模型介绍
特点：1.相比于MobileNetV1，先进行了1x1的卷积进行升维，目的在于获得更多特征，然后用3x3的空间卷积，最后再用1x1降维。核心思想是升维再降维，参数量更少。
2.为了避免Relu对特征的破坏，在在3x3网络结构前利用1x1卷积升维，在3x3网络结构后，再利用1x1卷积降维后，不再进行Relu6层，直接进行残差网络的加法。
PS：Relu对于负的输入，输出全为零；而本来特征就已经被“压缩”，再经过Relu的话，又要“损失”一部分特征，因此这里不采用Relu。
模型网络
t：1x1升维扩张的倍数。
c：输出通道数。
n：进行多少次。
s：步长。
如下图是bottleneck的结构，这里我们需要注意的是，步长维1和2的情况是不同的。stride=2时，不采用shortcut。因为维度变化，无法相加。
代码实现+图片预测
我们先将代码块分开说明。
首先，我们来看一下MoblieNet的更新点吧。
从上面我们知道： 1. (1,1)的pointwise卷积，提高通道数，结果Relu 2. (3,3)的depthwise卷积，提特征，结果Relu 3. (1,1)的pointwise卷积，单纯降维，不Relu 4.输入与输出的shortcut 所以在步长(2,2)的时候，要注意使用correct_pad填充特征图， 避免卷积出小数宽高特征图的情况。
这就是第一步啦，1x1扩张，relu6。
if block_id: # Expand x = layers.Conv2D(expansion * in_channels, kernel_size=1, padding='same', use_bias=False, activation=None, name=prefix + 'expand')(x) x = layers.BatchNormalization(axis=channel_axis, epsilon=1e-3, momentum=0.999, name=prefix + 'expand_BN')(x) x = layers.ReLU(6., name=prefix + 'expand_relu')(x) else: prefix = 'expanded_conv_'
在这里我们进行填充0的操作，为的是在3x3卷积后，长宽缩短到1/2，correct_pad是计算padding。
if stride == 2: x = ZeroPadding2D(padding=correct_pad(x, 3), name=prefix + 'pad')(x)
然后就是可分离卷积了。
# part2 可分离卷积 x = DepthwiseConv2D(kernel_size=3, strides=stride, activation=None, use_bias=False, padding='same' if stride == 1 else 'valid', name=prefix + 'depthwise')(x) x = BatchNormalization(epsilon=1e-3, momentum=0.999, name=prefix + 'depthwise_BN')(x) x = Activation(relu6, name=prefix + 'depthwise_relu')(x)
最后就是进行降维，这里不用relu。当stride=1以及输入输出通道数相同时，进行shortcut。
# part3压缩特征，而且不使用relu函数，保证特征不被破坏 x = Conv2D(pointwise_filters, kernel_size=1, padding='same', use_bias=False, activation=None, name=prefix + 'project')(x) x = BatchNormalization(epsilon=1e-3, momentum=0.999, name=prefix + 'project_BN')(x) if in_channels == pointwise_filters and stride == 1: return Add(name=prefix + 'add')([inputs, x]) return x
下面是代码整合
#-------------------------------------------------------------# # MobileNetV2的网络部分 #-------------------------------------------------------------# import math import numpy as np import tensorflow as tf from tensorflow.keras import backend from keras import backend as K from keras.preprocessing import image from keras.models import Model from keras.layers.normalization import BatchNormalization from keras.layers import Conv2D, Add, ZeroPadding2D, GlobalAveragePooling2D, Dropout, Dense from keras.layers import MaxPooling2D,Activation,DepthwiseConv2D,Input,GlobalMaxPooling2D from keras.applications import imagenet_utils from keras.applications.imagenet_utils import decode_predictions from keras.utils.data_utils import get_file # TODO Change path to v1.1 BASE_WEIGHT_PATH = ('https://github.com/JonathanCMitchell/mobilenet_v2_keras/' 'releases/download/v1.1/') # relu6！ def relu6(x): return K.relu(x, max_value=6) # 用于计算padding的大小 def correct_pad(inputs, kernel_size): img_dim = 1 input_size = backend.int_shape(inputs)[img_dim:(img_dim + 2)] if isinstance(kernel_size, int): kernel_size = (kernel_size, kernel_size) if input_size[0] is None: adjust = (1, 1) else: adjust = (1 - input_size[0] % 2, 1 - input_size[1] % 2) correct = (kernel_size[0] // 2, kernel_size[1] // 2) return ((correct[0] - adjust[0], correct[0]), (correct[1] - adjust[1], correct[1])) # 使其结果可以被8整除，因为使用到了膨胀系数α def _make_divisible(v, divisor, min_value=None): if min_value is None: min_value = divisor new_v = max(min_value, int(v + divisor / 2) // divisor * divisor) if new_v < 0.9 * v: new_v += divisor return new_v def MobileNetV2(input_shape=[224,224,3], alpha=1.0, include_top=True, weights='imagenet', classes=1000): rows = input_shape[0] img_input = Input(shape=input_shape) # stem部分 # 224,224,3 -> 112,112,32 first_block_filters = _make_divisible(32 * alpha, 8) x = ZeroPadding2D(padding=correct_pad(img_input, 3), name='Conv1_pad')(img_input) x = Conv2D(first_block_filters, kernel_size=3, strides=(2, 2), padding='valid', use_bias=False, name='Conv1')(x) x = BatchNormalization(epsilon=1e-3, momentum=0.999, name='bn_Conv1')(x) x = Activation(relu6, name='Conv1_relu')(x) # 112,112,32 -> 112,112,16 x = _inverted_res_block(x, filters=16, alpha=alpha, stride=1, expansion=1, block_id=0) # 112,112,16 -> 56,56,24 x = _inverted_res_block(x, filters=24, alpha=alpha, stride=2, expansion=6, block_id=1) x = _inverted_res_block(x, filters=24, alpha=alpha, stride=1, expansion=6, block_id=2) # 