我是靠谱客的博主 魁梧台灯,最近开发中收集的这篇文章主要介绍Keras YOLOv3代码详解(三):目标检测的流程图和源代码+中文注释,觉得挺不错的,现在分享给大家,希望可以做个参考。

概述

Keras YOLOv3源代码下载地址:https://github.com/qqwweee/keras-yolo3

YOLOv3论文地址:https://pjreddie.com/media/files/papers/YOLOv3.pdf

关于darknet53网络的图像识别,请查看源文件:

keras-yolo3-masteryolo.py

keras-yolo3-masteryolo3model.py

keras-yolo3-masteryolo3utils.py

上一周,我们介绍了YOLOv3的目标识别原理,请参考:

Keras YOLOv3代码详解(二):目标检测原理解析

 

这篇文章的主要内容包括:(一)目标检测的核心函数,(二)目标检测流程图,(三)源代码+中文注释。

(一)目标检测的核心函数

YOLOv3的目标检测源代码,核心的函数包括:

detect_image()

generate()

yolo_eval()

yolo_model()

yolo_boxes_and_scores()

yolo_head()

yolo_correct_boxes()

等。

其中,yolo_model()已经在前面的文章中详细介绍过,请参考:

Keras YOLO v3代码详解(一):darknet53网络结构分析+Netron工具

 

(二)目标检测流程

YOLOv3的目标识别的源代码流程大致如下:

(1)设置缺省值并初始化:

(2)detect_image()将图片缩放成416x416大小,调用yolo_model(),生成13x13、26x26与52x52等3个feature map的输出,对这3个feature map进行预测,调用yolo_eval()函数得到目标框、目标框得分和类别,然后使用Pillow对发现的每一类对象的每一个目标框,绘制标签、框和文字:

(3)在yolo_eval()函数中调用yolo_boxes_and_scores()得到目标框、目标框得分和类别。而在yolo_boxes_and_scores()函数中,先调用yolo_head()函数计算每一个网格的目标对象的中心点坐标box_xy和目标框的宽与高box_wh,以及目标框的置信度box_confidence和类别置信度box_class_probs;然后调用yolo_correct_boxes(),将box_xy, box_wh转换为输入图片上的真实坐标,输出boxes是框的左下、右上两个坐标(y_min, x_min, y_max, x_max):

(4)完整的代码流程图如下图所示:

 

(三)源代码+中文注释

(3.1)yolo.py

class YOLO(object):

    _defaults = {

        "model_path": 'model_data/yolo.h5', #已经训练好的模型

        "anchors_path": 'model_data/yolo_anchors.txt', #通过聚类算法得到的3组9个anchor box,分别用于13x13、26x26、52x52的feature map

        "classes_path": 'model_data/coco_classes.txt', #可识别的COCO类别列表,一共80个

        "score" : 0.3, #框置信度阈值阈值,小于阈值则目标框被抛弃

        "iou" : 0.45, #IOU(Intersection over Union)阈值,大于阈值的重叠框被删除

        "model_image_size" : (416, 416), #输入图片的“标准尺寸”,不同于这个尺寸的输入会首先调整到标准大小

        "gpu_num" : 1, #GPU数量,通常是指Nvidia的GPU

    }


    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() #由generate()函数完成目标检测


    def generate(self):

        #读取:model路径、anchor box、coco类别,加载模型yolo.h5

        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.

        #对于80种coco目标,确定每一种目标框的绘制颜色,即:将(x/80, 1.0, 1.0)的颜色转换为RGB格式,并随机调整颜色以便于肉眼识别,其中:一个1.0表示饱和度,一个1.0表示亮度

        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, ))

        #若GPU个数大于等于2,调用multi_gpu_model()

        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):

        #开始计时
        start = timer()

        #调用letterbox_image()函数,即:先生成一个用“绝对灰”R128-G128-B128“填充的416x416新图片,然后用按比例缩放(采样方法:BICUBIC)后的输入图片粘贴,粘贴不到的部分保留为灰色

        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:

