【YOLOv8改进- Backbone主干】YOLOv8 更换主干网络之EfficientNet，高效的卷积神经网络，降低参数量

YOLO大王

于 2024-07-15 23:00:05 发布

阅读量469

点赞数 7

文章标签： YOLO 网络 cnn 目标检测神经网络 yolov8

原创文章，禁止任何形式转载！

本文链接：https://blog.csdn.net/shangyanaf/article/details/140451442

版权

YOLOv8目标检测创新改进与实战案例专栏

专栏目录： YOLOv8有效改进系列及项目实战目录包含卷积，主干注意力，检测头等创新机制以及各种目标检测分割项目实战案例

专栏链接: YOLOv8基础解析+创新改进+实战案例

介绍

摘要

卷积神经网络（ConvNets）通常在固定的资源预算下开发，如果有更多资源可用，则会进行扩展以提高准确性。在本文中，我们系统地研究了模型扩展，并发现仔细平衡网络的深度、宽度和分辨率可以带来更好的性能。基于这一观察，我们提出了一种新的扩展方法，使用一个简单但非常有效的复合系数均匀扩展深度、宽度和分辨率的所有维度。我们展示了这种方法在扩展MobileNets和ResNet时的有效性。

为了进一步提高，我们使用神经架构搜索设计了一个新的基准网络，并将其扩展，获得了一系列称为EfficientNets的模型，这些模型比以前的ConvNets在准确性和效率方面都有了很大的提升。特别是，我们的EfficientNet-B7在ImageNet上实现了最先进的84.3%的top-1准确率，同时在推理时比现有的最佳ConvNet小8.4倍，快6.1倍。我们的EfficientNets在迁移学习任务中也表现良好，在CIFAR-100（91.7%）、Flowers（98.8%）和其他3个迁移学习数据集上实现了最先进的准确率，参数量减少了一个数量级。源码可在：https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet 找到。

文章链接

论文地址：论文地址

代码地址：代码地址

基本原理

卷积神经网络（ConvNets）通常是在固定资源预算下开发的，当有更多资源可用时，可以通过增加网络深度（depth）、网络宽度（width）和输入图像分辨率（resolution）来提高精度。然而，手动调整这些参数组合非常困难，尤其在计算资源有限的情况下，不同参数的组合空间太大，难以穷举。

创新方法
本文提出了一种新的模型缩放方法，通过使用一个简单而高效的复合系数，从深度（depth）、宽度（width）和分辨率（resolution）三个维度放大网络。相比传统方法，该方法不会随意缩放网络的维度，结合神经结构搜索技术，可以获得最优的一组参数（复合系数）。

复合模型扩张方法

问题定义：卷积网络N可以分为多个阶段，每个阶段由多个相同结构的卷积层组成。
优化目标：在资源有限的情况下，最大化精度（Accuracy）。更大的网络具有更大的宽度、深度或分辨率，可以获得更高精度，但单一维度的扩展效果有限。
模型扩张的局限性：只对单一维度进行扩张的精度增益迅速饱和，说明需要平衡各个维度的扩张。

复合扩张方法的求解

求解参数：通过固定φ=1，通过网格搜索（grid search）得到最优的α、β、γ，得到基本模型EfficientNet-B0。
扩展模型：固定α、β、γ的值，通过调整φ的大小，获得EfficientNet-B1到B7。φ的大小决定了资源消耗的大小。

核心代码

# Copyright 2019 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""Contains definitions EfficientNet."""
from absl import logging
import tensorflow.compat.v1 as tf

import sys
sys.path.append('tpu/models/official/efficientnet')
from modeling.architecture import efficientnet_constants
from modeling.architecture import nn_blocks
from modeling.architecture import nn_ops
from official.efficientnet import efficientnet_builder


class Efficientnet(object):
  """Class to build EfficientNet family models."""

  def __init__(self,
               model_name):
    """EfficientNet initialization function.

    Args:
      model_name: string, the EfficientNet model name, e.g., `efficient-b0`.
    """
    self._model_name = model_name

  def __call__(self, inputs, is_training=False):
    """Returns features at various levels for EfficientNet model.

    Args:
      inputs: a `Tesnor` with shape [batch_size, height, width, 3] representing
        a batch of images.
      is_training: `bool` if True, the model is in training mode.

    Returns:
      a `dict` containing `int` keys for continuous feature levels [2, 3, 4, 5].
      The values are corresponding feature hierarchy in EfficientNet with shape
      [batch_size, height_l, width_l, num_filters].
    """
    _, endpoints = efficientnet_builder.build_model(
        inputs,
        self._model_name,
        training=is_training,
        override_params=None)
    u2 = endpoints['reduction_2']
    u3 = endpoints['reduction_3']
    u4 = endpoints['reduction_4']
    u5 = endpoints['reduction_5']
    return {2: u2, 3: u3, 4: u4, 5: u5}


class BlockSpec(object):
  """A container class that specifies the block configuration for EfficientNet."""

