目标检测–解读我在重写yolov3的时候遇到的困难
一、特征提取部分
对于darknet50是很好的特征提取，但是我个人认为在我们平常restnet50做主干网络就行了，没必要darknet50，因为darknet50设计出来本来就是分类1000类的，我个人也不崇尚于大网络，我和我师兄测试过长沙理工的交通数据集，darnet50并不是很理想。所以我个人的建议就是换成restnet50，mobilenet等小型网络跑分类10以下，这个我们测试过确实比较好，在交通数据集上我师兄告诉我，他loss在原来的8，下降到了5。
darknet50特征提取介绍
这里我的这张图我盗用我师兄的一张图，哎因为懒得画了，高手不要喷我
darknet50我认为是yolov2和yolov1以来最大的改进吧，其实我特别崇拜何凯明，就是因为他的残差网络，不管在cv届还是什么。我感觉都对识别效果带来了巨大的改进，darknet53引入了3x3的卷积和1x1的卷积，一共有53个卷积层，所以我们称它为Darknet-53。 跳连的思想，其实我还是觉得挺好的，缓解了在深度神经网络中增加深度带来的梯度消失问题。
我认为darknet53还有个最大的改进就是引入了Leaky ReLU激活函数，以前的Relu激活函数负值为0，而Leaky Relu他的负值不在是0， 如下图比较。但是在这里我还是认为适合特别目标的网络，才是最好的。
代码实例
from functools import wraps
from keras.layers import Conv2D, Add, ZeroPadding2D, UpSampling2D, Concatenate, MaxPooling2D,DepthwiseConv2D, Input, Activation, Dropout, Reshape, BatchNormalization, \
GlobalAveragePooling2D, GlobalMaxPooling2D, Conv2D
from keras.layers.advanced_activations import LeakyReLU
from keras.layers.normalization import BatchNormalization
from keras.regularizers import l2
from utils.utils import compose
from keras import backend as K
def relu6(x):
return K.relu(x, max_value=6)
#--------------------------------------------------#
####### 单次卷积
#--------------------------------------------------#
@wraps(Conv2D)
def DarknetConv2D(*args, **kwargs):
darknet_conv_kwargs = {'kernel_regularizer': l2(5e-4)}
darknet_conv_kwargs['padding'] = 'valid' if kwargs.get('strides')==(2,2) else 'same'
darknet_conv_kwargs.update(kwargs)
return Conv2D(*args, **darknet_conv_kwargs)
#---------------------------------------------------#
####### 卷积块
####### DarknetConv2D + BatchNormalization + LeakyReLU
#---------------------------------------------------#
def DarknetConv2D_BN_Leaky(*args, **kwargs):
no_bias_kwargs = {'use_bias': False}
no_bias_kwargs.update(kwargs)
return compose(
DarknetConv2D(*args, **no_bias_kwargs),
BatchNormalization(),
LeakyReLU(alpha=0.1))
#---------------------------------------------------#
####### 卷积块
####### DarknetConv2D + BatchNormalization + LeakyReLU
#---------------------------------------------------#
def resblock_body(x, num_filters, num_blocks):
x = ZeroPadding2D(((1,0),(1,0)))(x)
x = DarknetConv2D_BN_Leaky(num_filters, (3,3), strides=(2,2))(x)
for i in range(num_blocks):
y = DarknetConv2D_BN_Leaky(num_filters//2, (1,1))(x)
y = DarknetConv2D_BN_Leaky(num_filters, (3,3))(y)
x = Add()([x,y])
return x
#---------------------------------------------------#
####### darknet53 的主体部分
#---------------------------------------------------#
def darknet_body(x):
x = DarknetConv2D_BN_Leaky(32, (3,3))(x)
x = resblock_body(x, 64, 1)
x = resblock_body(x, 128, 2)
x = resblock_body(x, 256, 8)
feat1 = x
x = resblock_body(x, 512, 8)
feat2 = x
x = resblock_body(x, 1024, 4)
feat3 = x
return feat1,feat2,feat3
附上我改进后的Restnet50部分：
from functools import wraps
