【目标检测】kera-yolo3模型计算mAP

Love The Way You Lie 2023-07-07 08:04 110阅读 0赞

这些是GitHub上的源码，下载之后均要修改一下。  
keras-yolo v3 源码：[https://github.com/qqwweee/keras-yolo3][https_github.com_qqwweee_keras-yolo3]  
mAP计算代码：[https://github.com/Cartucho/mAP][https_github.com_Cartucho_mAP]

--------------------

[【目标检测】交并比IoU、准确率precision、查全率recall、mAP][IoU_precision_recall_mAP]

上一篇博客[【目标检测】基于YOLOv3的海上船舶目标检测分类（Tensorflow/keras）][YOLOv3_Tensorflow_keras]记录了我训练模型的过程。

计算mAP，直接运行项目中`main.py`就好了。唯一难点就是要生成符合格式要求的`detection-results`和`ground-truth`。

--------------------

### 目录 ###

*  一、mAP项目构成
 *  二、 批量测试图片
 *  三、计算mAP
 *   *  3.1 生成detection-results
     *  3.2 生成ground-truth
     *  3.3 计算mAP

# 一、mAP项目构成 #

mAP项目如下：  
![在这里插入图片描述][20200222101315252.png]  
我们所需要了解的就是`input`文件，`input`目录下包含：`detection-results`、`ground-truth`、`images-optional`。

<table> 
 <thead> 
  <tr> 
   <th>文件夹</th> 
   <th>用途</th> 
  </tr> 
 </thead> 
 <tbody> 
  <tr> 
   <td>detection-results</td> 
   <td>模型预测的检测结果</td> 
  </tr> 
  <tr> 
   <td>ground-truth</td> 
   <td>图片本身的标记信息</td> 
  </tr> 
  <tr> 
   <td>images-optional</td> 
   <td>原始图片</td> 
  </tr> 
 </tbody> 
</table>

**detection-results 格式**  
每张图片保存在一个txt文件内，文件命名为图片的名称，每一行代表一个检测结果，格式为：`class score left top right bottom`。  
![在这里插入图片描述][20200222101739553.png]  
**ground-truth 格式**  
与detection-results 格式基本相同，只是缺少score这一项。

# 二、 批量测试图片 #

在`keras-yolo3项目`内建立新的python文件`test_yolo.py`（随便起的名，里面的代码是在yolo.py的基础上修改的）。该代码实现了对`test.txt`内图片的批量测试，并将结果保存在results目录下。  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70]  
result.txt文件内记录预测的结果。每一行代表一张图片的信息，依次代表图片名称、种类、得分、左、上、右、下。  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 1]  
代码同时复制一份原始图片到`mAP/input/images-optional`目录下。

代码如下：

# -*- coding: utf-8 -*-
    """ Class definition of YOLO_v3 style detection model on image and video """
    
    import colorsys
    import os
    import time
    from timeit import default_timer as timer
    
    import numpy as np
    from keras import backend as K
    from keras.models import load_model
    from keras.layers import Input
    from PIL import Image, ImageFont, ImageDraw
    
    from yolo3.model import yolo_eval, yolo_body, tiny_yolo_body
    from yolo3.utils import letterbox_image
    import os
    from keras.utils import multi_gpu_model
    
    
    # 创建创建一个存储检测结果的dir
    result_path = './result'
    if not os.path.exists(result_path):
        os.makedirs(result_path)
    
    # result如果之前存放的有文件，全部清除
    for i in os.listdir(result_path):
        path_file = os.path.join(result_path, i)
        if os.path.isfile(path_file):
            os.remove(path_file)
    
    # 创建一个记录检测结果的文件
    txt_path = result_path + '/result.txt'
    file = open(txt_path, 'w')
    
    
    class YOLO(object):
        _defaults = { 
            "model_path": 'logs/000/trained_weights.h5',
            "anchors_path": 'model_data/yolo_anchors.txt',
            "classes_path": 'model_data/voc_classes.txt',
            "score" : 0.3,
            "iou" : 0.45,
            "model_image_size" : (416, 416),
            "gpu_num" : 1,
        }
    
