-
Notifications
You must be signed in to change notification settings - Fork 0
/
evaluate_recon.py
172 lines (159 loc) · 8.85 KB
/
evaluate_recon.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
import os.path
import pandas as pd
import open3d as o3d
import numpy as np
import pybullet as p
import xml.dom.minidom as xmldom
import pymeshlab
def parse_xml(fn):
rpy = {}
xyz = {}
xml_file = xmldom.parse(fn)
rootNode = xml_file.documentElement
print(rootNode.nodeName)
links = rootNode.getElementsByTagName("link")
for link in links:
if link.hasAttribute("name") and (link.getAttribute("name")=="baseLink" or link.getAttribute("name")=="baselink"):
inertial = link.getElementsByTagName("inertial")[0]
visual = link.getElementsByTagName("visual")[0]
if len(inertial.getElementsByTagName("origin"))>0 :
origin = inertial.getElementsByTagName("origin")[0]
temp = list(map(float, origin.getAttribute("rpy").split(' ')))
rpy['inertial'] = np.asarray(p.getMatrixFromQuaternion(p.getQuaternionFromEuler(temp))).reshape([3, 3])
temp = list(map(float, origin.getAttribute("xyz").split(' ')))
xyz['inertial'] = np.asarray(temp)
if len(visual.getElementsByTagName("origin")) > 0:
origin = visual.getElementsByTagName("origin")[0]
temp = list(map(float, origin.getAttribute("rpy").split(' ')))
rpy['visual'] = np.asarray(p.getMatrixFromQuaternion(p.getQuaternionFromEuler(temp))).reshape([3, 3])
temp = list(map(float, origin.getAttribute("xyz").split(' ')))
xyz['visual'] = np.asarray(temp)
break
print(rpy, xyz)
return rpy, xyz
def compute_unsigned_distance_and_closest_goemetry(scene, query_points: np.ndarray):
closest_points = scene.compute_closest_points(query_points)
distance = np.linalg.norm(query_points - closest_points['points'].numpy(),
axis=-1)
return distance, closest_points['geometry_ids'].numpy()
def compute_signed_distance_and_closest_goemetry(scene, query_points: np.ndarray):
closest_points = scene.compute_closest_points(query_points)
distance = np.linalg.norm(query_points - closest_points['points'].numpy(),
axis=-1)
rays = np.concatenate([query_points, np.ones_like(query_points)], axis=-1)
intersection_counts = scene.count_intersections(rays).numpy()
is_inside = intersection_counts % 2 == 1
distance[is_inside] *= -1
return distance, closest_points['geometry_ids'].numpy()
YcbObjects = ["YcbMustardBottle",
"YcbGelatinBox",
"YcbPottedMeatCan",
"YcbTomatoSoupCan",
"YcbCrackerBox",
"YcbSugarBox",
"YcbBanana",
"YcbTennisBall"]
DataPath = ["Data", "Data_stuck"]
SegMethods = ["BackFlow", "OSVOS"]
ReconMethods = ["point-to-plane", "robot-joints"]
if __name__ == '__main__':
mean_usdf = []
mean_sdf = []
mean_squared_sdf = []
mean_hausdorff_distance = []
min_hausdorff_distance = []
max_hausdorff_distance = []
RMS_hausdorff_distance =[]
obj_record = []
data_record = []
seg_record = []
recon_record = []
coordinate = o3d.geometry.TriangleMesh.create_coordinate_frame(size=0.3, origin=[0, 0, 0])
for obj in YcbObjects:
if obj == "YcbSugarBox":
mesh_path = os.path.join(os.path.dirname(os.path.abspath(os.getcwd())),
"pybullet-object-models", "pybullet_object_models",
"ycb_objects", obj, "textured.obj")
elif obj == "YcbTennisBall":
mesh_path = os.path.join(os.path.dirname(os.path.abspath(os.getcwd())),
"pybullet-object-models", "pybullet_object_models",
"ycb_objects", obj, "textured_reoriented.obj")
else:
mesh_path = os.path.join(os.path.dirname(os.path.abspath(os.getcwd())),
"pybullet-object-models", "pybullet_object_models",
"ycb_objects", obj, "textured_simple_reoriented.obj")