from keras.layers import Conv2D, Add, ZeroPadding2D, UpSampling2D, Concatenate, MaxPooling2D,DepthwiseConv2D, Input, Activation, Dropout, Reshape, BatchNormalization, \
GlobalAveragePooling2D, GlobalMaxPooling2D, Conv2D
from keras.layers.advanced_activations import LeakyReLU
from keras.layers.normalization import BatchNormalization
from keras.regularizers import l2
from util import compose
from keras import backend as K
from keras import layers
def identity_block(input_tensor, kernel_size, filters, stage, block):
filters1, filters2, filters3 = filters
conv_name_base = 'res' + str(stage) + block + '_branch'
bn_name_base = 'bn' + str(stage) + block + '_branch'
# 降维
x = Conv2D(filters1, (1, 1), name=conv_name_base + '2a')(input_tensor)
x = BatchNormalization(name=bn_name_base + '2a')(x)
x = Activation('relu')(x)
# 3x3卷积
x = Conv2D(filters2, kernel_size,padding='same', name=conv_name_base + '2b')(x)
x = BatchNormalization(name=bn_name_base + '2b')(x)
x = Activation('relu')(x)
# 升维
x = Conv2D(filters3, (1, 1), name=conv_name_base + '2c')(x)
x = BatchNormalization(name=bn_name_base + '2c')(x)
x = layers.add([x, input_tensor])
x = Activation('relu')(x)
return x
def conv_block(input_tensor, kernel_size, filters, stage, block, strides=(2, 2)):
# 64,64,256
filters1, filters2, filters3 = filters
conv_name_base = 'res' + str(stage) + block + '_branch'
bn_name_base = 'bn' + str(stage) + block + '_branch'
# 降维
x = Conv2D(filters1, (1, 1), strides=strides,
name=conv_name_base + '2a')(input_tensor)
x = BatchNormalization(name=bn_name_base + '2a')(x)
x = Activation('relu')(x)
# 3x3卷积
x = Conv2D(filters2, kernel_size, padding='same',
name=conv_name_base + '2b')(x)
x = BatchNormalization(name=bn_name_base + '2b')(x)
x = Activation('relu')(x)
# 升维
x = Conv2D(filters3, (1, 1), name=conv_name_base + '2c')(x)
x = BatchNormalization(name=bn_name_base + '2c')(x)
# 残差边
shortcut = Conv2D(filters3, (1, 1), strides=strides,
name=conv_name_base + '1')(input_tensor)
shortcut = BatchNormalization(name=bn_name_base + '1')(shortcut)
x = layers.add([x, shortcut])
x = Activation('relu')(x)
return x
#---------------------------------------------------#
####### retnet50 的主体部分
#---------------------------------------------------#
def darknet_body(x,):
x = ZeroPadding2D((3, 3))(x)
# [208,208,64]
x = Conv2D(64, (7, 7), strides=(2, 2), name='conv1')(x)
x = BatchNormalization(name='bn_conv1')(x)
x = Activation('relu')(x)
# [104,104,64]
x = MaxPooling2D((3, 3), strides=(2, 2))(x)
# [104,104,256]
x = conv_block(x, 3, [64, 64, 256], stage=2, block='a', strides=(1, 1))
x = identity_block(x, 3, [64, 64, 256], stage=2, block='b')
x = identity_block(x, 3, [64, 64, 256], stage=2, block='c')
# [52,52,256]
x = conv_block(x, 3, [128, 128, 256], stage=3, block='a')
x = identity_block(x, 3, [128, 128,256], stage=3, block='b')
x = identity_block(x, 3, [128, 128, 256], stage=3, block='c')
x = identity_block(x, 3, [128, 128, 256], stage=3, block='d')
feat1 = x
