본문 바로가기
📂 프로젝트/◾ PAPERS

[ResNet] 논문 리뷰 및 구현 (코드 설명 포함)

by 이 정규 2023. 2. 8.
728x90
반응형

ResNet

Description

  • ResNet의 특징:
  • ResNet Architecture:

  • ResNet - Residual Block
    1. Plain network: VGGNet based
      • Convolution layer: 3x3, stried = 2
      • Global Average Pooling applied
      • FC layer(1000), softmax
    2. Residual network: Plain Network based
      • shortcut connection 이 추가
      • 차원이 같을 때 : 직접 값을 전달해 준다
      • 차원이 증가할 때 :
        1. zero padding으로 차원을 맞춰준다
        2. 1x1 Conv layer로 차원을 맞춰준다(VGG or GoogleNet)
      • stride = 2
  • ResNet Projection(Bottleneck Building block):
    • residual network 방식:
      1. Identity 방식
      2. 1x1 convolution 연산을 통해 차원을 스케일링 해주는 Projection 방식
  • Augmentation
    • Input image shape = 224 x 224 x 3
    • Mean subtraction of RGB per channel
  • Layer
    • Batch Normalization 적용 (between Conv and Activation function)
    • Dropout = None
    • ResNet18, ResNet34 layer = Basic Building Block 사용
    • ResNet50, ResNet101, ResNet 152 = Bottleneck Builiding Block 사용
    • Max Pooling = 1개 (kernel size = 3x3, stride = 2)
    • Global Average Pooling = 1개 (stride = 1)
    • residual network에서 차원이 증가하는 경우 :
      1. zerro - padding 으로 차원 증가에 대응: 추가적으로 parameters가 없다.
      2. 1x1 convolution으로 차원 스케일링 해준 뒤 다시 원래 차원으로 스케일링 진행
  • Hyper-parameter
    • Optimizer = SGD -> Adam으로 변경
    • Momentum = 0.9
    • Weight Decay = 0.0001
    • Batch size = 256
    • learning rate = learning rate scheduler 사용(0.1에서 시작해서 x10씩 줄여나감) -> 0.001로 설정
    • Epoch = not mentioned -> 10 적용
    • 60 × 104 iterations 만큼 학습 (논문)
    • Dropout = None
  • Fully Connected Layer(FC Layer):
    • 1개의 512 channel
  • Dataset:
    • 논문 : ImageNet Large Scale Visual Recognition Challenge(ILSVRC)-2012, CIFAR-10
    • 구현 : CIFAR-10
  • CIFAR-10 Result(paper):
    • ResNet은 기존의 plain network와는 다르게 layer가 깊어질수록 성능 향상이 나타남
    • layer가 너무 깊은 경우 error 가 증가
  • System Environment:
    • Google Colab Pro

Reference

Load Modules

# Utils
import numpy as np
from tqdm import tqdm 
import matplotlib.pyplot as plt

# Torch
import torch
import torch.nn as nn
from torch import Tensor
from typing import Optional
import torch.nn.functional as F
from torchsummary import summary
from torchvision import transforms
import torchvision

# TensorBoard
from torch.utils.tensorboard import SummaryWriter
! pip install tensorboardX
from tensorboardX import SummaryWriter

Set Hyperparameters

batch_size = 128
num_epochs = 10
learning_rate = 0.001

Load Data - CIFAR10

train_set = torchvision.datasets.CIFAR10(root='./cifar10', train=True, download=True, transform=transforms.ToTensor())
test_set = torchvision.datasets.CIFAR10(root='./cifar10', train=False, download=True, transform=transforms.ToTensor())

Mean subtraction of RGB per channel

train_meanRGB = [np.mean(x.numpy(), axis=(1,2)) for x, _ in train_set]
train_stdRGB = [np.std(x.numpy(), axis=(1,2)) for x, _ in train_set]

train_meanR = np.mean([m[0] for m in train_meanRGB])
train_meanG = np.mean([m[1] for m in train_meanRGB])
train_meanB = np.mean([m[2] for m in train_meanRGB])
train_stdR = np.mean([s[0] for s in train_stdRGB])
train_stdG = np.mean([s[1] for s in train_stdRGB])
train_stdB = np.mean([s[2] for s in train_stdRGB])

print(train_meanR, train_meanG, train_meanB)
print(train_stdR, train_stdG, train_stdB)