  def __init__(self, num_repeats, block_fn, expand_ratio, kernel_size, se_ratio,
               output_filters, act_fn):
    self.num_repeats = num_repeats
    self.block_fn = block_fn
    self.expand_ratio = expand_ratio
    self.kernel_size = kernel_size
    self.se_ratio = se_ratio
    self.output_filters = output_filters
    self.act_fn = act_fn


def build_block_specs(block_specs=None):
  """Builds the list of BlockSpec objects for EfficientNet."""
  if not block_specs:
    block_specs = efficientnet_constants.EFFICIENTNET_X_B0_BLOCK_SPECS
  if len(block_specs) != efficientnet_constants.EFFICIENTNET_NUM_BLOCKS:
    raise ValueError(
        'The block_specs of EfficientNet must be a length {} list.'.format(
            efficientnet_constants.EFFICIENTNET_NUM_BLOCKS))
  logging.info('Building EfficientNet block specs: %s', block_specs)
  return [BlockSpec(*b) for b in block_specs]


class EfficientNetX(object):
  """Class to build EfficientNet and X family models."""

  def __init__(self,
               block_specs=build_block_specs(),
               batch_norm_activation=nn_ops.BatchNormActivation(),
               data_format='channels_last'):
    """EfficientNet initialization function.

    Args:
      block_specs: a list of BlockSpec objects that specifies the EfficientNet
        network. By default, the previously discovered EfficientNet-A1 is used.
      batch_norm_activation: an operation that includes a batch normalization
        layer followed by an optional activation layer.
      data_format: An optional string from: "channels_last", "channels_first".
        Defaults to "channels_last".
    """
    self._block_specs = block_specs
    self._batch_norm_activation = batch_norm_activation
    self._data_format = data_format

  def __call__(self, images, is_training=False):
    """Generate a multiscale feature pyramid.

    Args:
      images: The input image tensor.
      is_training: `bool` if True, the model is in training mode.

    Returns:
      a `dict` containing `int` keys for continuous feature levels
      [min_level, min_level + 1, ..., max_level]. The values are corresponding
      features with shape [batch_size, height_l, width_l,
      endpoints_num_filters].
    """
    x = images
    with tf.variable_scope('efficientnet'):
      x = nn_ops.conv2d_fixed_padding(
          inputs=x,
          filters=32,
          kernel_size=3,
          strides=2,
          data_format=self._data_format)
      x = tf.identity(x, 'initial_conv')
      x = self._batch_norm_activation(x, is_training=is_training)

      endpoints = []
      for i, block_spec in enumerate(self._block_specs):
        bn_act = nn_ops.BatchNormActivation(activation=block_spec.act_fn)
        with tf.variable_scope('block_{}'.format(i)):
          for j in range(block_spec.num_repeats):
            strides = (1 if j > 0 else
                       efficientnet_constants.EFFICIENTNET_STRIDES[i])

            if block_spec.block_fn == 'conv':
              x = nn_ops.conv2d_fixed_padding(
                  inputs=x,
                  filters=block_spec.output_filters,
                  kernel_size=block_spec.kernel_size,
                  strides=strides,
                  data_format=self._data_format)
              x = bn_act(x, is_training=is_training)
            elif block_spec.block_fn == 'mbconv':
              x_shape = x.get_shape().as_list()
              in_filters = (
                  x_shape[1]
                  if self._data_format == 'channel_first' else x_shape[-1])
              x = nn_blocks.mbconv_block(
                  inputs=x,
                  in_filters=in_filters,
                  out_filters=block_spec.output_filters,
                  expand_ratio=block_spec.expand_ratio,
                  strides=strides,
                  kernel_size=block_spec.kernel_size,
                  se_ratio=block_spec.se_ratio,
                  batch_norm_activation=bn_act,
                  data_format=self._data_format,
                  is_training=is_training)
            elif block_spec.block_fn == 'fused_mbconv':
              x_shape = x.get_shape().as_list()
              in_filters = (
                  x_shape[1]
                  if self._data_format == 'channel_first' else x_shape[-1])
              x = nn_blocks.fused_mbconv_block(
                  inputs=x,
                  in_filters=in_filters,
                  out_filters=block_spec.output_filters,
                  expand_ratio=block_spec.expand_ratio,
                  strides=strides,
                  kernel_size=block_spec.kernel_size,
                  se_ratio=block_spec.se_ratio,
                  batch_norm_activation=bn_act,
                  data_format=self._data_format,
                  is_training=is_training)
            else:
              raise ValueError('Un-supported block_fn `{}`!'.format(
                  block_spec.block_fn))
          x = tf.identity(x, 'endpoints')
          endpoints.append(x)

    return {2: endpoints[1], 3: endpoints[2], 4: endpoints[4], 5: endpoints[6]}

下载YoloV8代码

直接下载

GitHub地址

Git Clone

git clone https://github.com/ultralytics/ultralytics

安装环境

进入代码根目录并安装依赖。

pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple

在最新版本中，官方已经废弃了requirements.txt文件，转而将所有必要的代码和依赖整合进了ultralytics包中。因此，用户只需安装这个单一的ultralytics库，就能获得所需的全部功能和环境依赖。

pip install ultralytics

引入代码

在根目录下的ultralytics/nn/目录，新建一个 backbone目录，然后新建一个以 EfficientNet为文件名的py文件，把代码拷贝进去。

import re
import math
import collections
from functools import partial
import torch
from torch import nn
from torch.nn import functional as F
from torch.utils import model_zoo

GlobalParams = collections.namedtuple('GlobalParams', [
    'width_coefficient', 'depth_coefficient', 'image_size', 'dropout_rate',
    'num_classes', 'batch_norm_momentum', 'batch_norm_epsilon',
    'drop_connect_rate', 'depth_divisor', 'min_depth', 'include_top'])