# x = resblock_body(x, 512, 8)
# 26,26,512
x = conv_block(x, 3, [256, 256, 512], stage=4, block='a')
x = identity_block(x, 3, [256, 256, 512], stage=4, block='b')
x = identity_block(x, 3, [256, 256, 512], stage=4, block='c')
x = identity_block(x, 3, [256, 256, 512], stage=4, block='d')
x = identity_block(x, 3, [256, 256, 512], stage=4, block='e')
x = identity_block(x, 3, [256, 256, 512], stage=4, block='f')
feat2 = x
# [13,13,1024]
x = conv_block(x, 3, [512, 512, 2048], stage=5, block='a')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='b')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='c')
# x = resblock_body(x, 1024, 4)
feat3 = x
return feat1,feat2,feat3
darknet50特征提取后分类
这里输入图像是416X416,在原始论文里作者提到过为什么用416作为输入，好像说为为了留个奇数点，预测大目标，不过这里我有点没太懂起作者的意思。在特征提取部分利用部分，yolov3用了多尺度融合，多个特征层进行检测三个特征层的shape分别为(52,52,256)、(26,26,512)、(13,13,1024)。这里我师兄这张图是voc数据集，voc数据集只有20类，最后输出结果shape分别为(13,13,75)，(26,26,75)，(52,52,75)，最后一个维度为75是因为该图是基于voc数据集的，它的类为20种，yolo3只有针对每一个特征层存在3个先验框，所以最后维度为3x25；5表示x_offset、y_offset和预测框长宽，加上自信度。
如图
代码如下：
from functools import wraps
import numpy as np
import tensorflow as tf
from keras import backend as K
from keras.layers import Conv2D, Add, ZeroPadding2D, UpSampling2D, Concatenate, MaxPooling2D
from keras.layers.advanced_activations import LeakyReLU
from keras.layers.normalization import BatchNormalization
from keras.models import Model
from keras.regularizers import l2
from nets.darknet53 import darknet_body
from utils.utils import compose
#--------------------------------------------------#
####### 单次卷积
#--------------------------------------------------#
@wraps(Conv2D)
def DarknetConv2D(*args, **kwargs):
darknet_conv_kwargs = {'kernel_regularizer': l2(5e-4)}
darknet_conv_kwargs['padding'] = 'valid' if kwargs.get('strides')==(2,2) else 'same'
darknet_conv_kwargs.update(kwargs)
return Conv2D(*args, **darknet_conv_kwargs)
#---------------------------------------------------#
####### 卷积块
####### DarknetConv2D + BatchNormalization + LeakyReLU
#---------------------------------------------------#
def DarknetConv2D_BN_Leaky(*args, **kwargs):
no_bias_kwargs = {'use_bias': False}
no_bias_kwargs.update(kwargs)
return compose(
DarknetConv2D(*args, **no_bias_kwargs),
BatchNormalization(),
LeakyReLU(alpha=0.1))
#---------------------------------------------------#
####### 特征层->最后的输出
#---------------------------------------------------#
def make_last_layers(x, num_filters, out_filters):
# 五次卷积
x = DarknetConv2D_BN_Leaky(num_filters, (1,1))(x)
x = DarknetConv2D_BN_Leaky(num_filters*2, (3,3))(x)
x = DarknetConv2D_BN_Leaky(num_filters, (1,1))(x)
x = DarknetConv2D_BN_Leaky(num_filters*2, (3,3))(x)
x = DarknetConv2D_BN_Leaky(num_filters, (1,1))(x)
# 将最后的通道数调整为outfilter
y = DarknetConv2D_BN_Leaky(num_filters*2, (3,3))(x)
y = DarknetConv2D(out_filters, (1,1))(y)
return x, y
#---------------------------------------------------#
####### 特征层->最后的输出
