-
Notifications
You must be signed in to change notification settings - Fork 16
/
draw_rebuttal_fig.py
160 lines (145 loc) · 8.55 KB
/
draw_rebuttal_fig.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
import os
from utils import draw_figures
import numpy as np
import utils.se3 as se3
import matplotlib.pyplot as plt
from matplotlib.font_manager import FontProperties
from matplotlib.ticker import FuncFormatter
import torch
def to_percent(y,position):
return str(int(100*y))+"%"#这里可以用round()函数设置取几位小数
#好的图需要手动存
def show_graph(pc, node, size=2, theta_1=0, theta_2=0, c0='#b30000',savename=None, close=True):
if type(pc) == torch.Tensor:
pc = pc.detach().cpu().numpy()
if pc.shape[0]==3:
pc = pc.transpose()
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1, projection='3d')
ax.scatter(pc[:,0], pc[:,1], pc[:,2],s=size, alpha=1, c=c0)
plt.axis("off")
ax.view_init(theta_1, theta_2)
ax.auto_scale_xyz([-1,1],[-1,1],[-1,1])
for i in range(len(node[0])):
ax.plot([pc[node[0]][i,0], pc[node[1]][i,0]],
[pc[node[0]][i,1], pc[node[1]][i,1]],
[pc[node[0]][i,2], pc[node[1]][i,2]], color='b', alpha=0.4)
plt.show(fig)
if savename is not None:
plt.savefig(savename, format='svg', bbox_inches='tight', transparent=True, dpi=600)
if close:
plt.close(fig)
def show_2graph(pc0, pc1, node0, node1, size=1, theta_1=0, theta_2=0, c0='#1f4f89', c1='#b30000', savename=None, close=True):
if type(pc0) == torch.Tensor:
pc0 = pc0.detach().cpu().numpy()
if pc0.shape[0]==3:
pc0 = pc0.transpose()
if type(pc1) == torch.Tensor:
pc1 = pc1.detach().cpu().numpy()
if pc1.shape[0]==3:
pc1 = pc1.transpose()
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1, projection='3d')
ax.scatter(pc0[:, 0], pc0[:, 1], pc0[:, 2], s=size, c=c0, alpha=1)
ax.scatter(pc1[:, 0], pc1[:, 1], pc1[:, 2], s=size, c=c1, alpha=1)
for i in range(len(node0[0])):
ax.plot([pc0[node0[0]][i,0], pc0[node0[1]][i,0]],
[pc0[node0[0]][i,1], pc0[node0[1]][i,1]],
[pc0[node0[0]][i,2], pc0[node0[1]][i,2]], color=c0, linewidth = 1)
for i in range(len(node1[0])):
ax.plot([pc1[node1[0]][i,0], pc1[node1[1]][i,0]],
[pc1[node1[0]][i,1], pc1[node1[1]][i,1]],
[pc1[node1[0]][i,2], pc1[node1[1]][i,2]], color=c1, linewidth = 1)
plt.axis("off")
ax.view_init(theta_1, theta_2)
ax.auto_scale_xyz([-1,1],[-1,1],[-1,1])
# plt.show(fig)
if savename is not None:
plt.savefig(savename, format='pdf', bbox_inches='tight', transparent=True, dpi=1200)
if close:
plt.show(fig)
plt.close(fig)
exp=2
ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
if exp==2:
data = np.load(os.path.join(ROOT_DIR,'creat_data/PGM_DGCNN_3dmatchSeen_NoPreW[\'xyz\', \'gxyz\']_attentiontransformer_crop/realdata_30.npy'),allow_pickle=True).item()
pre_result=np.load(os.path.join(ROOT_DIR,'output/PGM_DGCNN_3dmatchSeen_NoPreW[\'xyz\', \'gxyz\']_attentiontransformer_crop/eval_log_2021-01-25-13-26-36_metric.npy'),allow_pickle=True).item()
plt.figure() # 图片长宽和清晰度
plt.hist(pre_result['r_mae'],50,weights=[1./len(pre_result['r_mae'])]*len(pre_result['r_mae']))
fomatter=FuncFormatter(to_percent)
plt.gca().yaxis.set_major_formatter(fomatter)
plt.title("Distribution of Rotation Error", fontdict={'family' : 'Times New Roman', 'size': 16})
plt.ylabel('Ratio', fontdict={'family' : 'Times New Roman', 'size' : 13})
plt.xlabel('Rotation Error (degree)', fontdict={'family' : 'Times New Roman', 'size' : 13})
plt.yticks(fontproperties = 'Times New Roman', size = 13)
plt.xticks(fontproperties = 'Times New Roman', size = 13)
plt.savefig('fig/3dmatch/hist_r.svg', format='svg', bbox_inches='tight', transparent=True, dpi=1200)
plt.show()
plt.figure() # 图片长宽和清晰度
plt.hist(pre_result['t_mae'],50,weights=[1./len(pre_result['t_mae'])]*len(pre_result['t_mae']))
fomatter=FuncFormatter(to_percent)
plt.gca().yaxis.set_major_formatter(fomatter)
plt.title("Distribution of Translation Error", fontdict={'family' : 'Times New Roman', 'size': 16})
plt.ylabel('Ratio', fontdict={'family' : 'Times New Roman', 'size' : 13})
plt.xlabel('Translation Error (m)', fontdict={'family' : 'Times New Roman', 'size' : 13})
plt.yticks(fontproperties = 'Times New Roman', size = 13)