        @classmethod
        def get_defaults(cls, n):
            if n in cls._defaults:
                return cls._defaults[n]
            else:
                return "Unrecognized attribute name '" + n + "'"
    
        def __init__(self, **kwargs):
            self.__dict__.update(self._defaults) # set up default values
            self.__dict__.update(kwargs) # and update with user overrides
            self.class_names = self._get_class()
            self.anchors = self._get_anchors()
            self.sess = K.get_session()
            self.boxes, self.scores, self.classes = self.generate()
    
        def _get_class(self):
            classes_path = os.path.expanduser(self.classes_path)
            with open(classes_path) as f:
                class_names = f.readlines()
            class_names = [c.strip() for c in class_names]
            return class_names
    
        def _get_anchors(self):
            anchors_path = os.path.expanduser(self.anchors_path)
            with open(anchors_path) as f:
                anchors = f.readline()
            anchors = [float(x) for x in anchors.split(',')]
            return np.array(anchors).reshape(-1, 2)
    
        def generate(self):
            model_path = os.path.expanduser(self.model_path)
            assert model_path.endswith('.h5'), 'Keras model or weights must be a .h5 file.'
    
            # Load model, or construct model and load weights.
            num_anchors = len(self.anchors)
            num_classes = len(self.class_names)
            is_tiny_version = num_anchors==6 # default setting
            try:
                self.yolo_model = load_model(model_path, compile=False)
            except:
                self.yolo_model = tiny_yolo_body(Input(shape=(None,None,3)), num_anchors//2, num_classes) \
                    if is_tiny_version else yolo_body(Input(shape=(None,None,3)), num_anchors//3, num_classes)
                self.yolo_model.load_weights(self.model_path) # make sure model, anchors and classes match
            else:
                assert self.yolo_model.layers[-1].output_shape[-1] == \
                    num_anchors/len(self.yolo_model.output) * (num_classes + 5), \
                    'Mismatch between model and given anchor and class sizes'
    
            print('{} model, anchors, and classes loaded.'.format(model_path))
    
            # Generate colors for drawing bounding boxes.
            hsv_tuples = [(x / len(self.class_names), 1., 1.)
                          for x in range(len(self.class_names))]
            self.colors = list(map(lambda x: colorsys.hsv_to_rgb(*x), hsv_tuples))
            self.colors = list(
                map(lambda x: (int(x[0] * 255), int(x[1] * 255), int(x[2] * 255)),
                    self.colors))
            np.random.seed(10101)  # Fixed seed for consistent colors across runs.
            np.random.shuffle(self.colors)  # Shuffle colors to decorrelate adjacent classes.
            np.random.seed(None)  # Reset seed to default.
    
            # Generate output tensor targets for filtered bounding boxes.
            self.input_image_shape = K.placeholder(shape=(2, ))
            if self.gpu_num>=2:
                self.yolo_model = multi_gpu_model(self.yolo_model, gpus=self.gpu_num)
            boxes, scores, classes = yolo_eval(self.yolo_model.output, self.anchors,
                    len(self.class_names), self.input_image_shape,
                    score_threshold=self.score, iou_threshold=self.iou)
            return boxes, scores, classes
    
        def detect_image(self, image):
            start = timer()
    
            if self.model_image_size != (None, None):
                assert self.model_image_size[0]%32 == 0, 'Multiples of 32 required'
                assert self.model_image_size[1]%32 == 0, 'Multiples of 32 required'
                boxed_image = letterbox_image(image, tuple(reversed(self.model_image_size)))
            else:
                new_image_size = (image.width - (image.width % 32),
                                  image.height - (image.height % 32))
                boxed_image = letterbox_image(image, new_image_size)
            image_data = np.array(boxed_image, dtype='float32')
    
            print(image_data.shape)
            image_data /= 255.
            image_data = np.expand_dims(image_data, 0)  # Add batch dimension.
    