#---------------------------------------------------#
def yolo_body(inputs, num_anchors, num_classes):
# 生成darknet53的主干模型
feat1,feat2,feat3 = darknet_body(inputs)
darknet = Model(inputs, feat3)
# 第一个特征层
# y1=(batch_size,13,13,3,85)
x, y1 = make_last_layers(darknet.output, 512, num_anchors*(num_classes+5))
x = compose(
DarknetConv2D_BN_Leaky(256, (1,1)),
UpSampling2D(2))(x)
x = Concatenate()([x,feat2])
# 第二个特征层
# y2=(batch_size,26,26,3,85)
x, y2 = make_last_layers(x, 256, num_anchors*(num_classes+5))
x = compose(
DarknetConv2D_BN_Leaky(128, (1,1)),
UpSampling2D(2))(x)
x = Concatenate()([x,feat1])
# 第三个特征层
# y3=(batch_size,52,52,3,85)
x, y3 = make_last_layers(x, 128, num_anchors*(num_classes+5))
return Model(inputs, [y1,y2,y3])`
二、预测框解码过程
预测结果对应着三个预测框的位置，我们先将其reshape一下，其结果为**(N,13,13,3,25)，(N,26,26,3,25)，(N,52,52,3,25)。**
最后一个维度中的25包含了4+1+20，分别代表x_offset、y_offset、h和w、置信度、分类结果。
yolo3的解码过程就是将每个网格点加上它对应的x_offset和y_offset，加完后的结果就是预测框的中心，然后再利用 先验框和h、w结合 计算出预测框的长和宽，当计算出位置后，还要经过非极大抑制筛选，这样就能得到整个预测框的位置了。在此附上作者原始论文中的图：
代码如下：
```python
#---------------------------------------------------#
# 将预测值的每个特征层调成真实值
#---------------------------------------------------#
def yolo_head(feats, anchors, num_classes, input_shape, calc_loss=False):
num_anchors = len(anchors)
# [1, 1, 1, num_anchors, 2]
anchors_tensor = K.reshape(K.constant(anchors), [1, 1, 1, num_anchors, 2])
# 获得x，y的网格
# (13, 13, 1, 2)
grid_shape = K.shape(feats)[1:3] # height, width
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))
# (batch_size,13,13,3,25)
feats = K.reshape(feats, [-1, grid_shape[0], grid_shape[1], num_anchors, num_classes + 5])
# 将预测值调成真实值
# box_xy对应框的中心点
# box_wh对应框的宽和高
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))
box_confidence = K.sigmoid(feats[..., 4:5])
box_class_probs = K.sigmoid(feats[..., 5:])
# 在计算loss的时候返回如下参数
if calc_loss == True:
return grid, feats, box_xy, box_wh
return box_xy, box_wh, box_confidence, box_class_probs
#---------------------------------------------------#
# 对box进行调整，使其符合真实图片的样子
#---------------------------------------------------#
def yolo_correct_boxes(box_xy, box_wh, input_shape, image_shape):
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
])
boxes *= K.concatenate([image_shape, image_shape])
return boxes
#---------------------------------------------------#
# 获取每个box和它的得分
#---------------------------------------------------#
def yolo_boxes_and_scores(feats, anchors, num_classes, input_shape, image_shape):
# 将预测值调成真实值
# box_xy对应框的中心点
# box_wh对应框的宽和高
# -1,13,13,3,2; -1,13,13,3,2; -1,13,13,3,1; -1,13,13,3,20
box_xy, box_wh, box_confidence, box_class_probs = yolo_head(feats, anchors, num_classes, input_shape)
# 将box_xy、和box_wh调节成y_min,y_max,xmin,xmax
boxes = yolo_correct_boxes(box_xy, box_wh, input_shape, image_shape)
# 获得得分和box
boxes = K.reshape(boxes, [-1, 4])
box_scores = box_confidence * box_class_probs
