思想很简单，将标注的yolo数据转下格式，转为[类别，xmin，ymin，xmax，ymax]

转换valid后的信息，两个信息进行对比 完事

 

具体的，在终端执行：

./darknet detector valid cfg/voc.data cfg/yolo-voc.cfg backup/yolo-voc_final.weights

tip:在后面加入-thresh 0.25可以设置显示置信度的大小

在result文件夹下会生成以类别命名的txt文档，如下，在这里我一共就三个类别：smoke、red、white

comp4_det_test_smoke.txt

comp4_det_test_red.txt

comp4_det_test_white.txt

解析几个文件中的信息，如comp4_det_test_smoke.txt里面保存的是所有图片中识别为smoke的信息[文件名、置信度、xmin，ymin，xmax，ymax]。

运行如下代码：

import os

# txt_file为配置文件.data中的valid
txt_file = '/home/share/liubo/darknet/yanhuo/eval.txt'
f = open(txt_file)
lines = f.readlines()
for line in lines:
    line = line.split('/')[-1][0:-5]
    # test_out_file 为转换后保存的结果地址
    test_out_file = '/home/share/liubo/darknet-yolov3/results/test_acc_yanhuo'
    # 下面3个with需要自己的修改，修改成自己对应的类别
    with open(os.path.join(test_out_file , line + '.txt'), "a") as new_f:
        f1 = open('/home/share/liubo/darknet-yolov3/results/comp4_det_test_smoke.txt', 'r')
        f1_lines = f1.readlines()
        for f1_line in f1_lines:
            f1_line = f1_line.split()
            if line == f1_line[0]:
                new_f.write("%s %s %s %s %s %s\n" % ('smoke', f1_line[1], f1_line[2], f1_line[3], f1_line[4], f1_line[5]))
    with open(os.path.join(test_out_file , line + '.txt'), "a") as new_f:
        f1 = open('/home/share/liubo/darknet-yolov3/results/comp4_det_test_white.txt', 'r')
        f1_lines = f1.readlines()
        for f1_line in f1_lines:
            f1_line = f1_line.split()
            # print(line.split('.')[0] + ' ' + f1_line[0])
            if line == f1_line[0]:
                new_f.write("%s %s %s %s %s %s\n" % ('white', f1_line[1], f1_line[2], f1_line[3], f1_line[4], f1_line[5]))
    with open(os.path.join(test_out_file , line + '.txt'), "a") as new_f:
        f1 = open('/home/share/liubo/darknet-yolov3/results/comp4_det_test_red.txt', 'r')
        f1_lines = f1.readlines()
        for f1_line in f1_lines:
            f1_line = f1_line.split()
            if line == f1_line[0]:
                new_f.write("%s %s %s %s %s %s\n" % ('red', f1_line[1], f1_line[2], f1_line[3], f1_line[4], f1_line[5]))

运行代码后，效果如下：

                                转换前                                                                                  转换后

此时out_file中保存了所有的检测信息，接下来对yolo数据格式进行转换

执行以下代码：

import os
from PIL import Image
import numpy as np
# label_img为数据集的labels地址，img_path为数据集images的地址
label_img = '/home/share/liubo/darknet/yanhuo/labels/'
img_path = '/home/share/liubo/darknet/yanhuo/images/'
classes = {
    0:'smoke',
    1:'white',
    2:'red'
}
for line in lines:
    line = line.split('/')[-1][0:-5] + '.txt'
    txt = label_img + line
    img = np.array(Image.open(img_path + line.split('/')[-1][0:-4] + '.jpg'))
    sh, sw = img.shape[0], img.shape[1]
    # gt_out_file为转换后的地址
    gt_out_file = '/home/share/liubo/darknet-yolov3/results/gt_yanhuo'
    with open(os.path.join(gt_out_file , line ), "a") as new_f:
        f1 = open(txt)
        f1_lines = f1.readlines()
        for f1_line in f1_lines:
            f1_line = f1_line.split()
            x = float(f1_line[1]) * sw
            y = float(f1_line[2]) * sh
            w = float(f1_line[3]) * sw
            h = float(f1_line[4]) * sh
            xmin = x+1-w/2
            ymin = y+1-h/2
            xmax = x+1+w/2
            ymax = y+1+h/2
            new_f.write("%s %s %s %s %s\n" % (classes[int(f1_line[0])], xmin ,ymin,xmax,ymax))

