Improving Object Detection With One Line of Code
2017-09-08 17:22
381 查看
一篇讲通过改进NMS来提高检测效果的论文。
论 文 地 址: 《Improving Object Detection With One Line of Code》
Github地址: https://github.com/bharatsingh430/soft-nms
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。
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)
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
论 文 地 址: 《Improving Object Detection With One Line of Code》
Github地址: https://github.com/bharatsingh430/soft-nms
动机:
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。
方法:
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] p d491 os = 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 keep1
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
相关文章推荐
- 目标检测--Improving Object Detection With One Line of Code
- 目标检测:Improving Object Detection With One Line of Code
- Improving Object Detection With One Line of Code
- How to train models of Object Detection with Discriminatively Trained Part Based Models
- 【CV论文阅读】+【搬运工】LocNet: Improving Localization Accuracy for Object Detection + A Theoretical analysis of feature pooling in Visual Recognition
- 论文提要“Improving Object Detection with DCN via Bayesian Optimization and Structured Prediction”
- How to train models of Object Detection with Discriminatively Trained Part Based Models
- KeyPoint of 《Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks》
- How to train models of Object Detection with Discriminatively Trained Part Based Models
- 在windows下训练models of Object Detection with Discriminatively Trained Part Based Models
- How to train models of Object Detection with Discriminatively Trained Part Based Models
- How to train models of Object Detection with Discriminatively Trained Part Based Models
- Rapid Object Detection With A Cascade of Boosted Classifiers Based on Haar-like Features
- Object Detection with 10 lines of code - Image AI
- Tutorial: OpenCV haartraining (Rapid Object Detection With A Cascade of Boosted Classifiers Based on
- How to train models of Object Detection with Discriminatively Trained Part Based Model
- 【原创】如何让SharePoint2010的内联代码支持.Net framework 3.5[How to embed inline code in aspx with .net framework 3.5 syntax]
- Sending SMS from Dynamics NAV with 5 lines of code
- mutating the [[Prototype]] of an object will cause your code to run very slowly
- vim separate with . and replace all to end of line