box_scores = K.reshape(box_scores, [-1, num_classes])
return boxes, box_scores
#---------------------------------------------------#
# 图片预测
#---------------------------------------------------#
def yolo_eval(yolo_outputs,
anchors,
num_classes,
image_shape,
max_boxes=20,
score_threshold=.6,
iou_threshold=.5):
# 获得特征层的数量
num_layers = len(yolo_outputs)
# 特征层1对应的anchor是678
# 特征层2对应的anchor是345
# 特征层3对应的anchor是012
anchor_mask = [[6,7,8], [3,4,5], [0,1,2]]
input_shape = K.shape(yolo_outputs[0])[1:3] * 32
boxes = []
box_scores = []
# 对每个特征层进行处理
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)
# 将每个特征层的结果进行堆叠
boxes = K.concatenate(boxes, axis=0)
box_scores = K.concatenate(box_scores, axis=0)
mask = box_scores >= score_threshold
max_boxes_tensor = K.constant(max_boxes, dtype='int32')
boxes_ = []
scores_ = []
classes_ = []
for c in range(num_classes):
# 取出所有box_scores >= score_threshold的框，和成绩
class_boxes = tf.boolean_mask(boxes, mask[:, c])
class_box_scores = tf.boolean_mask(box_scores[:, c], mask[:, c])
# 非极大抑制，去掉box重合程度高的那一些
nms_index = tf.image.non_max_suppression(
class_boxes, class_box_scores, max_boxes_tensor, iou_threshold=iou_threshold)
# 获取非极大抑制后的结果
# 下列三个分别是
# 框的位置，得分与种类
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)
boxes_ = K.concatenate(boxes_, axis=0)
scores_ = K.concatenate(scores_, axis=0)
classes_ = K.concatenate(classes_, axis=0)
return boxes_, scores_, classes_
三、训练
首先在数据预处理部分对数据进行了数据增强，不过我感觉yolo的数据增强并不是很好，我曾经用yolo的数据增强去参加比赛发现，yolo的数据增强提升可能只有0.001，不过最新的autoaugment 感觉还不错，我也最近在研究怎么改上去。回归正题，因为voc数据集是左上角和右下角坐标，我们需要把数据处理成中心点、w和h。首先我们建立了建立全为0的y_true，y_true是一个列表，包含三个特征层，shape分别为(m,13,13,3,25),(m,26,26,3,25),(m,52,52,3,25)，然后对每一张图片处理，将每一张图片中的真实框的wh和先验框的wh对比，计算IOU值，选取其中IOU最高的一个，得到其所属特征层及其网格点的位置，在对应的y_true中将内容进行保存。
代码如下：
def preprocess_true_boxes(true_boxes, input_shape, anchors, num_classes):
assert (true_boxes[..., 4]<num_classes).all(), 'class id must be less than num_classes'
# 一共有三个特征层数
num_layers = len(anchors)//3
# 先验框
# 678为116,90, 156,198, 373,326
# 345为30,61, 62,45, 59,119
# 012为10,13, 16,30, 33,23,
anchor_mask = [[6,7,8], [3,4,5], [0,1,2]] if num_layers==3 else [[3,4,5], [1,2,3]]
true_boxes = np.array(true_boxes, dtype='float32')
input_shape = np.array(input_shape, dtype='int32') # 416,416
# 读出xy轴，读出长宽
# 中心点(m,n,2)
boxes_xy = (true_boxes[..., 0:2] + true_boxes[..., 2:4]) // 2
boxes_wh = true_boxes[..., 2:4] - true_boxes[..., 0:2]
# 计算比例
true_boxes[..., 0:2] = boxes_xy/input_shape[:]
true_boxes[..., 2:4] = boxes_wh/input_shape[:]
# m张图
m = true_boxes.shape[0]
# 得到网格的shape为13,13;26,26;52,52
grid_shapes = [input_shape//{0:32, 1:16, 2:8}[l] for l in range(num_layers)]
# y_true的格式为(m,13,13,3,85)(m,26,26,3,85)(m,52,52,3,85)
y_true = [np.zeros((m,grid_shapes[l][0],grid_shapes[l][1],len(anchor_mask[l]),5+num_classes),
dtype='float32') for l in range(num_layers)]
# [1,9,2]
anchors = np.expand_dims(anchors, 0)
anchor_maxes = anchors / 2.