urdf_path = os.path.join(os.path.dirname(os.path.abspath(os.getcwd())),
"pybullet-object-models", "pybullet_object_models",
"ycb_objects", obj, "model.urdf")
rpy, xyz = parse_xml(urdf_path)
mesh = o3d.io.read_triangle_mesh(mesh_path, True)
ms = pymeshlab.MeshSet()
ms.load_new_mesh(mesh_path)
scene = o3d.t.geometry.RaycastingScene()
mesh_cpu = o3d.t.geometry.TriangleMesh.from_legacy(mesh)
_ = scene.add_triangles(mesh_cpu)
count = 0
for d in DataPath:
for s in SegMethods:
for r in ReconMethods:
count += 1
print(d, obj, s, r)
path = os.path.join(os.getcwd(), d, obj)
pcd = o3d.io.read_point_cloud(os.path.join(path, obj + '_' + s + '_' + r + '.ply'))
pcd1 = o3d.io.read_point_cloud(os.path.join(path, obj + '_' + s + '_' + r + '.ply'))
robot_joints = pd.read_csv(path + "/robot_joints.csv", index_col='filenames')
ObjectR = robot_joints['ObjectRotationMatrices'][1]
ObjectPositions = robot_joints['ObjectPositions'][1]
ObjectOrientation = list(map(float, ObjectR[1:len(ObjectR) - 1].split(',')))
ObjectOrientation = np.reshape(np.array(ObjectOrientation), [3, 3])
ObjectPositions = np.array( list(map(float, ObjectPositions[1:len(ObjectPositions) - 1].split(','))))
# o3d.visualization.draw_geometries([mesh,pcd1,coordinate])
points = ( np.dot(np.linalg.inv(ObjectOrientation), (np.asarray(pcd.points) - ObjectPositions).T) ).T
if 'inertial' in rpy:
points = ( np.dot( rpy['inertial'], points.T)).T + xyz['inertial']
if 'visual' in rpy:
points = (np.dot(np.linalg.inv(rpy['visual']), (points- xyz['visual']).T )).T
pcd.points = o3d.utility.Vector3dVector(points)
o3d.io.write_point_cloud(os.path.join(path, obj + '_' + s + '_' + r + '_origin.ply'), pcd)
ms.load_new_mesh(os.path.join(path, obj + '_' + s + '_' + r + '_origin.ply'))
temp_haus_dis = ms.hausdorff_distance(sampledmesh=0, targetmesh=count, samplenum=8000)
ms.set_current_mesh(new_curr_id=count)
ms.delete_current_mesh()
mean_hausdorff_distance.append(temp_haus_dis['mean'])
min_hausdorff_distance.append(temp_haus_dis['min'])
max_hausdorff_distance.append(temp_haus_dis['max'])
RMS_hausdorff_distance.append(temp_haus_dis['RMS'])
print("mean hausdorff distance ", mean_hausdorff_distance[-1])
print("rms hausdorff distance ", RMS_hausdorff_distance[-1])
# points = pcd.points + np.asarray([0.2,0,0])
# pcd1.points = o3d.utility.Vector3dVector(points)
# o3d.visualization.draw_geometries([mesh,pcd1,coordinate])
# compute range
usdf, _ = compute_unsigned_distance_and_closest_goemetry(scene, query_points = np.asarray(pcd.points).astype(np.float32))
sdf, _ = compute_signed_distance_and_closest_goemetry(scene, query_points = np.asarray(pcd.points).astype(np.float32))
temp_usdf = np.mean(usdf)
temp_sdf = np.mean(sdf)
temp_ssdf = np.mean(np.square(sdf))
print("mean usdf ", temp_usdf)
print("mean sdf ", temp_sdf)
print('mean squared sdf', temp_ssdf)
mean_usdf.append(temp_usdf)
mean_sdf.append(temp_sdf)
mean_squared_sdf.append(temp_ssdf)
obj_record.append(obj)
data_record.append(d)
seg_record.append(s)
recon_record.append(r)
result_df = pd.DataFrame({'Data Path': data_record,
'Object Name': obj_record,
'Segmentation Method': seg_record,
'Reconstruction Method': recon_record,
'mean usdf': mean_usdf,
'mean sdf': mean_sdf,
'mean squared sdf': mean_squared_sdf,
'mean haus dist': mean_hausdorff_distance,
'min haus dist': min_hausdorff_distance,
'max haus dist': max_hausdorff_distance,
'RMS haus dist': RMS_hausdorff_distance
})
result_df.to_csv(os.path.join(os.getcwd(), "recon_comparison.csv"), index=False)