plt.xticks(fontproperties = 'Times New Roman', size = 13)
plt.savefig('fig/3dmatch/hist_t.svg', format='svg', bbox_inches='tight', transparent=True, dpi=1200)
plt.show()
# fig = plt.figure()
# ax1 = fig.add_subplot(1, 2, 1)
#
# ax1.hist(pre_result['r_mae'],50,weights=[1./len(pre_result['r_mae'])]*len(pre_result['r_mae']))
# fomatter=FuncFormatter(to_percent)
# plt.gca().yaxis.set_major_formatter(fomatter)
# plt.title("Distribution of Rotation Error", fontdict={'family' : 'Times New Roman', 'size': 16})
# plt.ylabel('Ratio', fontdict={'family' : 'Times New Roman', 'size' : 13})
# plt.xlabel('Rotation Error $^\circ$', fontdict={'family' : 'Times New Roman', 'size' : 13})
# plt.yticks(fontproperties = 'Times New Roman', size = 13)
# plt.xticks(fontproperties = 'Times New Roman', size = 13)
#
# ax2 = fig.add_subplot(1, 2, 2)
# ax2.hist(pre_result['t_mae'],50,weights=[1./len(pre_result['t_mae'])]*len(pre_result['t_mae']))
# fomatter=FuncFormatter(to_percent)
# plt.gca().yaxis.set_major_formatter(fomatter)
# plt.title("Distribution of Translation Error", fontdict={'family' : 'Times New Roman', 'size': 16})
# plt.ylabel('Ratio', fontdict={'family' : 'Times New Roman', 'size' : 13})
# plt.xlabel('Translation Error $^\circ$', fontdict={'family' : 'Times New Roman', 'size' : 13})
# plt.yticks(fontproperties = 'Times New Roman', size = 13)
# plt.xticks(fontproperties = 'Times New Roman', size = 13)
# plt.savefig('fig/3dmatch/hist_t.pdf', format='pdf', bbox_inches='tight', transparent=True, dpi=1200)
# plt.show()
# print(np.where((pre_result['r_mae'] < 1) * (pre_result['t_mae'] < 0.1)))
# # print(pre_result['r_mae'][(pre_result['r_mae'] < 1) * (pre_result['t_mae'] < 0.1)])
#
# pc_index = 505
# point_size = 0.1
# theta_2 = 90
# theta_1 = 106
# c0 = '#1f4f89'
# c1 = '#b30000'
#
# draw_figures.show_pointcloud_2part(data['p1'][pc_index], data['p2'][pc_index], c0=c0,c1=c1, size=point_size, theta_1=theta_1, theta_2=theta_2, savename ='fig/3dmatch/_before' + str(pc_index) + '.svg')
# draw_figures.show_pointcloud_2part(se3.transform(pre_result['pre_transform'][pc_index], data['p1'][pc_index, :, :3]), data['p2'][pc_index], c0=c0,c1=c1, size=point_size, theta_1=theta_1, theta_2=theta_2, savename ='fig/3dmatch/_reg' + str(pc_index) + '.svg')
# draw_figures.show_pointcloud_2part(se3.transform(pre_result['gt_transform'][pc_index], data['raw'][pc_index, :, :3]),
# data['raw'][pc_index, :, 3:6], c0=c0,c1=c1, size=point_size, theta_1=theta_1, theta_2=theta_2, savename ='fig/3dmatch/_reg_gt_all' + str(pc_index) + '.svg')
# draw_figures.show_pointcloud_2part(se3.transform(pre_result['pre_transform'][pc_index], data['raw'][pc_index, :, :3]),
# data['raw'][pc_index, :, 3:6], c0=c0,c1=c1, size=point_size, theta_1=theta_1, theta_2=theta_2, savename ='fig/3dmatch/_reg_pre_all' + str(pc_index) + '.svg')
elif exp==3:
data = np.load(os.path.join(ROOT_DIR,'creat_data/PGM_DGCNN_ModelNet40Seen_NoPreW[\'xyz\', \'gxyz\']_attentiontransformer_crop/Creat_dataset_log_2020-09-29-14-46-57_metric.npy'),allow_pickle=True).item()
pre_result=np.load(os.path.join(ROOT_DIR,'output/PGM_DGCNN_ModelNet40Seen_NoPreW[\'xyz\', \'gxyz\']_attentiontransformer_crop/eval_log_2020-10-14-22-11-45_metric.npy'),allow_pickle=True).item()
print(np.where((pre_result['r_mae'] < 1) * (pre_result['t_mae'] < 0.1)))
# print(pre_result['r_mae'][(pre_result['r_mae'] < 1) * (pre_result['t_mae'] < 0.1)])
index = 0
theta_2 = 18
theta_1 = -140
# index=21
# theta_2 = 89
# theta_1 = -172
file_num=index//4+1
array_num=index-(file_num-1)*4
graph_array=np.load(os.path.join(ROOT_DIR,'output/PGM_DGCNN_ModelNet40Seen_NoPreW[\'xyz\', \'gxyz\']_attentiontransformer_crop/graph/graph'+str(file_num)+'.npy'),allow_pickle=True).item()
graph={k: graph_array[k][array_num] for k in graph_array}
node0 = np.where(graph['A_srcz'] > 0.08)
node1 = np.where(graph['A_tgtz'] > 0.08)
show_2graph(se3.transform(pre_result['gt_transform'][index], data['p1'][index, :, :3]), data['p2'][index], node0, node1, theta_1=theta_1, theta_2=theta_2,savename='fig/3dmatch/vis_graph' + str(index) + '.pdf')