[pytorch] ⛓ HRNet : Powerful Pose Estimator & Object Detector parallel structure

😎 오늘은 Pose Estimation / Object Detection / Sementic Segmentaion에 Task에 강력한 HRNet의 Code를 분석해보도록 하겠습니다.

https://arxiv.org/abs/1908.07919
https://github.com/leoxiaobin/deep-high-resolution-net.pytorch
https://github.com/HRNet/HRNet-Object-Detection/tree/master/mmdet/models (mmdet)

🙄 Network의 구조를 간단히 설명하면,
1. Conv Block을 쌓아올리고,
2. 이를 Transition으로 분기처리(병렬화)하고,
3. Conv Block을 쌓고,
4. Fuse로 다른 Resolution끼리 섞은 후 다시 분기처리(병렬화)하고,
5. Conv Block을 쌓고,
6. Fuse로 다른 Resolution끼리 섞은 후 다시 분기처리(병렬화)하고,
7. Conv Block을 쌓고,
8. Fuse로 다른 Resolution끼리 섞은 후 다시 분기처리(병렬화)합니다.

아래 코드에서 중요히 볼 대목은 HighResolutionModule / _make_transition_layer 입니다.

# ------------------------------------------------------------------------------
# Copyright (c) Microsoft
# Licensed under the MIT License.
# Written by Bin Xiao (Bin.Xiao@microsoft.com)
# ------------------------------------------------------------------------------

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import os
import logging

import torch
import torch.nn as nn


BN_MOMENTUM = 0.1
logger = logging.getLogger(__name__)


def conv3x3(in_planes, out_planes, stride=1):
    """3x3 convolution with padding"""
    return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
                     padding=1, bias=False)

# residual block
class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(BasicBlock, self).__init__()
        self.conv1 = conv3x3(inplanes, planes, stride)
        self.bn1 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = conv3x3(planes, planes)
        self.bn2 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        residual = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)

        if self.downsample is not None:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)

        return out

# normal한 residual bottleneck block
class Bottleneck(nn.Module):
    expansion = 4

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(Bottleneck, self).__init__()
        self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride,
                               padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM)
        self.conv3 = nn.Conv2d(planes, planes * self.expansion, kernel_size=1,
                               bias=False)
        self.bn3 = nn.BatchNorm2d(planes * self.expansion,
                                  momentum=BN_MOMENTUM)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        residual = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        out = self.relu(out)

        out = self.conv3(out)
        out = self.bn3(out)

        if self.downsample is not None:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)

        return out

HighResolutionModule

class HighResolutionModule(nn.Module):
    def __init__(self, num_branches, blocks, num_blocks, num_inchannels,
                 num_channels, fuse_method, multi_scale_output=True):
        super(HighResolutionModule, self).__init__()
        self._check_branches(
            num_branches, blocks, num_blocks, num_inchannels, num_channels)

        self.num_inchannels = num_inchannels
        self.fuse_method = fuse_method
        self.num_branches = num_branches

        self.multi_scale_output = multi_scale_output

        self.branches = self._make_branches(
            num_branches, blocks, num_blocks, num_channels)
        self.fuse_layers = self._make_fuse_layers()
        self.relu = nn.ReLU(True)

    def _check_branches(self, num_branches, blocks, num_blocks,
                        num_inchannels, num_channels):
        if num_branches != len(num_blocks):
            error_msg = 'NUM_BRANCHES({}) <> NUM_BLOCKS({})'.format(
                num_branches, len(num_blocks))
            logger.error(error_msg)
            raise ValueError(error_msg)

        if num_branches != len(num_channels):
            error_msg = 'NUM_BRANCHES({}) <> NUM_CHANNELS({})'.format(
                num_branches, len(num_channels))
            logger.error(error_msg)
            raise ValueError(error_msg)

        if num_branches != len(num_inchannels):
            error_msg = 'NUM_BRANCHES({}) <> NUM_INCHANNELS({})'.format(
                num_branches, len(num_inchannels))
            logger.error(error_msg)
            raise ValueError(error_msg)

    def _make_one_branch(self, branch_index, block, num_blocks, num_channels,
                         stride=1):
        downsample = None
        if stride != 1 or \
           self.num_inchannels[branch_index] != num_channels[branch_index] * block.expansion:
            downsample = nn.Sequential(
                nn.Conv2d(
                    self.num_inchannels[branch_index],
                    num_channels[branch_index] * block.expansion,
                    kernel_size=1, stride=stride, bias=False
                ),
                nn.BatchNorm2d(
                    num_channels[branch_index] * block.expansion,
                    momentum=BN_MOMENTUM
                ),
            )

