-
Notifications
You must be signed in to change notification settings - Fork 13
/
metrics.py
146 lines (109 loc) · 3.83 KB
/
metrics.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
import numpy as np
from shapely.geometry import Polygon
from pcdet.ops.iou3d_nms.iou3d_nms_utils import boxes_bev_iou_cpu
class AverageMeter(object):
"""Computes and stores the average and current value"""
def __init__(self):
self.reset()
def reset(self):
self.val = 0
self.avg = 0
self.sum = 0
self.count = 0
def update(self, val, n=1):
self.val = val
self.sum += val * n
self.count += n
self.avg = self.sum / self.count
def estimateAccuracy(box_a, box_b, dim=3):
if dim == 3:
return np.linalg.norm(box_a.center - box_b.center, ord=2)
elif dim == 2:
return np.linalg.norm(
box_a.center[[0, 2]] - box_b.center[[0, 2]], ord=2)
def fromBoxToPoly(box):
return Polygon(tuple(box.corners()[[0, 2]].T[[0, 1, 5, 4]]))
# def corners2box(box):
def estimateIOU3d(box_a, box_b):
xyz_a = box_a.center
xyz_b = box_b.center
wlh_a = box_a.wlh
wlh_b = box_b.wlh
box3d_a = np.array([xyz_a[0], xyz_a[1], xyz_a[2], wlh_a[1], wlh_a[0], wlh_a[2], box_a.orientation.radians])
box3d_b = np.array([xyz_b[0], xyz_b[1], xyz_b[2], wlh_b[1], wlh_b[0], wlh_b[2], box_b.orientation.radians])
iou_2d = boxes_bev_iou_cpu(box3d_a[None, ...], box3d_b[None, ...])[0][0]
diff_h = np.abs(xyz_a[2] - xyz_b[2])
# assert wlh_a == wlh_b
assert np.sum(abs(wlh_a - wlh_b)) <= 1e-10
iou_h = 1.0 * (wlh_a[2] - diff_h) / (wlh_a[2] + diff_h)
return iou_2d * iou_h
def estimateOverlap(box_a, box_b, dim=2):
# if box_a == box_b:
# return 1.0
Poly_anno = fromBoxToPoly(box_a)
Poly_subm = fromBoxToPoly(box_b)
box_inter = Poly_anno.intersection(Poly_subm)
box_union = Poly_anno.union(Poly_subm)
if dim == 2:
return box_inter.area / box_union.area
else:
ymax = min(box_a.center[1], box_b.center[1])
ymin = max(box_a.center[1] - box_a.wlh[2],
box_b.center[1] - box_b.wlh[2])
inter_vol = box_inter.area * max(0, ymax - ymin)
anno_vol = box_a.wlh[0] * box_a.wlh[1] * box_a.wlh[2]
subm_vol = box_b.wlh[0] * box_b.wlh[1] * box_b.wlh[2]
overlap = inter_vol * 1.0 / (anno_vol + subm_vol - inter_vol)
return overlap
class Success(object):
"""Computes and stores the Success"""
def __init__(self, n=21, max_overlap=1):
self.max_overlap = max_overlap
self.Xaxis = np.linspace(0, self.max_overlap, n)
self.reset()
def reset(self):
self.overlaps = []
def add_overlap(self, val):
self.overlaps.append(val)
@property
def count(self):
return len(self.overlaps)
@property
def value(self):
succ = [
np.sum(i >= thres
for i in self.overlaps).astype(float) / self.count
for thres in self.Xaxis
]
return np.array(succ)
@property
def average(self):
if len(self.overlaps) == 0:
return 0
return np.trapz(self.value, x=self.Xaxis) * 100 / self.max_overlap
class Precision(object):
"""Computes and stores the Precision"""
def __init__(self, n=21, max_accuracy=2):
self.max_accuracy = max_accuracy
self.Xaxis = np.linspace(0, self.max_accuracy, n)
self.reset()
def reset(self):
self.accuracies = []
def add_accuracy(self, val):
self.accuracies.append(val)
@property
def count(self):
return len(self.accuracies)
@property
def value(self):
prec = [
np.sum(i <= thres
for i in self.accuracies).astype(float) / self.count
for thres in self.Xaxis
]
return np.array(prec)
@property
def average(self):
if len(self.accuracies) == 0:
return 0
return np.trapz(self.value, x=self.Xaxis) * 100 / self.max_accuracy