anchor_mins = -anchor_maxes
# 长宽要大于0才有效
valid_mask = boxes_wh[..., 0]>0
for b in range(m):
# 对每一张图进行处理
wh = boxes_wh[b, valid_mask[b]]
if len(wh)==0: continue
# [n,1,2]
wh = np.expand_dims(wh, -2)
box_maxes = wh / 2.
box_mins = -box_maxes
# 计算真实框和哪个先验框最契合
intersect_mins = np.maximum(box_mins, anchor_mins)
intersect_maxes = np.minimum(box_maxes, anchor_maxes)
intersect_wh = np.maximum(intersect_maxes - intersect_mins, 0.)
intersect_area = intersect_wh[..., 0] * intersect_wh[..., 1]
box_area = wh[..., 0] * wh[..., 1]
anchor_area = anchors[..., 0] * anchors[..., 1]
iou = intersect_area / (box_area + anchor_area - intersect_area)
# 维度是(n) 感谢 消尽不死鸟 的提醒
best_anchor = np.argmax(iou, axis=-1)
for t, n in enumerate(best_anchor):
for l in range(num_layers):
if n in anchor_mask[l]:
# floor用于向下取整
i = np.floor(true_boxes[b,t,0]*grid_shapes[l][1]).astype('int32')
j = np.floor(true_boxes[b,t,1]*grid_shapes[l][0]).astype('int32')
# 找到真实框在特征层l中第b副图像对应的位置
k = anchor_mask[l].index(n)
c = true_boxes[b,t, 4].astype('int32')
y_true[l][b, j, i, k, 0:4] = true_boxes[b,t, 0:4]
y_true[l][b, j, i, k, 4] = 1
y_true[l][b, j, i, k, 5+c] = 1
return y_true
三、loss值的计算
作者计算计算其中所有真实框与预测框的IOU，取出每个网络点中IOU最大的先验框，如果这个最大的IOU都小于ignore_thresh，则保留，一般来说ignore_thresh取0.5，该步的目的是为了平衡负样本。首先我个人认为作者处理正负样本平衡的时候并有处理得相当好，可能是我基础不够没有理解到作者的深刻用意吧，我尝试换成focal loss，我们在口罩识别上的效果并没有以前好。
作者利用y_true取出该特征层中真实存在目标的点的位置(m,13,13,3,1)及其对应的种类(m,13,13,3,20)，将yolo_outputs的预测值输出进行处理，得到reshape后的预测值y_pre，shape分别为(m,13,13,3,25),(m,26,26,3,25),(m,52,52,3,25)，还获取真实框编码后的值，后面用于计算loss。作者总共计算了3个loss的值
1.编码后的长宽与xy轴偏移量与预测值的差距。
2.测结果中置信度的值与1对比；实际不存在的框。
3.种类预测结果与实际结果的对比。
代码如下：
import numpy as np
import tensorflow as tf
from keras import backend as K
# ---------------------------------------------------#
# 将预测值的每个特征层调成真实值
# ---------------------------------------------------#
def yolo_head(feats, anchors, num_classes, input_shape, calc_loss=False):
num_anchors = len(anchors)
# [1, 1, 1, num_anchors, 2]
anchors_tensor = K.reshape(K.constant(anchors), [1, 1, 1, num_anchors, 2])
# 获得x，y的网格
# (13, 13, 1, 2)
grid_shape = K.shape(feats)[1:3] # height, width
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))
# (batch_size,13,13,3,85)
feats = K.reshape(feats, [-1, grid_shape[0], grid_shape[1], num_anchors, num_classes + 5])
# 将预测值调成真实值
# box_xy对应框的中心点
# box_wh对应框的宽和高
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))
box_confidence = K.sigmoid(feats[..., 4:5])
box_class_probs = K.sigmoid(feats[..., 5:])
# 在计算loss的时候返回如下参数
if calc_loss == True:
return grid, feats, box_xy, box_wh
return box_xy, box_wh, box_confidence, box_class_probs
# ---------------------------------------------------#
# 用于计算每个预测框与真实框的iou
# ---------------------------------------------------#
def box_iou(b1, b2):
# 13,13,3,1,4
# 计算左上角的坐标和右下角的坐标
b1 = K.expand_dims(b1, -2)
b1_xy = b1[..., :2]
b1_wh = b1[..., 2:4]
b1_wh_half = b1_wh / 2.