        layers = []
        layers.append(
            block(
                self.num_inchannels[branch_index],
                num_channels[branch_index],
                stride,
                downsample
            )
        )
        # 여기가 새로운 in channel 갱신 부분
        self.num_inchannels[branch_index] = \
            num_channels[branch_index] * block.expansion
        for i in range(1, num_blocks[branch_index]):
            layers.append(
                block(
                    self.num_inchannels[branch_index],
                    num_channels[branch_index]
                )
            )

        return nn.Sequential(*layers)

    def _make_branches(self, num_branches, block, num_blocks, num_channels):
        branches = []

        for i in range(num_branches):
            branches.append(
                self._make_one_branch(i, block, num_blocks, num_channels)
            )

        return nn.ModuleList(branches)

    # 이게 핵심 : 섞섞
    def _make_fuse_layers(self):
        if self.num_branches == 1:
            return None

        num_branches = self.num_branches
        num_inchannels = self.num_inchannels
        fuse_layers = []
        for i in range(num_branches if self.multi_scale_output else 1):
            fuse_layer = []
            for j in range(num_branches):
                # j는 주는 쪽, i는 받는 쪽
                # 큰 resolution에서 작은 reolution으로 1x1 conv + upsampling
                if j > i:
                    fuse_layer.append(
                        nn.Sequential(
                            nn.Conv2d(
                                num_inchannels[j],
                                num_inchannels[i],
                                1, 1, 0, bias=False
                            ),
                            nn.BatchNorm2d(num_inchannels[i]),
                            nn.Upsample(scale_factor=2**(j-i), mode='nearest')
                        )
                    )
                # same resolution
                elif j == i:
                    fuse_layer.append(None)
                # 작은 resolution에서 큰 resolution으로 3x3 conv (downsampling 포함)
                else:
                    conv3x3s = []
                    for k in range(i-j):
                        if k == i - j - 1:
                            num_outchannels_conv3x3 = num_inchannels[i]
                            conv3x3s.append(
                                nn.Sequential(
                                    nn.Conv2d(
                                        num_inchannels[j],
                                        num_outchannels_conv3x3,
                                        3, 2, 1, bias=False
                                    ),
                                    nn.BatchNorm2d(num_outchannels_conv3x3)
                                )
                            )
                        else:
                            num_outchannels_conv3x3 = num_inchannels[j]
                            conv3x3s.append(
                                nn.Sequential(
                                    nn.Conv2d(
                                        num_inchannels[j],
                                        num_outchannels_conv3x3,
                                        3, 2, 1, bias=False
                                    ),
                                    nn.BatchNorm2d(num_outchannels_conv3x3),
                                    nn.ReLU(True)
                                )
                            )
                    fuse_layer.append(nn.Sequential(*conv3x3s))
            fuse_layers.append(nn.ModuleList(fuse_layer))

        return nn.ModuleList(fuse_layers)

    def get_num_inchannels(self):
        return self.num_inchannels

    def forward(self, x):
        # 원래 resolution
        if self.num_branches == 1:
            return [self.branches[0](x[0])]
        # transition 있는 경우부터
        for i in range(self.num_branches):
            x[i] = self.branches[i](x[i])

        x_fuse = []

        for i in range(len(self.fuse_layers)):
            # 같은 resolution이면 conv 안하고 그냥 이동
            y = x[0] if i == 0 else self.fuse_layers[i][0](x[0])
            for j in range(1, self.num_branches):
                if i == j:
                    y = y + x[j] # 걍 더해주기
                else:
                    y = y + self.fuse_layers[i][j](x[j]) # forwrd 치고 더해주기
            x_fuse.append(self.relu(y))
        
        # 이렇게 되면 결국 fused된 layer가 병렬로 들어가게 됨
        return x_fuse


blocks_dict = {
    'BASIC': BasicBlock,
    'BOTTLENECK': Bottleneck
}

_make_transition_layer in PoseHighResolutionNet

# 여기가 본체!
class PoseHighResolutionNet(nn.Module):
    # 설정 start
    def __init__(self, cfg='', **kwargs):
        self.inplanes = 64
        #extra = cfg['MODEL']['EXTRA']
        super(PoseHighResolutionNet, self).__init__()

        # stem net
        # 3 2 1 => H W 절반
        self.conv1 = nn.Conv2d(3, 64, kernel_size=3, stride=2, padding=1,
                               bias=False)
        self.bn1 = nn.BatchNorm2d(64, momentum=BN_MOMENTUM)
        # 3 2 1 => H W 절반
        self.conv2 = nn.Conv2d(64, 64, kernel_size=3, stride=2, padding=1,
                               bias=False)
        self.bn2 = nn.BatchNorm2d(64, momentum=BN_MOMENTUM)
        self.relu = nn.ReLU(inplace=True)
        # bottleneck?
        self.layer1 = self._make_layer(Bottleneck, 64, 4)

        #self.stage2_cfg = extra['STAGE2'] #woong
        self.stage2_cfg={'NUM_CHANNELS':[32,64],'BLOCK':'BASIC', 'NUM_MODULES':1,'NUM_BRANCHES':2, 
        'NUM_BLOCKS':[4,4], 'NUM_CHANNELS':[32,64],'FUSE_METHOD':'SUM'  }
        num_channels = self.stage2_cfg['NUM_CHANNELS']
        #print('num_channels:', num_channels)
        #return
        block = blocks_dict[self.stage2_cfg['BLOCK']]
        #print('##block:',BasicBlock)
        #return
        num_channels = [
            num_channels[i] * block.expansion for i in range(len(num_channels))
        ]
        self.transition1 = self._make_transition_layer([256], num_channels)
        self.stage2, pre_stage_channels = self._make_stage(
            self.stage2_cfg, num_channels)

        #self.stage3_cfg = extra['STAGE3']
        self.stage3_cfg={'NUM_CHANNELS':[32,64,128],'BLOCK':'BASIC', 'NUM_MODULES':4,'NUM_BRANCHES':3, 
        'NUM_BLOCKS':[4,4,4], 'NUM_CHANNELS':[32,64,128],'FUSE_METHOD':'SUM'  }
        num_channels = self.stage3_cfg['NUM_CHANNELS']
        block = blocks_dict[self.stage3_cfg['BLOCK']]
        num_channels = [
            num_channels[i] * block.expansion for i in range(len(num_channels))
        ]
        self.transition2 = self._make_transition_layer(
            pre_stage_channels, num_channels)
        self.stage3, pre_stage_channels = self._make_stage(
            self.stage3_cfg, num_channels)

        #self.stage4_cfg = extra['STAGE4']
        self.stage4_cfg={'NUM_CHANNELS':[32,64,128,256],'BLOCK':'BASIC', 'NUM_MODULES':3,'NUM_BRANCHES':4,
         'NUM_BLOCKS':[4,4,4,4], 'NUM_CHANNELS':[32,64,128,256],'FUSE_METHOD':'SUM'  }
        num_channels = self.stage4_cfg['NUM_CHANNELS']
        block = blocks_dict[self.stage4_cfg['BLOCK']]
        num_channels = [
            num_channels[i] * block.expansion for i in range(len(num_channels))
        ]
        self.transition3 = self._make_transition_layer(
            pre_stage_channels, num_channels)
        self.stage4, pre_stage_channels = self._make_stage(
            self.stage4_cfg, num_channels, multi_scale_output=False)
        
        extra={'FINAL_CONV_KERNEL':1}
        self.final_layer = nn.Conv2d(
            in_channels=pre_stage_channels[0],
            out_channels=16, #woong #cfg['MODEL']['NUM_JOINTS'],
            kernel_size= extra['FINAL_CONV_KERNEL'],
            stride=1,
            padding=1 if extra['FINAL_CONV_KERNEL'] == 3 else 0
        )

        self.pretrained_layers = ['conv1', 'bn1', 'conv2', 'bn2', 'layer1', 'transition1', 'stage2', 
        'transition2', 'stage3', 'transition3', 'stage4']#woong # extra['PRETRAINED_LAYERS']

    # transition?
    def _make_transition_layer(
            self, num_channels_pre_layer, num_channels_cur_layer):
        num_branches_cur = len(num_channels_cur_layer)
        num_branches_pre = len(num_channels_pre_layer)

        transition_layers = []
        # layer1 - transition1 사이에는 fuse가 없었음 : 그를 위한 else의 conv3x3 downsampling
        for i in range(num_branches_cur):
            if i < num_branches_pre:
                if num_channels_cur_layer[i] != num_channels_pre_layer[i]:
                    transition_layers.append(
                        nn.Sequential(
                            nn.Conv2d(
                                num_channels_pre_layer[i],
                                num_channels_cur_layer[i],
                                3, 1, 1, bias=False
                            ),
                            nn.BatchNorm2d(num_channels_cur_layer[i]),
                            nn.ReLU(inplace=True)
                        )
                    )
                else:
                    transition_layers.append(None)
            else:
                conv3x3s = []
                for j in range(i+1-num_branches_pre):
                    inchannels = num_channels_pre_layer[-1]
                    outchannels = num_channels_cur_layer[i] \
                        if j == i-num_branches_pre else inchannels
                    conv3x3s.append(
                        nn.Sequential(
                            nn.Conv2d(
                                inchannels, outchannels, 3, 2, 1, bias=False
                            ),
                            nn.BatchNorm2d(outchannels),
                            nn.ReLU(inplace=True)
                        )
                    )
                transition_layers.append(nn.Sequential(*conv3x3s))