            #model_image_size定义的宽和高必须是32的整倍数;若没有定义model_image_size,将输入图片的尺寸调整到32的整倍数,并调用letterbox_image()函数进行缩放

            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. #将缩放后图片的数值除以255,做归一化

        #将(416,416,3)数组调整为(1,416,416,3)元组,满足YOLOv3输入的张量格式
        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={  #输入参数

                self.yolo_model.input: image_data, #输入图片

                self.input_image_shape: [image.size[1], image.size[0]], #图片尺寸416x416

                K.learning_phase(): 0 #学习模式:0测试/1训练

            })


        print('Found {} boxes for {}'.format(len(out_boxes), 'img'))

        #设置字体
        font = ImageFont.truetype(font='font/FiraMono-Medium.otf',

                    size=np.floor(3e-2 * image.size[1] + 0.5).astype('int32'))
        
        #设置目标框线条的宽度
        thickness = (image.size[0] + image.size[1]) // 300


        #对于c个目标类别中的每个目标框i,调用Pillow画图
        for i, c in reversed(list(enumerate(out_classes))):

            predicted_class = self.class_names[c]    #目标类别的名字

            box = out_boxes[i]        #目标框

            score = out_scores[i]     #目标框的置信度评分


            label = '{} {:.2f}'.format(predicted_class, score)

            draw = ImageDraw.Draw(image)  #输出:绘制输入的原始图片

            label_size = draw.textsize(label, font)    #返回label的宽和高(多少个pixels).


            top, left, bottom, right = box
            
            #目标框的上、左两个坐标小数点后一位四舍五入
            top = max(0, np.floor(top + 0.5).astype('int32'))

            left = max(0, np.floor(left + 0.5).astype('int32'))

            #目标框的下、右两个坐标小数点后一位四舍五入,与图片的尺寸相比,取最小值
            bottom = min(image.size[1], np.floor(bottom + 0.5).astype('int32'))

            right = min(image.size[0], np.floor(right + 0.5).astype('int32'))

            print(label, (left, top), (right, bottom))


            #确定标签(label)起始点位置:左、下
            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.
            
            #画目标框,线条宽度为thickness
            for i in range(thickness):

                draw.rectangle(

                    [left + i, top + i, right - i, bottom - i],

                    outline=self.colors[c])

            #画标签框
            draw.rectangle(

                [tuple(text_origin), tuple(text_origin + label_size)],

                fill=self.colors[c])

            #填写标签内容
            draw.text(text_origin, label, fill=(0, 0, 0), font=font)

            del draw

        #结束计时
        end = timer()

        print(end - start)

        return image

(3.2)yolo3model.py

def yolo_head(feats, anchors, num_classes, input_shape, calc_loss=False):

    #feats,即:feature maps

    """Convert final layer features to bounding box parameters."""

    num_anchors = len(anchors) #num_anchors = 3

    # Reshape to batch, height, width, num_anchors, box_params.

    anchors_tensor = K.reshape(K.constant(anchors), [1, 1, 1, num_anchors, 2])


    grid_shape = K.shape(feats)[1:3] # height, width #13x13或26x26或52x52

    #通过arange、reshape、tile的组合,根据grid_shape(13x13、26x26或52x52)创建y轴的0~N-1的组合grid_y,再创建x轴的0~N-1的组合grid_x,将两者拼接concatenate,形成NxN的grid(13x13、26x26或52x52)

    grid_y = K.tile(K.reshape(K.arange(0, stop=grid_shape[0]), [-1, 1, 1, 1]),

        [1, grid_shape[1], 1, 1])

    grid_x = K.tile(K.reshape(K.arange(0, stop=grid_shape[1]), [1, -1, 1, 1]),

        [grid_shape[0], 1, 1, 1])

    grid = K.concatenate([grid_x, grid_y])

    grid = K.cast(grid, K.dtype(feats))


    #从待处理的feature map的最后一维数据中,先将num_anchors这个维度与num_classes+5这个维度的数据分离,再取出4个框值tx、ty(最后一维数据的0:1)、tw和th(最后一维数据的2:3)

    feats = K.reshape(

        feats, [-1, grid_shape[0], grid_shape[1], num_anchors, num_classes + 5])