56,56,24 -> 28,28,32 x = _inverted_res_block(x, filters=32, alpha=alpha, stride=2, expansion=6, block_id=3) x = _inverted_res_block(x, filters=32, alpha=alpha, stride=1, expansion=6, block_id=4) x = _inverted_res_block(x, filters=32, alpha=alpha, stride=1, expansion=6, block_id=5) # 28,28,32 -> 14,14,64 x = _inverted_res_block(x, filters=64, alpha=alpha, stride=2, expansion=6, block_id=6) x = _inverted_res_block(x, filters=64, alpha=alpha, stride=1, expansion=6, block_id=7) x = _inverted_res_block(x, filters=64, alpha=alpha, stride=1, expansion=6, block_id=8) x = _inverted_res_block(x, filters=64, alpha=alpha, stride=1, expansion=6, block_id=9) # 14,14,64 -> 14,14,96 x = _inverted_res_block(x, filters=96, alpha=alpha, stride=1, expansion=6, block_id=10) x = _inverted_res_block(x, filters=96, alpha=alpha, stride=1, expansion=6, block_id=11) x = _inverted_res_block(x, filters=96, alpha=alpha, stride=1, expansion=6, block_id=12) # 14,14,96 -> 7,7,160 x = _inverted_res_block(x, filters=160, alpha=alpha, stride=2, expansion=6, block_id=13) x = _inverted_res_block(x, filters=160, alpha=alpha, stride=1, expansion=6, block_id=14) x = _inverted_res_block(x, filters=160, alpha=alpha, stride=1, expansion=6, block_id=15) # 7,7,160 -> 7,7,320 x = _inverted_res_block(x, filters=320, alpha=alpha, stride=1, expansion=6, block_id=16) if alpha > 1.0: last_block_filters = _make_divisible(1280 * alpha, 8) else: last_block_filters = 1280 # 7,7,320 -> 7,7,1280 x = Conv2D(last_block_filters, kernel_size=1, use_bias=False, name='Conv_1')(x) x = BatchNormalization(epsilon=1e-3, momentum=0.999, name='Conv_1_bn')(x) x = Activation(relu6, name='out_relu')(x) x = GlobalAveragePooling2D()(x) x = Dense(classes, activation='softmax', use_bias=True, name='Logits')(x) inputs = img_input model = Model(inputs, x, name='mobilenetv2_%0.2f_%s' % (alpha, rows)) # Load weights. if weights == 'imagenet': if include_top: model_name = ('mobilenet_v2_weights_tf_dim_ordering_tf_kernels_' + str(alpha) + '_' + str(rows) + '.h5') weight_path = BASE_WEIGHT_PATH + model_name weights_path = get_file( model_name, weight_path, cache_subdir='models') else: model_name = ('mobilenet_v2_weights_tf_dim_ordering_tf_kernels_' + str(alpha) + '_' + str(rows) + '_no_top' + '.h5') weight_path = BASE_WEIGHT_PATH + model_name weights_path = get_file( model_name, weight_path, cache_subdir='models') model.load_weights(weights_path) elif weights is not None: model.load_weights(weights) return model def _inverted_res_block(inputs, expansion, stride, alpha, filters, block_id): in_channels = backend.int_shape(inputs)[-1] pointwise_conv_filters = int(filters * alpha) pointwise_filters = _make_divisible(pointwise_conv_filters, 8) x = inputs prefix = 'block_{}_'.format(block_id) # part1 数据扩张 if block_id: # Expand x = Conv2D(expansion * in_channels, kernel_size=1, padding='same', use_bias=False, activation=None, name=prefix + 'expand')(x) x = BatchNormalization(epsilon=1e-3, momentum=0.999, name=prefix + 'expand_BN')(x) x = Activation(relu6, name=prefix + 'expand_relu')(x) else: prefix = 'expanded_conv_' if stride == 2: x = ZeroPadding2D(padding=correct_pad(x, 3), name=prefix + 'pad')(x) # part2 可分离卷积 x = DepthwiseConv2D(kernel_size=3, strides=stride, activation=None, use_bias=False, padding='same' if stride == 1 else 'valid', name=prefix + 'depthwise')(x) x = BatchNormalization(epsilon=1e-3, momentum=0.999, name=prefix + 'depthwise_BN')(x) x = Activation(relu6, name=prefix + 'depthwise_relu')(x) # part3压缩特征，而且不使用relu函数，保证特征不被破坏 x = Conv2D(pointwise_filters, kernel_size=1, padding='same', use_bias=False, activation=None, name=prefix + 'project')(x) x = BatchNormalization(epsilon=1e-3, momentum=0.999, name=prefix + 'project_BN')(x) if in_channels == pointwise_filters and stride == 1: return Add(name=prefix + 'add')([inputs, x]) return x def preprocess_input(x): x /= 255. x -= 0.5 x *= 2. return x if __name__ == '__main__': model = MobileNetV2(input_shape=(224, 224, 3)) model.summary() img_path = 'elephant.jpg' img = image.load_img(img_path, target_size=(224, 224)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) print('Input image shape:', x.shape) preds = model.predict(x) print(np.argmax(preds)) print('Predicted:', decode_predictions(preds, 1))
