forked from huahuasousou/dair_v2x_i_dataset_vis
-
Notifications
You must be signed in to change notification settings - Fork 0
/
convert.py
198 lines (162 loc) · 9.03 KB
/
convert.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
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
import os, sys
import shutil
import scipy.io as scio
import sys
import yaml
import shutil
import numpy as np
import pandas as pd
def read_config(config_path='./config/config.yaml'):
file = open(config_path, 'r', encoding="utf-8")
#读取文件中的所有数据
file_data = file.read()
file.close()
#指定Loader
config_data = yaml.load(file_data,Loader=yaml.FullLoader)
return config_data
class convert_datasets:
def __init__(self,config_data):
np.set_printoptions(linewidth=1000)#防止numpy转换str自动换行
self.output_path = os.path.join(config_data['output_path'],config_data['output_floder_name'])
#self.data_info_path=os.path.join(config_data['root_path'],config_data['data_info'])
#self.data_info=read_config(data_info_path)
self.source_image_path=os.path.join(config_data['root_path'],config_data['image_floder'])
self.source_velodyne_path=os.path.join(config_data['root_path'],config_data['velodyne_floder'])
self.source_calib_path=os.path.join(config_data['root_path'],config_data['calib_floder'])
self.source_camera_Tr_path=os.path.join(config_data['root_path'],config_data['lidar_to_camera_floder'])
self.source_camera_label=os.path.join(config_data['root_path'],config_data['camera_label_floder'])
self.source_velodyne_label=os.path.join(config_data['root_path'],config_data['virtuallidar_label_floder'])
self.calib_path=os.path.join(self.output_path,"calib")
self.image2_path=os.path.join(self.output_path,"image_2")
self.label2_path=os.path.join(self.output_path,"label_2")
self.label_vel_path=os.path.join(self.output_path,"label_velodyne")
self.velodyne_path=os.path.join(self.output_path,"velodyne")
self.create_floders_list=[]
self.create_floders_list.append(self.calib_path)
#self.create_floders_list.append(self.image2_path)
self.create_floders_list.append(self.label2_path)
self.create_floders_list.append(self.label_vel_path)
#self.create_floders_list.append(self.velodyne_path)
#print("create_floders list:",self.create_floders_list)
#self.files_list = os.listdir(self.source_image_path)
#self.files_list.sort()
#self.files_len=len(self.files_list)
#创建目录，返回路径
def create_path(self):
for floder in self.create_floders_list:
if not os.path.exists(floder):
os.makedirs(floder)
print("created the path:",floder)
def check_files(self):
#check image, velodyne and label are correspond.
pass
def calib_convert(self):
output_path=self.calib_path
calib_input_path=self.source_calib_path
tr_input_path=self.source_camera_Tr_path
self.calib_list = os.listdir(self.source_calib_path)
self.calib_list.sort() #no neceressory, save time
for index in self.calib_list:
calib=read_config(os.path.join(calib_input_path,index))
tr=read_config(os.path.join(tr_input_path,index))
rotation_matrix=np.mat(tr["rotation"])
translation_matrix=np.mat(tr["translation"])
Tr_matrix=np.c_[rotation_matrix,translation_matrix]
Tr_matrix=Tr_matrix.flatten()
Tr_matrix=str(Tr_matrix)
Tr_matrix=Tr_matrix.strip("[]")#delete "[" and "]"
Tr_matrix=Tr_matrix.replace(" "," ")# replace the "," to " "
#matrix="P2: "+matrix
Tr_matrix="Tr_velo_to_cam: "+Tr_matrix
#print(index,Tr_matrix)
#P2_matrix=str(calib["P"])#not P matrix
P2_matrix=str(calib["cam_K"])# should use cam_K matrix
P2_matrix=P2_matrix.strip("[]")
P2_matrix=P2_matrix.replace(",","")# replace the "," to " "
P2_matrix="P2: "+P2_matrix
output_name = os.path.splitext(index)
with open(os.path.join(output_path,output_name[0]+".txt"), 'wt') as f:
print("write file:",os.path.join(output_path,output_name[0]+".txt"))
f.write("P0: \n")
f.write("P1: \n")
f.write(P2_matrix+"\n")
f.write("P3: \n")
