用mAP评估yolov3测试结果时用到的两段小代码

mAP的代码链接

对yolov3的测试代码进行了一点修改，将要检测的图片批量输入，再将检测结果分别保存到对应的txt文件中。但如果如果没有检测到框，就不会生成结果txt。

import os

path_new = './input/detection-results/'
path ='./input/ground-truth/'
list = os.listdir (path)
list2 = os.listdir (path_new)
for j in range(len(list)):
    name_d = [str (i) for i in list[j]]
    name = ''.join(name_d)
    # name2 = name.replace('png','txt')
    if name in list2:
        continue
    else:
        # os.remove(path+name)
        print(path+name)

目的就是检测出哪个图片没有被检测出目标，可以考虑把它删除掉或者移动到其他文件夹，便于后续分析。

import os

path = './input/ground-truth/'
path_new ='./input/ground-truth/'
list = os.listdir (path)
for j in range(len(list)):
    name_d = [str (i) for i in list[j]]
    name = ''.join (name_d)
    new_name = path_new+name
    f = open (path + name)
    fns = [line.strip () for line in f]
    for i in range(len(fns)):
        line = fns[i]
        lines = line.split (' ')
        x,y,w,h,cls = float(lines[0]),float(lines[1]),float(lines[2]),float(lines[3]),lines[4]
        x1=x-w/2
        y1=y-h/2
        x2=x+w/2
        y2=y+h/2
        with open (new_name, 'a+') as g:
            i = cls +' '+ str(x1) +' '+str(y1) + ' ' + str(x2) +' '+str(y2)+'\n'
            g.write(str(i))
            print(str(i))
    g.close()

因为我的数据集格式和这个算法默认的格式不太匹配，所以把它规范一下~