BlockArgs = collections.namedtuple('BlockArgs', [
    'num_repeat', 'kernel_size', 'stride', 'expand_ratio',
    'input_filters', 'output_filters', 'se_ratio', 'id_skip'])

GlobalParams.__new__.__defaults__ = (None,) * len(GlobalParams._fields)
BlockArgs.__new__.__defaults__ = (None,) * len(BlockArgs._fields)

if hasattr(nn, 'SiLU'):
    Swish = nn.SiLU
else:
    class Swish(nn.Module):
        def forward(self, x):
            return x * torch.sigmoid(x)

class SwishImplementation(torch.autograd.Function):
    @staticmethod
    def forward(ctx, i):
        result = i * torch.sigmoid(i)
        ctx.save_for_backward(i)
        return result

    @staticmethod
    def backward(ctx, grad_output):
        i = ctx.saved_tensors[0]
        sigmoid_i = torch.sigmoid(i)
        return grad_output * (sigmoid_i * (1 + i * (1 - sigmoid_i)))

class MemoryEfficientSwish(nn.Module):
    def forward(self, x):
        return SwishImplementation.apply(x)

def round_filters(filters, global_params):
    multiplier = global_params.width_coefficient
    if not multiplier:
        return filters
    divisor = global_params.depth_divisor
    min_depth = global_params.min_depth
    filters *= multiplier
    min_depth = min_depth or divisor
    new_filters = max(min_depth, int(filters + divisor / 2) // divisor * divisor)
    if new_filters < 0.9 * filters:
        new_filters += divisor
    return int(new_filters)

def round_repeats(repeats, global_params):
    multiplier = global_params.depth_coefficient
    if not multiplier:
        return repeats
    return int(math.ceil(multiplier * repeats))

def drop_connect(inputs, p, training):
    assert 0 <= p <= 1, 'p must be in range of [0,1]'
    if not training:
        return inputs
    batch_size = inputs.shape[0]
    keep_prob = 1 - p
    random_tensor = keep_prob
    random_tensor += torch.rand([batch_size, 1, 1, 1], dtype=inputs.dtype, device=inputs.device)
    binary_tensor = torch.floor(random_tensor)
    output = inputs / keep_prob * binary_tensor
    return output

def get_width_and_height_from_size(x):
    if isinstance(x, int):
        return x, x
    if isinstance(x, list) or isinstance(x, tuple):
        return x
    else:
        raise TypeError()

def calculate_output_image_size(input_image_size, stride):
    if input_image_size is None:
        return None
    image_height, image_width = get_width_and_height_from_size(input_image_size)
    stride = stride if isinstance(stride, int) else stride[0]
    image_height = int(math.ceil(image_height / stride))
    image_width = int(math.ceil(image_width / stride))
    return [image_height, image_width]

def get_same_padding_conv2d(image_size=None):
    if image_size is None:
        return Conv2dDynamicSamePadding
    else:
        return partial(Conv2dStaticSamePadding, image_size=image_size)

class Conv2dDynamicSamePadding(nn.Conv2d):
    def __init__(self, in_channels, out_channels, kernel_size, stride=1, dilation=1, groups=1, bias=True):
        super().__init__(in_channels, out_channels, kernel_size, stride, 0, dilation, groups, bias)
        self.stride = self.stride if len(self.stride) == 2 else [self.stride[0]] * 2

    def forward(self, x):
        ih, iw = x.size()[-2:]
        kh, kw = self.weight.size()[-2:]
        sh, sw = self.stride
        oh, ow = math.ceil(ih / sh), math.ceil(iw / sw)
        pad_h = max((oh - 1) * self.stride[0] + (kh - 1) * self.dilation[0] + 1 - ih, 0)
        pad_w = max((ow - 1) * self.stride[1] + (kw - 1) * self.dilation[1] + 1 - iw, 0)
        if pad_h > 0 or pad_w > 0:
            x = F.pad(x, [pad_w // 2, pad_w - pad_w // 2, pad_h // 2, pad_h - pad_h // 2])
        return F.conv2d(x, self.weight, self.bias, self.stride, self.padding, self.dilation, self.groups)

class Conv2dStaticSamePadding(nn.Conv2d):
    def __init__(self, in_channels, out_channels, kernel_size, stride=1, image_size=None, **kwargs):
        super().__init__(in_channels, out_channels, kernel_size, stride, **kwargs)
        self.stride = self.stride if len(self.stride) == 2 else [self.stride[0]] * 2

        assert image_size is not None
        ih, iw = (image_size, image_size) if isinstance(image_size, int) else image_size
        kh, kw = self.weight.size()[-2:]
        sh, sw = self.stride
        oh, ow = math.ceil(ih / sh), math.ceil(iw / sw)
        pad_h = max((oh - 1) * self.stride[0] + (kh - 1) * self.dilation[0] + 1 - ih, 0)
        pad_w = max((ow - 1) * self.stride[1] + (kw - 1) * self.dilation[1] + 1 - iw, 0)
        if pad_h > 0 or pad_w > 0:
            self.static_padding = nn.ZeroPad2d((pad_w // 2, pad_w - pad_w // 2, pad_h // 2, pad_h - pad_h // 2))
        else:
            self.static_padding = nn.Identity()