f.write("R0_rect: \n")
f.write(Tr_matrix)
def label_convert(self,input_path,output_path):
files_list = os.listdir(input_path)
files_list.sort()
#print(files_list)
for index in files_list:
label=read_config(os.path.join(input_path,index))
output_name = os.path.splitext(index)
with open(os.path.join(output_path,output_name[0]+".txt"), 'wt') as f:
print("write file:",os.path.join(output_path,output_name[0]+".txt"))
for line in label:
#print("line: ",line,"\n")
f.write(line['type']+" ")
f.write(line['truncated_state']+" ")
f.write(line['occluded_state']+" ")
f.write(line['alpha']+" ")
f.write(line['2d_box']['xmin']+" ")
f.write(line['2d_box']['ymin']+" ")
f.write(line['2d_box']['xmax']+" ")
f.write(line['2d_box']['ymax']+" ")
f.write(line['3d_dimensions']['h']+" ")
f.write(line['3d_dimensions']['w']+" ")
f.write(line['3d_dimensions']['l']+" ")
f.write(line['3d_location']['x']+" ")
f.write(line['3d_location']['y']+" ")
f.write(line['3d_location']['z']+" ")
f.write(line['rotation']+" "+"\n")
def label_convert_fix_high(self,camera_label_path,lidar_label_path,output_path):
lidar_files_list = os.listdir(lidar_label_path)
lidar_files_list.sort()
camera_files_list = os.listdir(camera_label_path)
camera_files_list.sort()
#print(files_list)
for index in lidar_files_list:
lidar_label=read_config(os.path.join(lidar_label_path,index))
camera_label=read_config(os.path.join(camera_label_path,index))
output_name = os.path.splitext(index)
with open(os.path.join(output_path,output_name[0]+".txt"), 'wt') as f:
print("write file:",os.path.join(output_path,output_name[0]+".txt"))
for camera_line,lidar_line in zip(camera_label,lidar_label):
#print("camera_line: ",camera_line,"\n")
f.write(camera_line['type']+" ")
f.write(camera_line['truncated_state']+" ")
f.write(camera_line['occluded_state']+" ")
f.write(camera_line['alpha']+" ")
f.write(camera_line['2d_box']['xmin']+" ")
f.write(camera_line['2d_box']['ymin']+" ")
f.write(camera_line['2d_box']['xmax']+" ")
f.write(camera_line['2d_box']['ymax']+" ")
f.write(lidar_line['3d_dimensions']['h']+" ")
f.write(lidar_line['3d_dimensions']['w']+" ")
f.write(lidar_line['3d_dimensions']['l']+" ")
f.write(camera_line['3d_location']['x']+" ")
f.write(camera_line['3d_location']['y']+" ")
f.write(lidar_line['3d_location']['z']+" ")
f.write(camera_line['rotation']+" "+"\n")
def rename_floder(self,input_path,output_path):
print("copy file from ",input_path,"to the ",output_path)
shutil.copytree(input_path,output_path)
if __name__ == '__main__':
#You can select which folders you need to convert by commenting out parts of the code
config_data=read_config('./config/config.yaml') #read config file get the convert input and output path
dair2kitti=convert_datasets(config_data) #instantiate the object
dair2kitti.create_path() #create the subfloders
dair2kitti.calib_convert() #calib file convert to the kitti format
#camera coordinate label convert, but origin camera label have some issues, so don't use this.
#dair2kitti.label_convert(dair2kitti.source_camera_label,dair2kitti.label2_path)
#camera coordinate label convert, and fix the camera label issues
#base the camera label source, and replace the pos(z) and the size(l,w,h) by the lidar label source.
dair2kitti.label_convert_fix_high(dair2kitti.source_camera_label,dair2kitti.source_velodyne_label,dair2kitti.label2_path)
#velodyne coordinate label convert
dair2kitti.label_convert(dair2kitti.source_velodyne_label,dair2kitti.label_vel_path)
#Make sure the destination folder does not exist before converting
#copy and rename the camera image floder
dair2kitti.rename_floder(dair2kitti.source_image_path,dair2kitti.image2_path)
#copy and rename the velodyne pcd floder
dair2kitti.rename_floder(dair2kitti.source_velodyne_path,dair2kitti.velodyne_path)
#convert finished