b1_mins = b1_xy - b1_wh_half
b1_maxes = b1_xy + b1_wh_half
# 1,n,4
# 计算左上角和右下角的坐标
b2 = K.expand_dims(b2, 0)
b2_xy = b2[..., :2]
b2_wh = b2[..., 2:4]
b2_wh_half = b2_wh / 2.
b2_mins = b2_xy - b2_wh_half
b2_maxes = b2_xy + b2_wh_half
# 计算重合面积
intersect_mins = K.maximum(b1_mins, b2_mins)
intersect_maxes = K.minimum(b1_maxes, b2_maxes)
intersect_wh = K.maximum(intersect_maxes - intersect_mins, 0.)
intersect_area = intersect_wh[..., 0] * intersect_wh[..., 1]
b1_area = b1_wh[..., 0] * b1_wh[..., 1]
b2_area = b2_wh[..., 0] * b2_wh[..., 1]
iou = intersect_area / (b1_area + b2_area - intersect_area)
return iou
# ---------------------------------------------------#
# loss值计算
# ---------------------------------------------------#
def yolo_loss(args, anchors, num_classes, ignore_thresh=.5, print_loss=False):
# 一共有三层
num_layers = len(anchors) // 3
# 将预测结果和实际ground truth分开，args是[*model_body.output, *y_true]
# y_true是一个列表，包含三个特征层，shape分别为(m,13,13,3,85),(m,26,26,3,85),(m,52,52,3,85)。
# yolo_outputs是一个列表，包含三个特征层，shape分别为(m,13,13,3,85),(m,26,26,3,85),(m,52,52,3,85)。
y_true = args[num_layers:]
yolo_outputs = args[:num_layers]
# 先验框
# 678为116,90, 156,198, 373,326
# 345为30,61, 62,45, 59,119
# 012为10,13, 16,30, 33,23,
anchor_mask = [[6, 7, 8], [3, 4, 5], [0, 1, 2]] if num_layers == 3 else [[3, 4, 5], [1, 2, 3]]
# 得到input_shpae为416,416
input_shape = K.cast(K.shape(yolo_outputs[0])[1:3] * 32, K.dtype(y_true[0]))
# 得到网格的shape为13,13;26,26;52,52
grid_shapes = [K.cast(K.shape(yolo_outputs[l])[1:3], K.dtype(y_true[0])) for l in range(num_layers)]
loss = 0
# 取出每一张图片
# m的值就是batch_size
m = K.shape(yolo_outputs[0])[0]
mf = K.cast(m, K.dtype(yolo_outputs[0]))
# y_true是一个列表，包含三个特征层，shape分别为(m,13,13,3,85),(m,26,26,3,85),(m,52,52,3,85)。
# yolo_outputs是一个列表，包含三个特征层，shape分别为(m,13,13,3,85),(m,26,26,3,85),(m,52,52,3,85)。
for l in range(num_layers):
# 以第一个特征层(m,13,13,3,85)为例子
# 取出该特征层中存在目标的点的位置。(m,13,13,3,1)
object_mask = y_true[l][..., 4:5]
# 取出其对应的种类(m,13,13,3,80)
true_class_probs = y_true[l][..., 5:]
# 将yolo_outputs的特征层输出进行处理
# grid为网格结构(13,13,1,2)，raw_pred为尚未处理的预测结果(m,13,13,3,85)
# 还有解码后的xy，wh，(m,13,13,3,2)
grid, raw_pred, pred_xy, pred_wh = yolo_head(yolo_outputs[l],
anchors[anchor_mask[l]], num_classes, input_shape, calc_loss=True)
# 这个是解码后的预测的box的位置
# (m,13,13,3,4)
pred_box = K.concatenate([pred_xy, pred_wh])