            out_boxes, out_scores, out_classes = self.sess.run(
                [self.boxes, self.scores, self.classes],
                feed_dict={ 
                    self.yolo_model.input: image_data,
                    self.input_image_shape: [image.size[1], image.size[0]],
                    K.learning_phase(): 0
                })
    
            print('Found {} boxes for {}'.format(len(out_boxes), 'img'))
    
            font = ImageFont.truetype(font='font/FiraMono-Medium.otf',
                        size=np.floor(3e-2 * image.size[1] + 0.5).astype('int32'))
            thickness = (image.size[0] + image.size[1]) // 300
    
            # # 保存框检测出的框的个数 (添加)
            # file.write('find ' + str(len(out_boxes)) + ' target(s) \n')
    
            for i, c in reversed(list(enumerate(out_classes))):
                predicted_class = self.class_names[c]
                box = out_boxes[i]
                score = out_scores[i]
    
                label = '{} {:.2f}'.format(predicted_class, score)
                draw = ImageDraw.Draw(image)
                label_size = draw.textsize(label, font)
    
                top, left, bottom, right = box
                top = max(0, np.floor(top + 0.5).astype('int32'))
                left = max(0, np.floor(left + 0.5).astype('int32'))
                bottom = min(image.size[1], np.floor(bottom + 0.5).astype('int32'))
                right = min(image.size[0], np.floor(right + 0.5).astype('int32'))
    
                # # 写入检测位置（添加）
                # file.write(
                # predicted_class + ' score: ' + str(score) + ' \nlocation: top: ' + str(top) + '、 bottom: ' + str(
                # bottom) + '、 left: ' + str(left) + '、 right: ' + str(right) + '\n')
    
                file.write(predicted_class + ' ' + str(score) + ' ' + str(left) + ' ' + str(top) + ' ' + str(right) + ' ' + str(bottom) + ';')
    
                print(label, (left, top), (right, bottom))
    
                if top - label_size[1] >= 0:
                    text_origin = np.array([left, top - label_size[1]])
                else:
                    text_origin = np.array([left, top + 1])
    
                # My kingdom for a good redistributable image drawing library.
                for i in range(thickness):
                    draw.rectangle(
                        [left + i, top + i, right - i, bottom - i],
                        outline=self.colors[c])
                draw.rectangle(
                    [tuple(text_origin), tuple(text_origin + label_size)],
                    fill=self.colors[c])
                draw.text(text_origin, label, fill=(0, 0, 0), font=font)
                del draw
            end = timer()
            print(end - start)
            return image
    
        def close_session(self):
            self.sess.close()
    
    
    
    def detect_video(yolo, video_path, output_path=""):
        import cv2
        vid = cv2.VideoCapture(video_path)
        if not vid.isOpened():
            raise IOError("Couldn't open webcam or video")
        video_FourCC    = int(vid.get(cv2.CAP_PROP_FOURCC))        # 获得视频编码MPEG4/H264
        video_fps       = vid.get(cv2.CAP_PROP_FPS)
        video_size      = (int(vid.get(cv2.CAP_PROP_FRAME_WIDTH)),
                            int(vid.get(cv2.CAP_PROP_FRAME_HEIGHT)))
        isOutput = True if output_path != "" else False
        if isOutput:
            print("!!! TYPE:", type(output_path), type(video_FourCC), type(video_fps), type(video_size))
            out = cv2.VideoWriter(output_path, video_FourCC, video_fps, video_size)
        accum_time = 0
        curr_fps = 0
        fps = "FPS: ??"
        prev_time = timer()
        while True:
            return_value, frame = vid.read()
            image = Image.fromarray(frame)           # 从array转换成image
            image = yolo.detect_image(image)
            result = np.asarray(image)
            curr_time = timer()
            exec_time = curr_time - prev_time
            prev_time = curr_time
            accum_time = accum_time + exec_time
            curr_fps = curr_fps + 1
            if accum_time > 1:
                accum_time = accum_time - 1
                fps = "FPS: " + str(curr_fps)
                curr_fps = 0
            cv2.putText(result, text=fps, org=(3, 15), fontFace=cv2.FONT_HERSHEY_SIMPLEX,
                        fontScale=0.50, color=(255, 0, 0), thickness=2)
            cv2.namedWindow("result", cv2.WINDOW_NORMAL)
            cv2.imshow("result", result)
            if isOutput:
                out.write(result)
            if cv2.waitKey(1) & 0xFF == ord('q'):
                break
        yolo.close_session()
    