    # Adjust preditions to each spatial grid point and anchor size.

    #用sigmoid()函数计算目标框的中心点box_xy,用exp()函数计算目标框的宽和高box_wh
    #使用特征图尺寸(如:13x13、26x26或52x52)在水平x、垂直y两个维度对box_xy进行归一化,确定目标框的中心点的相对位置
    #使用标准图片尺寸(416x416)在宽和高两个维度对box_wh(因为,3组9个anchor box是基于416x416尺寸定义的)进行归一化,确定目标框的高和宽的相对位置

    box_xy = (K.sigmoid(feats[..., :2]) + grid) / K.cast(grid_shape[::-1], K.dtype(feats))

    box_wh = K.exp(feats[..., 2:4]) * anchors_tensor / K.cast(input_shape[::-1], K.dtype(feats))

    #用sigmoid()函数计算目标框的置信度box_confidence

    box_confidence = K.sigmoid(feats[..., 4:5])

    #用sigmoid()函数计算目标框的类别置信度box_class_probs

    box_class_probs = K.sigmoid(feats[..., 5:])


    if calc_loss == True:

        return grid, feats, box_xy, box_wh

    return box_xy, box_wh, box_confidence, box_class_probs


def yolo_correct_boxes(box_xy, box_wh, input_shape, image_shape):

    '''Get corrected boxes'''

    #将box_xy, box_wh转换为输入图片上的真实坐标,输出boxes是框的左下、右上两个坐标(y_min, x_min, y_max, x_max)

    #np.array[i:j:s],当s<0时,i缺省时,默认为-1;j缺省时,默认为-len(a)-1;所以array[::-1]相当于array[-1:-len(a)-1:-1],也就是从最后一个元素到第一个元素复制一遍,即倒序
    box_yx = box_xy[..., ::-1]

    box_hw = box_wh[..., ::-1]

    input_shape = K.cast(input_shape, K.dtype(box_yx))

    image_shape = K.cast(image_shape, K.dtype(box_yx))

    new_shape = K.round(image_shape * K.min(input_shape/image_shape))

    offset = (input_shape-new_shape)/2./input_shape

    scale = input_shape/new_shape

    box_yx = (box_yx - offset) * scale

    box_hw *= scale


    box_mins = box_yx - (box_hw / 2.)

    box_maxes = box_yx + (box_hw / 2.)

    boxes =  K.concatenate([

        box_mins[..., 0:1],  # y_min

        box_mins[..., 1:2],  # x_min

        box_maxes[..., 0:1],  # y_max

        box_maxes[..., 1:2]  # x_max

    ])


    # Scale boxes back to original image shape.

    boxes *= K.concatenate([image_shape, image_shape])

    return boxes



def yolo_boxes_and_scores(feats, anchors, num_classes, input_shape, image_shape):

'''Process Conv layer output'''

    #feats:需要处理的featue map

    #shape:(?,13,13,255),(?,26,26,255)或(?,52,52,255)

    #anchors:每层对应的3个anchor box      

    #num_classes:类别数(80)

    #input_shape:(416,416)

    #image_shape:图像尺寸

    box_xy, box_wh, box_confidence, box_class_probs = yolo_head(feats,

        anchors, num_classes, input_shape)

    boxes = yolo_correct_boxes(box_xy, box_wh, input_shape, image_shape)

    boxes = K.reshape(boxes, [-1, 4]) #形成框的列表boxes(?, 4)

    box_scores = box_confidence * box_class_probs #框的得分=框的置信度x类别置信度

    box_scores = K.reshape(box_scores, [-1, num_classes]) #形成框的得分列表box_scores(?, 80)

    return boxes, box_scores


def yolo_eval(yolo_outputs,

              anchors,

              num_classes,

              image_shape,

              max_boxes=20, #每张图每类最多检测到20个框

              score_threshold=.6,

              iou_threshold=.5):

"""Evaluate YOLO model on given input and return filtered boxes."""

