0.60348618, 0.65165804, 0.72084366, 0.64667457
]]) ]])
scannet_results = scannet_augment(scannet_results) scannet_results = scannet_augment(scannet_results)
scannet_point_cloud = scannet_results.get('point_cloud', None) scannet_points = scannet_results.get('points', None)
scannet_gt_boxes = scannet_results.get('gt_boxes', None) scannet_gt_bboxes_3d = scannet_results.get('gt_bboxes_3d', None)
expected_scannet_point_cloud = np.array( expected_scannet_points = np.array(
[[1.61240576, -0.15530836, 0.5811581, 0.5989725], [[1.61240576, -0.15530836, 0.5811581, 0.5989725],
[1.39417555, 0.43225122, 0.38729519, 0.40510958]]) [1.39417555, 0.43225122, 0.38729519, 0.40510958]])
expected_scannet_gt_boxes = np.array([[ expected_scannet_gt_bboxes_3d = np.array([[
0.55491157, 0.48676213, 0.65688646, 0.65879754, 0.60584609, 0.5163464 0.55491157, 0.48676213, 0.65688646, 0.65879754, 0.60584609, 0.5163464
], [ ], [
-0.04677942, 1.70358975, 0.60348618, 0.65777559, 0.72636927, 0.64667457 -0.04677942, 1.70358975, 0.60348618, 0.65777559, 0.72636927, 0.64667457
]]) ]])
assert np.allclose(scannet_point_cloud, expected_scannet_point_cloud) assert np.allclose(scannet_points, expected_scannet_points)
assert np.allclose(scannet_gt_boxes, expected_scannet_gt_boxes) assert np.allclose(scannet_gt_bboxes_3d, expected_scannet_gt_bboxes_3d)
tests/test_indoor_loading.py deleted 100644 → 0
View file @ bdce5236
import mmcv
from mmdet3d.datasets.pipelines.indoor_loading import (LoadAnnotations3D,
LoadPointsFromFile)
def test_load_points_from_file():
sunrgbd_info = mmcv.load('./tests/data/sunrgbd/sunrgbd_infos.pkl')
sunrgbd_load_points_from_file = LoadPointsFromFile(False, True,
[0.5, 0.5, 0.5])
sunrgbd_results = dict()
sunrgbd_results['data_path'] = './tests/data/sunrgbd/sunrgbd_trainval'
sunrgbd_results['info'] = sunrgbd_info[0]
sunrgbd_results = sunrgbd_load_points_from_file(sunrgbd_results)
sunrgbd_point_cloud = sunrgbd_results.get('point_cloud', None)
assert sunrgbd_point_cloud.shape == (1000, 4)
scannet_info = mmcv.load('./tests/data/scannet/scannet_infos.pkl')
scannet_load_data = LoadPointsFromFile(False, True, [0.5, 0.5, 0.5])
scannet_results = dict()
scannet_results[
'data_path'] = './tests/data/scannet/scannet_train_instance_data'
scannet_results['info'] = scannet_info[0]
scannet_results = scannet_load_data(scannet_results)
scannet_point_cloud = scannet_results.get('point_cloud', None)
scannet_pcl_color = scannet_results.get('pcl_color', None)
assert scannet_point_cloud.shape == (1000, 4)
assert scannet_pcl_color.shape == (1000, 3)
def test_load_annotations3D():
sunrgbd_info = mmcv.load('./tests/data/sunrgbd/sunrgbd_infos.pkl')
sunrgbd_load_annotations3D = LoadAnnotations3D()
sunrgbd_results = dict()
sunrgbd_results['data_path'] = './tests/data/sunrgbd/sunrgbd_trainval'
sunrgbd_results['info'] = sunrgbd_info[0]
sunrgbd_results = sunrgbd_load_annotations3D(sunrgbd_results)
sunrgbd_gt_boxes = sunrgbd_results.get('gt_boxes', None)
sunrgbd_gt_classes = sunrgbd_results.get('gt_classes', None)
sunrgbd_gt_boxes_mask = sunrgbd_results.get('gt_boxes_mask', None)
assert sunrgbd_gt_boxes.shape == (3, 7)
assert sunrgbd_gt_classes.shape == (3, 1)
assert sunrgbd_gt_boxes_mask.shape == (3, 1)
scannet_info = mmcv.load('./tests/data/scannet/scannet_infos.pkl')
scannet_load_annotations3D = LoadAnnotations3D()
scannet_results = dict()
scannet_results[
'data_path'] = './tests/data/scannet/scannet_train_instance_data'
scannet_results['info'] = scannet_info[0]
scannet_results = scannet_load_annotations3D(scannet_results)
scannet_gt_boxes = scannet_results.get('gt_boxes', None)
scannet_gt_classes = scannet_results.get('gt_classes', None)
scannet_gt_boxes_mask = scannet_results.get('gt_boxes_mask', None)
scannet_instance_labels = scannet_results.get('instance_labels', None)
scannet_semantic_labels = scannet_results.get('semantic_labels', None)
assert scannet_gt_boxes.shape == (27, 6)
assert scannet_gt_classes.shape == (27, 1)
assert scannet_gt_boxes_mask.shape == (27, 1)
assert scannet_instance_labels.shape == (1000, )
assert scannet_semantic_labels.shape == (1000, )
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment