Detectron2学习五：build_backbone_model配置及实现流程

class DefaultTrainer(SimpleTrainer):
    model = self.build_model(cfg)
                    ------->##/detectron2/engine/defaults.py
                           def build_model(cls,cfg):
                               model = build_model(cfg)
                                        ---------->##detectron2/modeling/meta_arch/build.py
                                                   def build_model(cfg):
                                                       meta_arch = cfg.MODEL.META_ARCHITECTURE
                                                       return META_ARCH_REGISTRY.get(meta_arch)(cfg)

####faster_rcnn_R_50_FPN.yaml
_BASE_: "../Base-RCNN-FPN.yaml"
MODEL:
  WEIGHTS: "detectron2://ImageNetPretrained/MSRA/R-50.pkl"
  MASK_ON: False
  RESNETS:
    DEPTH: 50
  ROI_HEADS:
    NUM_CLASSES: 20
...

###../Base-RCNN-FPN.yaml
MODEL:
  META_ARCHITECTURE: "GeneralizedRCNN"
  BACKBONE:
    NAME: "build_resnet_fpn_backbone"
  RESNETS:
    OUT_FEATURES: ["res2", "res3", "res4", "res5"]
  FPN:
    IN_FEATURES: ["res2", "res3", "res4", "res5"]
  ANCHOR_GENERATOR:
    SIZES: [[32], [64], [128], [256], [512]]  # One size for each in feature map
    ASPECT_RATIOS: [[0.5, 1.0, 2.0]]  # Three aspect ratios (same for all in feature maps)
...

__all__ = ["GeneralizedRCNN", "ProposalNetwork"]

@META_ARCH_REGISTRY.register()
class GeneralizedRCNN(nn.Module)：
    """
    Generalized R-CNN. Any models that contains the following three components:
    1. Per-image feature extraction (aka backbone)  图像特征提取主要是图像数据、通道、均值、方差的处理。
    2. Region proposal generation  ##候选区域的生成
    3. Per-region feature extraction and prediction ##特征提取及预测
    """
    
@META_ARCH_REGISTRY.register()
class ProposalNetwork(nn.Module):

 ##depth为配置文件中配置的MODEL.RESNETS.DEPTH
 ##当前默认支持res50,res101,res152三种网络结构，分为4个阶段，，每个阶段的配置blocks的个数如下
num_blocks_per_stage = {50: [3, 4, 6, 3], 101: [3, 4, 23, 3], 152: [3, 8, 36, 3]}[depth]

##4个阶段的输出名称配置如下：注意的是该名称与外面网络配置参数的名称要保持一致，否则找不到
##out_features为配置文件中的MODEL.RESNETS.OUT_FEATURES
stages = []
out_stage_idx = [{"res2": 2, "res3": 3, "res4": 4, "res5": 5}[f] for f in out_features]

#########题外延伸#############
resnet的配置不一定要配置为4个阶段，可根据需求进行适当修改。任意举例如：
num_blocks_per_stage={50,[2,3,4,3,4]}  ##5个阶段，意味着可以有5个特征层的输出
out_stage_idx=[{"res2": 2, "res3": 3, "res4": 4, "res5": 5，“res6”:6}[f] for f in out_features]

MODEL:
  ANCHOR_GENERATOR:
    SIZES: [[32], [64], [128], [256], [512]]  # One size for each in feature map
    ASPECT_RATIOS: [[0.5, 1.0, 2.0],[0.5, 1.0, 2.0,0.33,3.0],[0.5, 1.0, 2.0,0.33,3.0],[0.5, 1.0, 2.0],[0.5, 1.0, 2.0]]
    ##此处的aspect_ratio为实际运算时需要的比例值。与caffe里面实现的不同。
    ##caffe里面则是外部设置aspect_ratio=2,3，flip=True，实际内部计算时的值为：1,1,2,3,1/2,1/3.具体计算可自行百度
    ##个人认为caffe的这种实现方式比较方便。可修改相应源代码实现该操作。

##参数设置
MODEL:
  BACKBONE:
    NAME: "build_resnet_fpn_backbone"    
  RESNETS:
    OUT_FEATURES: ["res2", "res3", "res4", "res5"]
  FPN:
    IN_FEATURES: ["res2", "res3", "res4", "res5"]

# Apply Deformable Convolution in stages
# Specify if apply deform_conv on Res2, Res3, Res4, Res5
_C.MODEL.RESNETS.DEFORM_ON_PER_STAGE = [False, False, False, False]
# Use True to use modulated deform_conv (DeformableV2, https://arxiv.org/abs/1811.11168);
# Use False for DeformableV1.
_C.MODEL.RESNETS.DEFORM_MODULATED = False
# Number of groups in deformable conv.
_C.MODEL.RESNETS.DEFORM_NUM_GROUPS = 1

class FPN(Backbone):
    """
    This module implements Feature Pyramid Network.
    It creates pyramid features built on top of some input feature maps.
    """
    返回的结果dict[str->Tensor]
    与输入特征层in_features名称对应,输出的名称：['p2','p3','p4','p5','p6']
    其中最底层的特征输出'p6'，该层是通过对'p5'层进行下采样操作(max_pool(stride=2,kernel=1,padding=0))得到的。

    def __init__(
        self, bottom_up, in_features, out_channels, norm="", top_block=None, fuse_type="sum"
    ):
        """
        Args:
            bottom_up (Backbone): 输入的基网络，
            in_features (list[str]): 进行特征融合的层['res2','res3','res4','res5']]
            out_channels (int): number of channels in the output feature maps.
            norm (str): the normalization to use.
            top_block (nn.Module or None): 最后输出的层
            fuse_type (str): 特征融合时的计算方式，'sum','avg'
            #####其中，'avg':上采样后的feature map与当前融合的feature map 进行element_wise求平均操作((value1+value2/2))
                      'sum':上采样后的feature map与当前融合的feature map 进行element_wise求和操作
        """

top_down_features = F.interpolate(prev_features, scale_factor=2, mode="nearest")

######

def interpolate(input, size=None, scale_factor=None, mode='nearest', align_corners=None):
    
    The modes available for resizing are: `nearest`, `linear` (3D-only),
    `bilinear` (4D-only), `trilinear` (5D-only), `area`

    Args:
        input (Tensor): the input tensor，4D(N*C*H*W),5D(N*C*D*H*W)
        size (int or Tuple[int] or Tuple[int, int] or Tuple[int, int, int]):
            output spatial size.
        scale_factor (float or Tuple[float]): multiplier for spatial size. Has to match input size if it is a tuple.
        mode (string): algorithm used for upsampling:
            'nearest' | 'linear' | 'bilinear' | 'trilinear' | 'area'. Default: 'nearest'
        align_corners (bool, optional): if True, the corner pixels of the input
            and output tensors are aligned, and thus preserving the values at
            those pixels. This only has effect when :attr:`mode` is `linear`,
            `bilinear`, or `trilinear`. Default: False

    .. warning::
        With ``align_corners = True``, the linearly interpolating modes
        (`linear`, `bilinear`, and `trilinear`) don't proportionally align the
        output and input pixels, and thus the output values can depend on the
        input size. This was the default behavior for these modes up to version
        0.3.1. Since then, the default behavior is ``align_corners = False``.
        See :class:`~torch.nn.Upsample` for concrete examples on how this
        affects the outputs.

    .. include:: cuda_deterministic_backward.rst
    """