    def forward(self, x):
        x = self.static_padding(x)
        x = F.conv2d(x, self.weight, self.bias, self.stride, self.padding, self.dilation, self.groups)
        return x

def get_same_padding_maxPool2d(image_size=None):
    if image_size is None:
        return MaxPool2dDynamicSamePadding
    else:
        return partial(MaxPool2dStaticSamePadding, image_size=image_size)

class MaxPool2dDynamicSamePadding(nn.MaxPool2d):
    def __init__(self, kernel_size, stride, padding=0, dilation=1, return_indices=False, ceil_mode=False):
        super().__init__(kernel_size, stride, padding, dilation, return_indices, ceil_mode)
        self.stride = [self.stride] * 2 if isinstance(self.stride, int) else self.stride
        self.kernel_size = [self.kernel_size] * 2 if isinstance(self.kernel_size, int) else self.kernel_size
        self.dilation = [self.dilation] * 2 if isinstance(self.dilation, int) else self.dilation

    def forward(self, x):
        ih, iw = x.size()[-2:]
        kh, kw = self.kernel_size
        sh, sw = self.stride
        oh, ow = math.ceil(ih / sh), math.ceil(iw / sw)
        pad_h = max((oh - 1) * self.stride[0] + (kh - 1) * self.dilation[0] + 1 - ih, 0)
        pad_w = max((ow - 1) * self.stride[1] + (kw - 1) * self.dilation[1] + 1 - iw, 0)
        if pad_h > 0 or pad_w > 0:
            x = F.pad(x, [pad_w // 2, pad_w - pad_w // 2, pad_h // 2, pad_h - pad_h // 2])
        return F.max_pool2d(x, self.kernel_size, self.stride, self.padding,
                            self.dilation, self.ceil_mode, self.return_indices)

class MaxPool2dStaticSamePadding(nn.MaxPool2d):
    def __init__(self, kernel_size, stride, image_size=None, **kwargs):
        super().__init__(kernel_size, stride, **kwargs)
        self.stride = [self.stride] * 2 if isinstance(self.stride, int) else self.stride
        self.kernel_size = [self.kernel_size] * 2 if isinstance(self.kernel_size, int) else self.kernel_size
        self.dilation = [self.dilation] * 2 if isinstance(self.dilation, int) else self.dilation

        assert image_size is not None
        ih, iw = (image_size, image_size) if isinstance(image_size, int) else image_size
        kh, kw = self.kernel_size
        sh, sw = self.stride
        oh, ow = math.ceil(ih / sh), math.ceil(iw / sw)
        pad_h = max((oh - 1) * self.stride[0] + (kh - 1) * self.dilation[0] + 1 - ih, 0)
        pad_w = max((ow - 1) * self.stride[1] + (kw - 1) * self.dilation[1] + 1 - iw, 0)
        if pad_h > 0 or pad_w > 0:
            self.static_padding = nn.ZeroPad2d((pad_w // 2, pad_w - pad_w // 2, pad_h // 2, pad_h - pad_h // 2))
        else:
            self.static_padding = nn.Identity()

    def forward(self, x):
        x = self.static_padding(x)
        x = F.max_pool2d(x, self.kernel_size, self.stride, self.padding,
                         self.dilation, self.ceil_mode, self.return_indices)
        return x

class BlockDecoder(object):
    @staticmethod
    def _decode_block_string(block_string):
        assert isinstance(block_string, str)

        ops = block_string.split('_')
        options = {}
        for op in ops:
            splits = re.split(r'(\d.*)', op)
            if len(splits) >= 2:
                key, value = splits[:2]
                options[key] = value

        assert (('s' in options and len(options['s']) == 1) or
                (len(options['s']) == 2 and options['s'][0] == options['s'][1]))

        return BlockArgs(
            num_repeat=int(options['r']),
            kernel_size=int(options['k']),
            stride=[int(options['s'][0])],
            expand_ratio=int(options['e']),
            input_filters=int(options['i']),
            output_filters=int(options['o']),
            se_ratio=float(options['se']) if 'se' in options else None,
            id_skip=('noskip' not in block_string))

    @staticmethod
    def _encode_block_string(block):
        args = [
            'r%d' % block.num_repeat,
            'k%d' % block.kernel_size,
            's%d%d' % (block.strides[0], block.strides[1]),
            'e%s' % block.expand_ratio,
            'i%d' % block.input_filters,
            'o%d' % block.output_filters
        ]
        if 0 < block.se_ratio <= 1:
            args.append('se%s' % block.se_ratio)
        if block.id_skip is False:
            args.append('noskip')
        return '_'.join(args)

    @staticmethod
    def decode(string_list):
        assert isinstance(string_list, list)
        blocks_args = []
        for block_string in string_list:
            blocks_args.append(BlockDecoder._decode_block_string(block_string))
        return blocks_args