此时已经保存了所有检测结果和转换后的结果，接下来对应的txt进行比较计算。

import os


def compute_IOU(rec1, rec2):
    """
    计算两个矩形框的交并比。
    :param rec1: (x0,y0,x1,y1)      (x0,y0)代表矩形左上的顶点，（x1,y1）代表矩形右下的顶点。下同。
    :param rec2: (x0,y0,x1,y1)
    :return: 交并比IOU.
    """
    left_column_max = max(rec1[0], rec2[0])
    right_column_min = min(rec1[2], rec2[2])
    up_row_max = max(rec1[1], rec2[1])
    down_row_min = min(rec1[3], rec2[3])
    # 两矩形无相交区域的情况
    if left_column_max >= right_column_min or down_row_min <= up_row_max:
        return 0
    # 两矩形有相交区域的情况
    else:
        S1 = (rec1[2] - rec1[0]) * (rec1[3] - rec1[1])
        S2 = (rec2[2] - rec2[0]) * (rec2[3] - rec2[1])
        S_cross = (down_row_min - up_row_max) * (right_column_min - left_column_max)
        return S_cross / (S1 + S2 - S_cross)

# gt为yolo数据转换后的地址
gt = '/home/share/liubo/darknet-yolov3/results/gt_yanhuo/'
# test为检测结果转换后的地址
test = '/home/share/liubo/darknet-yolov3/results/test_acc_yanhuo/'
# count_gt为标注的所有数据框
count_gt = {

}
# count_test为检测的所有数据框
count_test = {

}
# count_yes_test为检测正确的数据框
count_yes_test = {

}
# count_no_test为检测错误的数据框
count_no_test = {

}
# 计数
for gt_ in os.listdir(gt):
    txt = gt + gt_
    f = open(txt)
    lines = f.readlines()
    for line in lines:
        line = line.split()
        name = line[0]
        if name not in count_gt:
            count_gt[name] = 0
        count_gt[name] += 1
for test_ in os.listdir(test):
    txt = test + test_
    f = open(txt)
    lines = f.readlines()
    for line in lines:
        line = line.split()
        name = line[0]
        if name not in count_test:
            count_test[name] = 0
        count_test[name] += 1
# 下面主要思想：遍历test结果，再遍历对应gt的结果，如果两个框的iou大于一定的阙址并且类别相同，视为正确
for test_ in os.listdir(test):
    f_test_txt = test + test_
    f_test = open(f_test_txt)
    f_test_lines = f_test.readlines()
    for f_test_line in f_test_lines:
        f_test_line = f_test_line.split()
        f_gt_txt = gt + test_
        f_gt = open(f_gt_txt)
        f_gt_lines = f_gt.readlines()
        flag = 1
        for f_gt_line in f_gt_lines:
            f_gt_line = f_gt_line.split()
            IOU = compute_IOU([float(f_gt_line[1]), float(f_gt_line[2]), float(f_gt_line[3]), float(f_gt_line[4])],
                              [float(f_test_line[2]), float(f_test_line[3]), float(f_test_line[4]), float(f_test_line[5])])
            if f_gt_line[0] == f_test_line[0] and IOU >= 0.5 and float(f_test_line[1]) >= 0.3:
                flag = 0
                if f_test_line[0] not in count_yes_test:
                    count_yes_test[f_test_line[0]] = 0
                count_yes_test[f_test_line[0]] += 1

        if flag == 1:
            if f_test_line[0] not in count_no_test:
                count_no_test[f_test_line[0]] = 0
            count_no_test[f_test_line[0]] += 1
# 有以下4个结果，就可以计算相关指标了
print(count_gt)
print(count_test)
print(count_yes_test)
print(count_no_test)

 

以上就是Yolov3计算准确率、误报率、漏检率等的全过程

写了好久之后才写的博客，有什么理解错误或者代码问题请及时交流 感谢~