        return nn.ModuleList(transition_layers)
    # layer 생성
    def _make_layer(self, block, planes, blocks, stride=1):
        downsample = None
        if stride != 1 or self.inplanes != planes * block.expansion:
            # for residual conn.
            downsample = nn.Sequential(
                nn.Conv2d(
                    self.inplanes, planes * block.expansion,
                    kernel_size=1, stride=stride, bias=False
                ),
                nn.BatchNorm2d(planes * block.expansion, momentum=BN_MOMENTUM),
            )

        layers = []
        layers.append(block(self.inplanes, planes, stride, downsample))
        self.inplanes = planes * block.expansion
        for i in range(1, blocks):
            layers.append(block(self.inplanes, planes))

        return nn.Sequential(*layers)

    # stage
    def _make_stage(self, layer_config, num_inchannels,
                    multi_scale_output=True):
        num_modules = layer_config['NUM_MODULES']
        num_branches = layer_config['NUM_BRANCHES']
        num_blocks = layer_config['NUM_BLOCKS']
        num_channels = layer_config['NUM_CHANNELS']
        block = blocks_dict[layer_config['BLOCK']]
        fuse_method = layer_config['FUSE_METHOD']

        modules = []
        for i in range(num_modules):
            # multi_scale_output is only used last module
            # 맨 마지막 transition 수행 여부?
            if not multi_scale_output and i == num_modules - 1:
                reset_multi_scale_output = False
            else:
                reset_multi_scale_output = True

            modules.append(
                HighResolutionModule(
                    num_branches,
                    block,
                    num_blocks,
                    num_inchannels,
                    num_channels,
                    fuse_method,
                    reset_multi_scale_output
                )
            )
            num_inchannels = modules[-1].get_num_inchannels()

        return nn.Sequential(*modules), num_inchannels

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.conv2(x)
        x = self.bn2(x)
        x = self.relu(x)
        x = self.layer1(x)

        x_list = []
        for i in range(self.stage2_cfg['NUM_BRANCHES']):
            if self.transition1[i] is not None:
                x_list.append(self.transition1[i](x))
            else:
                x_list.append(x)
        y_list = self.stage2(x_list)

        x_list = []
        for i in range(self.stage3_cfg['NUM_BRANCHES']):
            if self.transition2[i] is not None:
                x_list.append(self.transition2[i](y_list[-1]))
            else:
                x_list.append(y_list[i])
        y_list = self.stage3(x_list)

        x_list = []
        for i in range(self.stage4_cfg['NUM_BRANCHES']):
            if self.transition3[i] is not None:
                x_list.append(self.transition3[i](y_list[-1]))
            else:
                x_list.append(y_list[i])
        y_list = self.stage4(x_list)

        x = self.final_layer(y_list[0])

        return x

    def init_weights(self, pretrained=''):
        logger.info('=> init weights from normal distribution')
        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                # nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
                nn.init.normal_(m.weight, std=0.001)
                for name, _ in m.named_parameters():
                    if name in ['bias']:
                        nn.init.constant_(m.bias, 0)
            elif isinstance(m, nn.BatchNorm2d):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)
            elif isinstance(m, nn.ConvTranspose2d):
                nn.init.normal_(m.weight, std=0.001)
                for name, _ in m.named_parameters():
                    if name in ['bias']:
                        nn.init.constant_(m.bias, 0)

        if os.path.isfile(pretrained):
            pretrained_state_dict = torch.load(pretrained)
            logger.info('=> loading pretrained model {}'.format(pretrained))

            need_init_state_dict = {}
            for name, m in pretrained_state_dict.items():
                if name.split('.')[0] in self.pretrained_layers \
                   or self.pretrained_layers[0] is '*':
                    need_init_state_dict[name] = m
            self.load_state_dict(need_init_state_dict, strict=False)
        elif pretrained:
            logger.error('=> please download pre-trained models first!')
            raise ValueError('{} is not exist!'.format(pretrained))


def get_pose_net(cfg, is_train, **kwargs):
    model = PoseHighResolutionNet(cfg, **kwargs)

    if is_train and cfg['MODEL']['INIT_WEIGHTS']:
        model.init_weights(cfg['MODEL']['PRETRAINED'])
    print(model)  #woong
    return model