    @staticmethod
    def encode(blocks_args):
        block_strings = []
        for block in blocks_args:
            block_strings.append(BlockDecoder._encode_block_string(block))
        return block_strings

def efficientnet_params(model_name):
    params_dict = {
        'efficientnet-b0': (1.0, 1.0, 224, 0.2),
        'efficientnet-b1': (1.0, 1.1, 240, 0.2),
        'efficientnet-b2': (1.1, 1.2, 260, 0.3),
        'efficientnet-b3': (1.2, 1.4, 300, 0.3),
        'efficientnet-b4': (1.4, 1.8, 380, 0.4),
        'efficientnet-b5': (1.6, 2.2, 456, 0.4),
        'efficientnet-b6': (1.8, 2.6, 528, 0.5),
        'efficientnet-b7': (2.0, 3.1, 600, 0.5),
        'efficientnet-b8': (2.2, 3.6, 672, 0.5),
        'efficientnet-l2': (4.3, 5.3, 800, 0.5),
    }
    return params_dict[model_name]

def efficientnet(width_coefficient=None, depth_coefficient=None, image_size=None,
                 dropout_rate=0.2, drop_connect_rate=0.2, num_classes=1000, include_top=True):
    blocks_args = [
        'r1_k3_s11_e1_i32_o16_se0.25',
        'r2_k3_s22_e6_i16_o24_se0.25',
        'r2_k5_s22_e6_i24_o40_se0.25',
        'r3_k3_s22_e6_i40_o80_se0.25',
        'r3_k5_s11_e6_i80_o112_se0.25',
        'r4_k5_s22_e6_i112_o192_se0.25',
        'r1_k3_s11_e6_i192_o320_se0.25',
    ]
    blocks_args = BlockDecoder.decode(blocks_args)

    global_params = GlobalParams(
        width_coefficient=width_coefficient,
        depth_coefficient=depth_coefficient,
        image_size=image_size,
        dropout_rate=dropout_rate,

        num_classes=num_classes,
        batch_norm_momentum=0.99,
        batch_norm_epsilon=1e-3,
        drop_connect_rate=drop_connect_rate,
        depth_divisor=8,
        min_depth=None,
        include_top=include_top,
    )

    return blocks_args, global_params

def get_model_params(model_name, override_params):
    if model_name.startswith('efficientnet'):
        w, d, s, p = efficientnet_params(model_name)
        blocks_args, global_params = efficientnet(
            width_coefficient=w, depth_coefficient=d, dropout_rate=p, image_size=s)
    else:
        raise NotImplementedError('model name is not pre-defined: {}'.format(model_name))
    if override_params:
        global_params = global_params._replace(**override_params)
    return blocks_args, global_params

url_map = {
    'efficientnet-b0': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b0-355c32eb.pth',
    'efficientnet-b1': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b1-f1951068.pth',
    'efficientnet-b2': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b2-8bb594d6.pth',
    'efficientnet-b3': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b3-5fb5a3c3.pth',
    'efficientnet-b4': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b4-6ed6700e.pth',
    'efficientnet-b5': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b5-b6417697.pth',
    'efficientnet-b6': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b6-c76e70fd.pth',
    'efficientnet-b7': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/efficientnet-b7-dcc49843.pth',
}

url_map_advprop = {
    'efficientnet-b0': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b0-b64d5a18.pth',
    'efficientnet-b1': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b1-0f3ce85a.pth',
    'efficientnet-b2': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b2-6e9d97e5.pth',
    'efficientnet-b3': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b3-cdd7c0f4.pth',
    'efficientnet-b4': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b4-44fb3a87.pth',
    'efficientnet-b5': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b5-86493f6b.pth',
    'efficientnet-b6': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b6-ac80338e.pth',
    'efficientnet-b7': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b7-4652b6dd.pth',
    'efficientnet-b8': 'https://github.com/lukemelas/EfficientNet-PyTorch/releases/download/1.0/adv-efficientnet-b8-22a8fe65.pth',
}

def load_pretrained_weights(model, model_name, weights_path=None, load_fc=True, advprop=False, verbose=True):
    if isinstance(weights_path, str):
        state_dict = torch.load(weights_path)
    else:
        url_map_ = url_map_advprop if advprop else url_map
        state_dict = model_zoo.load_url(url_map_[model_name])

    if load_fc:
        ret = model.load_state_dict(state_dict, strict=False)
        assert not ret.missing_keys, 'Missing keys when loading pretrained weights: {}'.format(ret.missing_keys)
    else:
        state_dict.pop('_fc.weight')
        state_dict.pop('_fc.bias')
        ret = model.load_state_dict(state_dict, strict=False)
        assert set(ret.missing_keys) == set(
            ['_fc.weight', '_fc.bias']), 'Missing keys when loading pretrained weights: {}'.format(ret.missing_keys)
    assert not ret.unexpected_keys, 'Missing keys when loading pretrained weights: {}'.format(ret.unexpected_keys)

    if verbose:
        print('Loaded pretrained weights for {}'.format(model_name))

VALID_MODELS = (
    'efficientnet-b0', 'efficientnet-b1', 'efficientnet-b2', 'efficientnet-b3',
    'efficientnet-b4', 'efficientnet-b5', 'efficientnet-b6', 'efficientnet-b7',
    'efficientnet-b8',
    'efficientnet-l2'
)

class MBConvBlock(nn.Module):
    def __init__(self, block_args, global_params, image_size=None):
        super().__init__()
        self._block_args = block_args
        self._bn_mom = 1 - global_params.batch_norm_momentum
        self._bn_eps = global_params.batch_norm_epsilon
        self.has_se = (self._block_args.se_ratio is not None) and (0 < self._block_args.se_ratio <= 1)
        self.id_skip = block_args.id_skip

        inp = self._block_args.input_filters
        oup = self._block_args.input_filters * self._block_args.expand_ratio
        if self._block_args.expand_ratio != 1:
            Conv2d = get_same_padding_conv2d(image_size=image_size)
            self._expand_conv = Conv2d(in_channels=inp, out_channels=oup, kernel_size=1, bias=False)
            self._bn0 = nn.BatchNorm2d(num_features=oup, momentum=self._bn_mom, eps=self._bn_eps)

        k = self._block_args.kernel_size
        s = self._block_args.stride
        Conv2d = get_same_padding_conv2d(image_size=image_size)
        self._depthwise_conv = Conv2d(
            in_channels=oup, out_channels=oup, groups=oup,
            kernel_size=k, stride=s, bias=False)
        self._bn1 = nn.BatchNorm2d(num_features=oup, momentum=self._bn_mom, eps=self._bn_eps)
        image_size = calculate_output_image_size(image_size, s)

        if self.has_se:
            Conv2d = get_same_padding_conv2d(image_size=(1, 1))
            num_squeezed_channels = max(1, int(self._block_args.input_filters * self._block_args.se_ratio))
            self._se_reduce = Conv2d(in_channels=oup, out_channels=num_squeezed_channels, kernel_size=1)
            self._se_expand = Conv2d(in_channels=num_squeezed_channels, out_channels=oup, kernel_size=1)

        final_oup = self._block_args.output_filters
        Conv2d = get_same_padding_conv2d(image_size=image_size)
        self._project_conv = Conv2d(in_channels=oup, out_channels=final_oup, kernel_size=1, bias=False)
        self._bn2 = nn.BatchNorm2d(num_features=final_oup, momentum=self._bn_mom, eps=self._bn_eps)
        self._swish = MemoryEfficientSwish()

    def forward(self, inputs, drop_connect_rate=None):
        x = inputs
        if self._block_args.expand_ratio != 1:
            x = self._expand_conv(inputs)
            x = self._bn0(x)
            x = self._swish(x)

        x = self._depthwise_conv(x)
        x = self._bn1(x)
        x = self._swish(x)

        if self.has_se:
            x_squeezed = F.adaptive_avg_pool2d(x, 1)
            x_squeezed = self._se_reduce(x_squeezed)
            x_squeezed = self._swish(x_squeezed)
            x_squeezed = self._se_expand(x_squeezed)
            x = torch.sigmoid(x_squeezed) * x

        x = self._project_conv(x)
        x = self._bn2(x)

        input_filters, output_filters = self._block_args.input_filters, self._block_args.output_filters
        if self.id_skip and self._block_args.stride == 1 and input_filters == output_filters:
            if drop_connect_rate:
                x = drop_connect(x, p=drop_connect_rate, training=self.training)
            x = x + inputs
        return x

    def set_swish(self, memory_efficient=True):
        self._swish = MemoryEfficientSwish() if memory_efficient else Swish()

class EfficientNet(nn.Module):
    def __init__(self, blocks_args=None, global_params=None):
        super().__init__()
        assert isinstance(blocks_args, list), 'blocks_args should be a list'
        assert len(blocks_args) > 0, 'block args must be greater than 0'
        self._global_params = global_params
        self._blocks_args = blocks_args

        bn_mom = 1 - self._global_params.batch_norm_momentum
        bn_eps = self._global_params.batch_norm_epsilon

        image_size = global_params.image_size
        Conv2d = get_same_padding_conv2d(image_size=image_size)

        in_channels = 3
        out_channels = round_filters(32, self._global_params)
        self._conv_stem = Conv2d(in_channels, out_channels, kernel_size=3, stride=2, bias=False)
        self._bn0 = nn.BatchNorm2d(num_features=out_channels, momentum=bn_mom, eps=bn_eps)
        image_size = calculate_output_image_size(image_size, 2)

        self._blocks = nn.ModuleList([])
        for block_args in self._blocks_args:
            block_args = block_args._replace(
                input_filters=round_filters(block_args.input_filters, self._global_params),
                output_filters=round_filters(block_args.output_filters, self._global_params),
                num_repeat=round_repeats(block_args.num_repeat, self._global_params)
            )

            self._blocks.append(MBConvBlock(block_args, self._global_params, image_size=image_size))
            image_size = calculate_output_image_size(image_size, block_args.stride)
            if block_args.num_repeat > 1:
                block_args = block_args._replace(input_filters=block_args.output_filters, stride=1)
            for _ in range(block_args.num_repeat - 1):
                self._blocks.append(MBConvBlock(block_args, self._global_params, image_size=image_size))

        in_channels = block_args.output_filters
        out_channels = round_filters(1280, self._global_params)
        Conv2d = get_same_padding_conv2d(image_size=image_size)
        self._conv_head = Conv2d(in_channels, out_channels, kernel_size=1, bias=False)
        self._bn1 = nn.BatchNorm2d(num_features=out_channels, momentum=bn_mom, eps=bn_eps)

        self._avg_pooling = nn.AdaptiveAvgPool2d(1)
        if self._global_params.include_top:
            self._dropout = nn.Dropout(self._global_params.dropout_rate)
            self._fc = nn.Linear(out_channels, self._global_params.num_classes)

        self._swish = MemoryEfficientSwish()
        self.channels = [i.size(1) for i in self.forward(torch.randn(1, 3, 640, 640))]

    def set_swish(self, memory_efficient=True):
        self._swish = MemoryEfficientSwish() if memory_efficient else Swish()
        for block in self._blocks:
            block.set_swish(memory_efficient)