    
    # 批量处理文件
    if __name__ == '__main__':
        # 读取test文件
        with open("A05_helmet/ImageSets/Main/test.txt", 'r') as f:  # 打开文件
            test_list = f.readlines()  # 读取文件
            test_list = [x.strip() for x in test_list if x.strip() != '']  # 去除/n
            # print(test_list)
    
        t1 = time.time()
        yolo = YOLO()
    
        for filename in test_list:
            image_path = 'A05_helmet/JPEGImages/'+filename+'.jpg'
            portion = os.path.split(image_path)
            # file.write(portion[1]+' detect_result：\n')
            file.write(image_path + ' ')
            image = Image.open(image_path)
            image_mAP_save_path = 'E:/Activities/fwwb2019/code/mAP-master/input/images-optional/'
            image.save(image_mAP_save_path + filename + '.jpg')
            r_image = yolo.detect_image(image)
            file.write('\n')
            #r_image.show() 显示检测结果
            image_save_path = './result/result_'+portion[1]
            print('detect result save to....:'+image_save_path)
            r_image.save(image_save_path)
    
        time_sum = time.time() - t1
        # file.write('time sum: '+str(time_sum)+'s')
        print('time sum:',time_sum)
        file.close()
        yolo.close_session()

# 三、计算mAP #

首先在mAP项目中建立名为cal\_mAP文件夹，在其中分别建立如下两个python文件。分别用于生成符合detection-result和ground-truth的格式要求的txt文件。  
![在这里插入图片描述][20200222102314381.png]

## 3.1 生成detection-results ##

从第二部分介绍的result，我们可以生成符合detection-results文件。下图为result.txt：  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 2]  
通过如下代码：

import re
    import os
    
    dir_project = os.path.abspath(os.path.join(os.getcwd(), "../.."))  # 获取上上级目录
    dir_result = '/result/result.txt'  # yolo批量处理结果的目录
    dir_detection_results = '/mAP/input/detection-results'  # detection-results目录
    surplus = 'A05_helmet/JPEGImages/'  # result.txt文件中图片名称多余的部分
    
    if __name__ == '__main__':
        with open(dir_project + dir_result, 'r') as f:  # 打开文件
            filename = f.readlines()  # 读取文件
    
        for i in range(len(filename)):
            filename[i] = re.sub(surplus, '', filename[i])        # 去除文件名多余的部分
    
        for i in range(len(filename)):  # 中按行存放的检测内容，为列表的形式
            r = filename[i].split('.jpg ')
            file = open(dir_project + dir_detection_results + '/' + r[0] + '.txt', 'w')
            t = r[1].split(';')
            # 去除空格和换行
            t.remove('\n')
    
            if len(t) == 0:            # 如果没有对象
                file.write('')
            else:
                for k in range(len(t)):
                    file.write(t[k] + '\n')

将其转化为如下txt文件，每一张图片用一个txt文本表示：  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 3]  
每一个txt文本内容如下图样式，以图片00242为例：  
![在这里插入图片描述][20200222105603719.png]

## 3.2 生成ground-truth ##

在[上一篇博客][Link 1]中我们生成了`test.txt`文件，我们则要将它转化为符合ground-truth格式要求的txt文本。  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 4]  
此文件中，没有score，其中“1”代表“hat”，“2”代表“person”。

