一篇讲通过改进NMS来提高检测效果的论文。
文章链接: 《Improving Object Detection With One Line of Code》
Github链接: https://github.com/bharatsingh430/soft-nms
Motivation
绝大部分目标检测方法,最后都要用到 NMS-非极大值抑制进行后处理。 通常的做法是将检测框按得分排序,然后保留得分最高的框,同时删除与该框重叠面积大于一定比例的其它框。
这种贪心式方法存在如下图所示的问题: 红色框和绿色框是当前的检测结果,二者的得分分别是0.95和0.80。如果按照传统的NMS进行处理,首先选中得分最高的红色框,然后绿色框就会因为与之重叠面积过大而被删掉。
另一方面,NMS的阈值也不太容易确定,设小了会出现下图的情况(绿色框因为和红色框重叠面积较大而被删掉),设置过高又容易增大误检。
思路:不要粗鲁地删除所有IOU大于阈值的框,而是降低其置信度。
Method
先直接上伪代码,如下图:如文章题目而言,就是用一行代码来替换掉原来的NMS。按照下图整个处理一遍之后,指定一个置信度阈值,然后最后得分大于该阈值的检测框得以保留
原来的NMS可以描述如下:将IOU大于阈值的窗口的得分全部置为0。
文章的改进有两种形式,一种是线性加权的:
一种是高斯加权的:
分析上面的两种改进形式,思想都是:M为当前得分最高框, bi 为待处理框, bi 和M的IOU越大, bi 的得分 si 就下降的越厉害。
具体地,下面是作者给出的代码:(当然不止一行T_T)
def cpu_soft_nms(np.ndarray[float, ndim=2] boxes, float sigma=0.5, float Nt=0.3, float threshold=0.001, unsigned int method=0):
cdef unsigned int N = boxes.shape[0]
cdef float iw, ih, box_area
cdef float ua
cdef int pos = 0
cdef float maxscore = 0
cdef int maxpos = 0
cdef float x1,x2,y1,y2,tx1,tx2,ty1,ty2,ts,area,weight,ov
for i in range(N):
maxscore = boxes[i, 4]
maxpos = i
tx1 = boxes[i,0]
ty1 = boxes[i,1]
tx2 = boxes[i,2]
ty2 = boxes[i,3]
ts = boxes[i,4]
pos = i + 1
# get max box
while pos < N:
if maxscore < boxes[pos, 4]:
maxscore = boxes[pos, 4]
maxpos = pos
pos = pos + 1
# add max box as a detection
boxes[i,0] = boxes[maxpos,0]
boxes[i,1] = boxes[maxpos,1]
boxes[i,2] = boxes[maxpos,2]
boxes[i,3] = boxes[maxpos,3]
boxes[i,4] = boxes[maxpos,4]
# swap ith box with position of max box
boxes[maxpos,0] = tx1
boxes[maxpos,1] = ty1
boxes[maxpos,2] = tx2
boxes[maxpos,3] = ty2
boxes[maxpos,4] = ts
tx1 = boxes[i,0]
ty1 = boxes[i,1]
tx2 = boxes[i,2]
ty2 = boxes[i,3]
ts = boxes[i,4]
pos = i + 1
# NMS iterations, note that N changes if detection boxes fall below threshold
while pos < N:
x1 = boxes[pos, 0]
y1 = boxes[pos, 1]
x2 = boxes[pos, 2]
y2 = boxes[pos, 3]
s = boxes[pos, 4]
area = (x2 - x1 + 1) * (y2 - y1 + 1)
iw = (min(tx2, x2) - max(tx1, x1) + 1)
if iw > 0:
ih = (min(ty2, y2) - max(ty1, y1) + 1)
if ih > 0:
ua = float((tx2 - tx1 + 1) * (ty2 - ty1 + 1) + area - iw * ih)
ov = iw * ih / ua #iou between max box and detection box
if method == 1: # linear
if ov > Nt:
weight = 1 - ov
else:
weight = 1
elif method == 2: # gaussian
weight = np.exp(-(ov * ov)/sigma)
else: # original NMS
if ov > Nt:
weight = 0
else:
weight = 1
boxes[pos, 4] = weight*boxes[pos, 4]
# if box score falls below threshold, discard the box by swapping with last box
# update N
if boxes[pos, 4] < threshold:
boxes[pos,0] = boxes[N-1, 0]
boxes[pos,1] = boxes[N-1, 1]
boxes[pos,2] = boxes[N-1, 2]
boxes[pos,3] = boxes[N-1, 3]
boxes[pos,4] = boxes[N-1, 4]
N = N - 1
pos = pos - 1
pos = pos + 1
keep = [i for i in range(N)]
return keep
- 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
Experiments
下图可以看出,基本可以获得平均1%的提升,且不增加额外的训练和计算负担。
高斯方差以及NMS的IOU阈值的敏感性测试: