-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_same_z_linear.py
222 lines (202 loc) · 8.46 KB
/
test_same_z_linear.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
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
import numpy as np
import time, sys, os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
os.environ["CUDA_VISIBLE_DEVICES"] = "2"
# 0 = all messages are logged(default behavior)
# 1 = INFO messages are not printed
# 2 = INFO and WARNING messages arenot printed
# 3 = INFO, WARNING, and ERROR messages arenot printed
import src.dataset_loader.KITTI_dataset_p2 as kitti
import src.net_core.darknet as Darknet
import src.module.nolbo_test as nolbo
import tensorflow as tf
import cv2
tf.get_logger().warning('test')
# WARNING:tensorflow:test
tf.get_logger().setLevel('ERROR')
tf.get_logger().warning('test')
imageSizeAndBatchListKITTI = [
# # [384,128,64],
# # [480,160,64],
# [576,192,50,16],
# [672,224,50,14],
# [768,256,50,14],
# [864,288,48,14],
# [960,288,48,12],
# [960,320,48,10],
# [960,352,38,10],
# [1056,352,42,10],
# [1088,352,42,10],
# [1088,384,40,10],
# [1152,384,40,9],
[1216,384,128,10],
# [1216,448,40,9],
# [1248,384,42,8],
# [1248,448,42,8],
# # #############################
# [160,128,64],
# [288,288,56],
# [320,256,48],
# [320,320,48],
# [416,416,40],
# [480,384,40],
# [448,448,40],
# [640,512,32],
# [800,640,24],
]
# imageSizeAndBatchListKITTI = [
# # [384,128,14],
# # [480,160,12],
# # [576,192,12],
# # [672,224,10],
# # [768,256,10],
# # [864,288,40],
# # [960,288,38],
# # [960,320,36],
# # [1056,352,36],
# [1152,384,32],
# [1216,384,32],
# ]
def train(
training_epoch = 1000,
learning_rate = 1e-4,
config = None, predictor_num=5,
save_path = None, load_path = None,
load_encoder_backbone_path = None, load_encoder_backbone_name = None,
load_decoder_path = None, load_decoder_name = None,
):
data_loader = kitti.dataLoader(predNumPerGrid=predictor_num, trainOrVal='val', is_train=False)
model = nolbo.nolbo_test(nolbo_structure=config,
backbone_style=Darknet.Darknet19)
# model = nolbo.nolbo(nolbo_structure=config,
# backbone_style=Darknet.Darknet19,
# learning_rate=learning_rate,
# IoU2D_loss=False, IoU3D_loss=False, exp=False,
# sqr_scale=True)
if load_path != None:
print('load weights...')
model.loadModel(load_path=load_path)
print('done!')
epoch, epoch_curr = 0, 0
print('start training...')
KITTIImagePath = '/home/yonsei/dataset/KITTI/data_object_image_2/training/image_2/'
KITTI_calib_path = '/home/yonsei/dataset/KITTI/data_object_calib/training/calib/'
KITTIDataListPath = '/home/yonsei/dataset/KITTI/trainvalsplit_3DOP_MONO3D/'
valPath = os.path.join(KITTIDataListPath, 'val.txt')
KITTIDataListPath = np.loadtxt(valPath, dtype='str')
for index, KITTIData in enumerate(KITTIDataListPath):
with open(os.path.join(KITTI_calib_path, KITTIData + '.txt')) as fp:
calib = fp.readlines()
P2 = np.array(calib[2].split(' ')[1:]).astype('float64')
projection_mat = np.identity(4)
projection_mat_inv = np.identity(4)
for i in range(3):
for j in range(4):
projection_mat[i, j] = P2[4 * i + j]
P2_pinv = np.linalg.pinv(projection_mat)
for i in range(4):
for j in range(3):
projection_mat_inv[i, j] = P2_pinv[i, j]
inputImage = cv2.imread(os.path.join(KITTIImagePath, KITTIData + '.png'), cv2.IMREAD_COLOR)
image_size = inputImage.shape[:2]
inputImage_net = cv2.resize(inputImage, (1216,384))
inputImage_net = inputImage_net/255.0
gridSize = int(1216 / 32), int(384 / 32)
anchor_path = '/home/yonsei/pyws/NOLBO_grid_anchor_no3Ddec/trash/NOLBO_2D_3D_IoU_exp_viewing/src/dataset_loader/anchor_same_z_'+str(predictor_num)+'/'
anchor_global_path = '/home/yonsei/pyws/NOLBO_grid_anchor_no3Ddec/trash/NOLBO_2D_3D_IoU_exp_viewing/src/dataset_loader/anchor_global_'+str(predictor_num)+'/'