# 找到负样本群组，第一步是创建一个数组，[]
# print(K.dtype(y_true[0]))
ignore_mask = tf.TensorArray(K.dtype(y_true[0]), size=1, dynamic_size=True)
object_mask_bool = K.cast(object_mask, 'bool')
# 对每一张图片计算ignore_mask
def loop_body(b, ignore_mask):
# 取出第b副图内，真实存在的所有的box的参数
# n,4
true_box = tf.boolean_mask(y_true[l][b, ..., 0:4], object_mask_bool[b, ..., 0])
# 计算预测结果与真实情况的iou
# pred_box为13,13,3,4
# 计算的结果是每个pred_box和其它所有真实框的iou
# 13,13,3,n
iou = box_iou(pred_box[b], true_box)
# 13,13,3,1
best_iou = K.max(iou, axis=-1)
# 判断预测框的iou小于ignore_thresh则认为该预测框没有与之对应的真实框
# 则被认为是这幅图的负样本
ignore_mask = ignore_mask.write(b, K.cast(best_iou < ignore_thresh, K.dtype(true_box)))
return b + 1, ignore_mask
# 遍历所有的图片
_, ignore_mask = K.control_flow_ops.while_loop(lambda b, *args: b < m, loop_body, [0, ignore_mask])
# 将每幅图的内容压缩，进行处理
ignore_mask = ignore_mask.stack()
# (m,13,13,3,1,1)
ignore_mask = K.expand_dims(ignore_mask, -1)
# 将真实框进行编码，使其格式与预测的相同，后面用于计算loss
raw_true_xy = y_true[l][..., :2] * grid_shapes[l][:] - grid
raw_true_wh = K.log(y_true[l][..., 2:4] / anchors[anchor_mask[l]] * input_shape[::-1])
# object_mask如果真实存在目标则保存其wh值
# switch接口，就是一个if/else条件判断语句
raw_true_wh = K.switch(object_mask, raw_true_wh, K.zeros_like(raw_true_wh))
box_loss_scale = 2 - y_true[l][..., 2:3] * y_true[l][..., 3:4]
xy_loss = object_mask * box_loss_scale * K.binary_crossentropy(raw_true_xy, raw_pred[..., 0:2],
from_logits=True)
wh_loss = object_mask * box_loss_scale * 0.5 * K.square(raw_true_wh - raw_pred[..., 2:4])
# 如果该位置本来有框，那么计算1与置信度的交叉熵
# 如果该位置本来没有框，而且满足best_iou<ignore_thresh，则被认定为负样本
# best_iou<ignore_thresh用于限制负样本数量
confidence_loss = object_mask * K.binary_crossentropy(object_mask, raw_pred[..., 4:5], from_logits=True) + \
(1 - object_mask) * K.binary_crossentropy(object_mask, raw_pred[..., 4:5],
from_logits=True) * ignore_mask
class_loss = object_mask * K.binary_crossentropy(true_class_probs, raw_pred[..., 5:], from_logits=True)
xy_loss = K.sum(xy_loss) / mf
wh_loss = K.sum(wh_loss) / mf
confidence_loss = K.sum(confidence_loss) / mf
class_loss = K.sum(class_loss) / mf
loss += xy_loss + wh_loss + confidence_loss + class_loss
if print_loss:
loss = tf.Print(loss, [loss, xy_loss, wh_loss, confidence_loss, class_loss, K.sum(ignore_mask)],
message='loss: ')
with tf.Session() as sess:
# sess.run(K.shape(a)[1:3])
print(sess.run(ignore_mask))
return loss