    #将anchor_box分为3组,分别分配给13x13、26x26、52x52等3个yolo_model输出的feature map

    num_layers = len(yolo_outputs)

    anchor_mask = [[6,7,8], [3,4,5], [0,1,2]] if num_layers==3 else [[3,4,5], [1,2,3]] # default setting

    input_shape = K.shape(yolo_outputs[0])[1:3] * 32

    boxes = []

    box_scores = []

    #分别对3个feature map运行

    for l in range(num_layers):

        _boxes, _box_scores = yolo_boxes_and_scores(yolo_outputs[l],

            anchors[anchor_mask[l]], num_classes, input_shape, image_shape)

        boxes.append(_boxes)

        box_scores.append(_box_scores)

       
    #将运算得到的目标框用concatenate()函数拼接为(?, 4)的元组,将目标框的置信度拼接为(?,1)的元组

    boxes = K.concatenate(boxes, axis=0)

    box_scores = K.concatenate(box_scores, axis=0)


    #计算MASK掩码,过滤小于score阈值的值,只保留大于阈值的值

    mask = box_scores >= score_threshold

    max_boxes_tensor = K.constant(max_boxes, dtype='int32')

    boxes_ = []

    scores_ = []

    classes_ = []

    for c in range(num_classes):

        # TODO: use keras backend instead of tf.

        class_boxes = tf.boolean_mask(boxes, mask[:, c]) #通过掩码MASK和类别C筛选框boxes

        class_box_scores = tf.boolean_mask(box_scores[:, c], mask[:, c]) #通过掩码MASK和类别C筛选scores

        nms_index = tf.image.non_max_suppression(

            class_boxes, class_box_scores, max_boxes_tensor, iou_threshold=iou_threshold) #运行非极大抑制non_max_suppression(),每类最多检测20个框


        #K.gather:根据索引nms_index选择class_boxes和class_box_scores,标出选出的框的类别classes

        class_boxes = K.gather(class_boxes, nms_index)

        class_box_scores = K.gather(class_box_scores, nms_index)

        classes = K.ones_like(class_box_scores, 'int32') * c

        boxes_.append(class_boxes)

        scores_.append(class_box_scores)

        classes_.append(classes)

    
    #用concatenate()函数把选出的class_boxes、class_box_scores和classes拼接,形成(?,4)、(?,1)和(?,80)的元组返回

    boxes_ = K.concatenate(boxes_, axis=0)

    scores_ = K.concatenate(scores_, axis=0)

    classes_ = K.concatenate(classes_, axis=0)


    return boxes_, scores_, classes_

(3.3)yolo3utils.py
 

def letterbox_image(image, size):

'''resize image with unchanged aspect ratio using padding'''

#先生成一个用“绝对灰”R128-G128-B128“填充的416x416新图片,然后用按比例缩放(采样方法:BICUBIC)后的输入图片粘贴,粘贴不到的部分保留为灰色

    iw, ih = image.size

    w, h = size

    scale = min(w/iw, h/ih)

    nw = int(iw*scale)

    nh = int(ih*scale)


    image = image.resize((nw,nh), Image.BICUBIC)

    new_image = Image.new('RGB', size, (128,128,128))

    new_image.paste(image, ((w-nw)//2, (h-nh)//2))

    return new_image

参考:

https://blog.csdn.net/KKKSQJ/article/details/83587138

(完)

最后

以上就是魁梧台灯为你收集整理的Keras YOLOv3代码详解(三):目标检测的流程图和源代码+中文注释的全部内容,希望文章能够帮你解决Keras YOLOv3代码详解(三):目标检测的流程图和源代码+中文注释所遇到的程序开发问题。

如果觉得靠谱客网站的内容还不错,欢迎将靠谱客网站推荐给程序员好友。

本图文内容来源于网友提供,作为学习参考使用,或来自网络收集整理,版权属于原作者所有。
点赞(57)

评论列表共有 0 条评论

立即
投稿
返回
顶部