anchor_z_global = np.load(anchor_global_path + 'global_anchor_xyz_s.npy')
anchor_bbox3D_global = np.load(anchor_global_path + 'global_anchor_bbox3D_s.npy')
anchor_xyz = np.load(anchor_path + 'anchor_xyz_{:d}{:d}.npy'.format(gridSize[0], gridSize[1]))
anchor_xyz[:,:,:,-1] = anchor_z_global[:,-1]
anchor_ry = np.load(anchor_path + 'anchor_ry_{:d}{:d}.npy'.format(gridSize[0], gridSize[1]))
anchor_alpha = np.load(anchor_path + 'anchor_alpha_{:d}{:d}.npy'.format(gridSize[0], gridSize[1]))
anchor_bbox3D = np.load(anchor_path + 'anchor_bbox3D_{:d}{:d}.npy'.format(gridSize[0], gridSize[1]))
anchor_bbox3D[:,:,:,:] = anchor_bbox3D_global[:,:]
inputImages = np.array([inputImage_net])
P2List = np.array([projection_mat])
P2InvList = np.array([projection_mat_inv])
objsPose, objsBbox3DSize, bbox3D8Points, objsBbox2D, rys, alphas, objness = \
model.getPred(
input_image=inputImages,
P2_gt=P2List, P2_inv_gt=P2InvList,
image_size=image_size, obj_thresh=0.01, IOU_thresh=0.1, is_exp=False,
anchor_xyz=anchor_xyz, anchor_bbox3D=anchor_bbox3D, anchor_theta=anchor_ry
)
def draw2Dbbox(image, bbox2d, color=(0, 255, 0), thickness=2):
p0 = (bbox2d[0], bbox2d[1])
p1 = (bbox2d[2], bbox2d[3])
cv2.rectangle(image, p0, p1, color=color, thickness=thickness)
# print(imageSizeList[0,0,0,0])
# img = cv2.resize(inputImages[0]*255, (int(imageSizeList[0,0,0,0,1]), int(imageSizeList[0,0,0,0,0])))
fp = open(save_path + 'data/{:06d}.txt'.format(index), "w")
print(index, KITTIData)
if len(objsPose) > 0:
for pose, b3DDim, box2D, alpha, ry, objn in zip(objsPose, objsBbox3DSize, objsBbox2D, alphas, rys, objness):
# print(file_name_list[0])
txtline = [
alpha[0],
box2D[0], box2D[1], box2D[2], box2D[3],
b3DDim[0], b3DDim[1], b3DDim[2],
pose[0,-1], pose[1,-1], pose[2,-1],
ry[0], objn[0]
]
# print(txtline)
fp.write("Car {:.2f} {} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f}\n".format(
0., int(0),
alpha[0],
box2D[0], box2D[1], box2D[2], box2D[3],
b3DDim[1], b3DDim[2], b3DDim[0], #l,h,w to h,w,l
pose[0,-1], pose[1,-1] + b3DDim[1]/2., pose[2,-1], #y -> y + h/2
ry[0], objn[0]))
draw2Dbbox(inputImage, box2D)
fp.close()
# print(save_path + 'image/' + file_name_list[0] + '.png')
cv2.imwrite(save_path + 'image/' + KITTIData + '.png', inputImage)
predictor_num = 12
# latent_dim = 16
# inst_num = 10
config = {
'encoder_backbone':{
'name' : 'nolbo_backbone',
'predictor_num': predictor_num,
'bbox2DXY_dim':2,
'bbox3D_dim':3, 'orientation_dim':1,
'localXYZ_dim':1,
'activation' : 'elu',
},
'encoder_head':{
'name' : 'nolbo_head',
'output_dim' : predictor_num*(1+2+3+(2*1+2)+(2+1)), # objness + b2Dxy + lhw + (2*z + logvarxy) + (sincos+var)
'filter_num_list':[1024,1024,1024,1024,512,512,512],
'filter_size_list':[3,3,3,3,1,1,1],
'activation':'elu',
},
# 'decoder':{
# 'name':'decoder',
# 'input_dim' : latent_dim,
# 'output_shape':[64,64,64,1],
# 'filter_num_list':[512,256,128,64,1],
# 'filter_size_list':[4,4,4,4,4],
# 'strides_list':[1,2,2,2,2],
# 'activation':'elu',
# 'final_activation':'sigmoid'
# },
# 'prior' : {
# 'name' : 'priornet',
# 'input_dim' : inst_num, # class num (one-hot vector)
# 'unit_num_list' : [32, 16],
# 'core_activation' : 'elu',
# 'const_log_var' : 0.0,
# }
}
if __name__ == '__main__':
sys.exit(train(
training_epoch=1000, learning_rate=1e-4,
config=config, predictor_num=predictor_num,
save_path='./data/same_z_linear/',
load_path='./weights_dist_exp/same_z_linear/',
# load_encoder_backbone_path='./weights/imagenet_and_place365_and_Object/',
# load_encoder_backbone_name='imagenet_backbone',
# load_decoder_path='./weights/imagenet_and_place365_and_Object/',
# load_decoder_name='decoder',
))