我们通过如下代码进行转换：

import re
    import os
    
    dir_project = os.path.abspath(os.path.join(os.getcwd(), "../.."))  # 获取上上级目录
    dir_ground_truth = '/mAP/input/ground-truth'  # detection-results目录
    surplus = 'A05_helmet/JPEGImages/'  # result.txt文件中图片名称多余的部分
    
    if __name__ == '__main__':
        with open(dir_project + '/test.txt', 'r') as f:  # 打开文件
            filename = f.readlines()  # 读取文件
            # print(filename)
    
        for i in range(len(filename)):
            filename[i] = re.sub(surplus, '', filename[i])    # 去除文件名多余的部分
    
        for i in range(len(filename)):  # 中按行存放的检测内容，为列表的形式
            r = filename[i].split('.jpg ')
            print(r[0])
            file = open(dir_project + dir_ground_truth + '/' + r[0] + '.txt', 'w')
            t = r[1].split(' ')
    
            for j in range(len(t)):
                class_t = t[j].split(',')[-1]
                pos_t = t[j].split(',')
                if class_t == '0' or class_t == '0\n':
                    file.write('person ' + pos_t[0] + ' ' + pos_t[1] + ' '+ pos_t[2] + ' '+ pos_t[3] + '\n')
                elif class_t == '1' or class_t == '1\n':
                    file.write('hat ' + pos_t[0] + ' ' + pos_t[1] + ' '+ pos_t[2] + ' '+ pos_t[3] + '\n')

转换成功后生成如下文件：  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 5]  
以图片00242为例：  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 6]

## 3.3 计算mAP ##

如果你的**类中包含空格**（如`cargo ship`、`ore carrier`），则需要进行小小的修改，参考本篇博客：[计算mAP去除类之间空格（remove\_space）][mAP_remove_space]。

之后可直接运行mAP项目中的`main.py`，可生成result文件夹，在该文件夹内保存各类结果。

--------------------

如果你的类中不包含空格(如`person`、`hat`），可直接运行mAP项目中的`main.py`，可生成result文件夹，在该文件夹内保存各类结果。  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 7]  
单一图像，预测和实际的对比：  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 8]  
各类结果都比较齐全，可慢慢分析。

参考：[【YOLOV3-keras-MAP】YOLOV3-keras版本的mAP计算][YOLOV3-keras-MAP_YOLOV3-keras_mAP]

[https_github.com_qqwweee_keras-yolo3]: https://github.com/qqwweee/keras-yolo3
[https_github.com_Cartucho_mAP]: https://github.com/Cartucho/mAP
[IoU_precision_recall_mAP]: https://blog.csdn.net/See_Star/article/details/104392492
[YOLOv3_Tensorflow_keras]: https://blog.csdn.net/See_Star/article/details/105820181
[20200222101315252.png]: https://img-blog.csdnimg.cn/20200222101315252.png
[20200222101739553.png]: https://img-blog.csdnimg.cn/20200222101739553.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/20200221204556107.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 1]: https://img-blog.csdnimg.cn/20200222100748841.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[20200222102314381.png]: https://img-blog.csdnimg.cn/20200222102314381.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 2]: https://img-blog.csdnimg.cn/20200222105459382.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 3]: https://img-blog.csdnimg.cn/20200222105540376.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[20200222105603719.png]: https://img-blog.csdnimg.cn/20200222105603719.png
[Link 1]: https://blog.csdn.net/See_Star/article/details/103595421
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 4]: https://img-blog.csdnimg.cn/20200222110133467.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 5]: https://img-blog.csdnimg.cn/20200222110335136.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 6]: https://img-blog.csdnimg.cn/20200222110358589.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[mAP_remove_space]: https://blog.csdn.net/See_Star/article/details/104719513
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 7]: https://img-blog.csdnimg.cn/2020022211062741.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy_size_16_color_FFFFFF_t_70 8]: https://img-blog.csdnimg.cn/20200222110812334.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L1NlZV9TdGFy,size_16,color_FFFFFF,t_70
[YOLOV3-keras-MAP_YOLOV3-keras_mAP]: https://blog.csdn.net/plsong_csdn/article/details/89502117