    def extract_endpoints(self, inputs):
        endpoints = dict()

        x = self._swish(self._bn0(self._conv_stem(inputs)))
        prev_x = x

        for idx, block in enumerate(self._blocks):
            drop_connect_rate = self._global_params.drop_connect_rate
            if drop_connect_rate:
                drop_connect_rate *= float(idx) / len(self._blocks)
            x = block(x, drop_connect_rate=drop_connect_rate)
            if prev_x.size(2) > x.size(2):
                endpoints['reduction_{}'.format(len(endpoints) + 1)] = prev_x
            elif idx == len(self._blocks) - 1:
                endpoints['reduction_{}'.format(len(endpoints) + 1)] = x
            prev_x = x

        x = self._swish(self._bn1(self._conv_head(x)))
        endpoints['reduction_{}'.format(len(endpoints) + 1)] = x

        return endpoints

    def forward(self, inputs):
        x = self._swish(self._bn0(self._conv_stem(inputs)))
        unique_tensors = {}
        for idx, block in enumerate(self._blocks):
            drop_connect_rate = self._global_params.drop_connect_rate
            if drop_connect_rate:
                drop_connect_rate *= float(idx) / len(self._blocks)
            x = block(x, drop_connect_rate=drop_connect_rate)
            width, height = x.shape[2], x.shape[3]
            unique_tensors[(width, height)] = x
        result_list = list(unique_tensors.values())[-4:]
        return result_list

    @classmethod
    def from_name(cls, model_name, in_channels=3, **override_params):
        cls._check_model_name_is_valid(model_name)
        blocks_args, global_params = get_model_params(model_name, override_params)
        model = cls(blocks_args, global_params)
        model._change_in_channels(in_channels)
        return model

    @classmethod
    def from_pretrained(cls, model_name, weights_path=None, advprop=False,
                        in_channels=3, num_classes=1000, **override_params):
        model = cls.from_name(model_name, num_classes=num_classes, **override_params)
        load_pretrained_weights(model, model_name, weights_path=weights_path,
                                load_fc=(num_classes == 1000), advprop=advprop)
        model._change_in_channels(in_channels)
        return model

    @classmethod
    def get_image_size(cls, model_name):
        cls._check_model_name_is_valid(model_name)
        _, _, res, _ = efficientnet_params(model_name)
        return res

    @classmethod
    def _check_model_name_is_valid(cls, model_name):
        if model_name not in VALID_MODELS:
            raise ValueError('model_name should be one of: ' + ', '.join(VALID_MODELS))

    def _change_in_channels(self, in_channels):
        if in_channels != 3:
            Conv2d = get_same_padding_conv2d(image_size=self._global_params.image_size)
            out_channels = round_filters(32, self._global_params)
            self._conv_stem = Conv2d(in_channels, out_channels, kernel_size=3, stride=2, bias=False)

def efficient(model_name='efficientnet-b0', pretrained=False):
    if pretrained:
        model = EfficientNet.from_pretrained('{}'.format(model_name))
    else:
        model = EfficientNet.from_name('{}'.format(model_name))
    return model

tasks注册

在ultralytics/nn/tasks.py中进行如下操作：

步骤1:

from ultralytics.nn.backbone.EfficientNet import  efficient

步骤2

修改def parse_model(d, ch, verbose=True):

        elif m in {efficient}:            
            m = m(*args)
            c2 = m.channels

在这里插入图片描述

增加代码1

在def parse_model(d, ch, verbose=True):增加t=m

    for i, (f, n, m, args) in enumerate(d["backbone"] + d["head"]):  # from, number, module, args
        t = m

修改代码

修改def parse_model(d, ch, verbose=True)中的：

        # m_ = nn.Sequential(*(m(*args) for _ in range(n))) if n > 1 else m(*args)  # module
        # t = str(m)[8:-2].replace("__main__.", "")  # module type
        # m.np = sum(x.numel() for x in m_.parameters())  # number params
        # m_.i, m_.f, m_.type = i, f, t  # attach index, 'from' index, type
        # if verbose:
        #     LOGGER.info(f"{i:>3}{str(f):>20}{n_:>3}{m.np:10.0f}  {t:<45}{str(args):<30}")  # print
        # save.extend(x % i for x in ([f] if isinstance(f, int) else f) if x != -1)  # append to savelist
        # layers.append(m_)
        # if i == 0:
        #     ch = []
        # ch.append(c2)

修改为

        if isinstance(c2, list):
      
            m_ = m
            m_.backbone = True
        else:
            m_ = nn.Sequential(*(m(*args) for _ in range(n))) if n > 1 else m(*args)  
            t = str(m)[8:-2].replace('__main__.', '')  
 
 
        m.np = sum(x.numel() for x in m_.parameters())   
        m_.i, m_.f, m_.type = i + 4 ,f,t
        if verbose:
            LOGGER.info(f'{i:>3}{str(f):>20}{n_:>3}{m.np:10.0f}  {t:<45}{str(args):<30}') 
 
        save.extend(x % (i + 4 ) for x in ([f] if isinstance(f, int) else f) if x != -1)  
        layers.append(m_)
        if i == 0:
            ch = []
        if isinstance(c2, list):
            ch.extend(c2)
            if len(c2) != 5:
                ch.insert(0, 0)
        else:
            ch.append(c2)

替换_predict_once方法

把_predict_once替换为下面的代码：

    def _predict_once(self, x, profile=False, visualize=False, embed=None):
        """
        Perform a forward pass through the network.
        Args:
            x (torch.Tensor): The input tensor to the model.
            profile (bool):  Print the computation time of each layer if True, defaults to False.
            visualize (bool): Save the feature maps of the model if True, defaults to False.
            embed (list, optional): A list of feature vectors/embeddings to return.
        Returns:
            (torch.Tensor): The last output of the model.
        """
        y, dt, embeddings = [], [], []   
        for m in self.model:
            if m.f != -1:  
                x = y[m.f] if isinstance(m.f, int) else [x if j == -1 else y[j] for j in m.f]  # from earlier layers
            if profile:
                self._profile_one_layer(m, x, dt)
            if hasattr(m, 'backbone'):
                x = m(x)
                if len(x) != 5:  # 0 - 5
                    x.insert(0, None)
                for index, i in enumerate(x):
                    if index in self.save:
                        y.append(i)
                    else:
                        y.append(None)
                x = x[-1]  
            else:
                x = m(x)  
                y.append(x if m.i in self.save else None)   
            if visualize:
                feature_visualization(x, m.type, m.i, save_dir=visualize)
            if embed and m.i in embed:
                embeddings.append(nn.functional.adaptive_avg_pool2d(x, (1, 1)).squeeze(-1).squeeze(-1))  # flatten
                if m.i == max(embed):
                    return torch.unbind(torch.cat(embeddings, 1), dim=0)
        return x

配置yolov8_EfficientNet.yaml

ultralytics/cfg/models/v8/yolov8_EfficientNet.yaml

# Ultralytics YOLO 🚀, AGPL-3.0 license
# YOLOv8 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
 
# Parameters
nc: 80  # number of classes
scales: # model compound scaling constants, i.e. 'model=yolov8n.yaml' will call yolov8.yaml with scale 'n'
  # [depth, width, max_channels]
  n: [0.33, 0.25, 1024]  # YOLOv8n summary: 225 layers,  3157200 parameters,  3157184 gradients,   8.9 GFLOPs
  s: [0.33, 0.50, 1024]  # YOLOv8s summary: 225 layers, 11166560 parameters, 11166544 gradients,  28.8 GFLOPs
  m: [0.67, 0.75, 768]   # YOLOv8m summary: 295 layers, 25902640 parameters, 25902624 gradients,  79.3 GFLOPs
  l: [1.00, 1.00, 512]   # YOLOv8l summary: 365 layers, 43691520 parameters, 43691504 gradients, 165.7 GFLOPs
  x: [1.00, 1.25, 512]   # YOLOv8x summary: 365 layers, 68229648 parameters, 68229632 gradients, 258.5 GFLOP
 
# YOLOv8.0n backbone
backbone:
  # [from, repeats, module, args]
  - [-1, 1, efficient, []]  # 4
  - [-1, 1, SPPF, [1024, 5]]  # 5
 
# YOLOv8.0n head
head:
  - [-1, 1, nn.Upsample, [None, 2, 'nearest']] # 6
  - [[-1, 3], 1, Concat, [1]]  # 7 cat backbone P4
  - [-1, 3, C2f, [512]]  # 8
 
  - [-1, 1, nn.Upsample, [None, 2, 'nearest']] # 9
  - [[-1, 2], 1, Concat, [1]]  # 10 cat backbone P3
  - [-1, 3, C2f, [256]]  # 11 (P3/8-small)
 
  - [-1, 1, Conv, [256, 3, 2]] # 12
  - [[-1, 8], 1, Concat, [1]]  # 13 cat head P4
  - [-1, 3, C2f, [512]]  # 14 (P4/16-medium)
 
  - [-1, 1, Conv, [512, 3, 2]] # 15
  - [[-1, 5], 1, Concat, [1]]  # 16 cat head P5
  - [-1, 3, C2f, [1024]]  # 17 (P5/32-large)
 
  - [[11, 14, 17], 1, Detect, [nc]]  # Detect(P3, P4, P5)

实验

脚本

import os
from ultralytics import YOLO
 
yaml = 'ultralytics/cfg/models/v8/yolov8_EfficientNet.yaml'

 
model = YOLO(yaml) 
 
model.info()
if __name__ == "__main__":
   
    results = model.train(data='ultralytics/datasets/original-license-plates.yaml',
                          name='yolov8_EfficientNet',
                          epochs=10, 
                          workers=8, 
                          batch=1)

结果

YOLO大王

关注

7
点赞
踩
6

收藏

觉得还不错? 一键收藏
打赏
0
评论
【YOLOv8改进- Backbone主干】YOLOv8 更换主干网络之EfficientNet，高效的卷积神经网络，降低参数量

卷积神经网络（ConvNets）通常在固定的资源预算下开发，如果有更多资源可用，则会进行扩展以提高准确性。在本文中，我们系统地研究了模型扩展，并发现仔细平衡网络的深度、宽度和分辨率可以带来更好的性能。基于这一观察，我们提出了一种新的扩展方法，使用一个简单但非常有效的复合系数均匀扩展深度、宽度和分辨率的所有维度。我们展示了这种方法在扩展MobileNets和ResNet时的有效性。
复制链接

扫一扫