Define the image transformation for data

train_transformer = transforms.Compose([transforms.Resize((224,224)),
                                        transforms.ToTensor(),
                                        transforms.Normalize(mean=[train_meanR, train_meanG, train_meanB], std=[train_stdR, train_stdG, train_stdB])])

train_set.transform = train_transformer
test_set.transform = train_transformer

Define DataLoader

trainloader = torch.utils.data.DataLoader(train_set, batch_size=batch_size, shuffle=True, num_workers=2)
testloader = torch.utils.data.DataLoader(test_set, batch_size=batch_size, shuffle=True, num_workers=2)

Created ResNet Model

  • BottleNeck
class BottleNeck(nn.Module):
    
    expansion = 4
    def __init__(self, in_channels, out_channels, stride=1):
        super().__init__()
        self.residual_function = nn.Sequential(
            nn.Conv2d(in_channels, out_channels, kernel_size=1, bias=False),
            nn.BatchNorm2d(out_channels),
            nn.ReLU(inplace=True),
            nn.Conv2d(out_channels, out_channels, stride=stride, kernel_size=3, padding=1, bias=False),
            nn.BatchNorm2d(out_channels),
            nn.ReLU(inplace=True),
            nn.Conv2d(out_channels, out_channels * BottleNeck.expansion, kernel_size=1, bias=False),
            nn.BatchNorm2d(out_channels * BottleNeck.expansion),
        )

        self.shortcut = nn.Sequential()

        if stride != 1 or in_channels != out_channels * BottleNeck.expansion:
            self.shortcut = nn.Sequential(
                nn.Conv2d(in_channels, out_channels * BottleNeck.expansion, stride=stride, kernel_size=1, bias=False),
                nn.BatchNorm2d(out_channels * BottleNeck.expansion)
            )

    def forward(self, x):
        return nn.ReLU(inplace=True)(self.residual_function(x) + self.shortcut(x))
  • ResNet50
class ResNet(nn.Module):

    def __init__(self, block, num_classes=10):
        super().__init__()

        self.in_channels = 64

        self.conv1 = nn.Sequential(
            nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False),
            nn.BatchNorm2d(64),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
        )
    
        self.conv2_x = self._make_layer(block, 64, 3, 1)
        self.conv3_x = self._make_layer(block, 128, 4, 2)
        self.conv4_x = self._make_layer(block, 256, 6, 2)
        self.conv5_x = self._make_layer(block, 512, 3, 2)
        self.avg_pool = nn.AdaptiveAvgPool2d((1, 1))
        self.fc = nn.Linear(512 * block.expansion, num_classes)

    def _make_layer(self, block, out_channels, num_blocks, stride):
        strides = [stride] + [1] * (num_blocks - 1)
        layers = []
        for stride in strides:
            layers.append(block(self.in_channels, out_channels, stride))
            self.in_channels = out_channels * block.expansion

        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv1(x)
        x = self.conv2_x(x)
        x = self.conv3_x(x)
        x = self.conv4_x(x)
        x = self.conv5_x(x)
        x = self.avg_pool(x)
        x = x.view(x.size(0), -1)
        x = self.fc(x)

        return x

Set Device and Model

use_cuda = torch.cuda.is_available()
print("use_cuda : ", use_cuda)

FloatTensor = torch.cuda.FloatTensor if use_cuda else torch.FloatTensor
device = torch.device("cuda:0" if use_cuda else "cpu")

net = ResNet(BottleNeck).to(device)

X = torch.randn(size=(3,224,224)).type(FloatTensor)
print(summary(net, (3,224,224)))

Loss and Optimizer

use_cuda = torch.cuda.is_available()
print("use_cuda : ", use_cuda)
device = torch.device("cuda:0" if use_cuda else "cpu")
model = ResNet(BottleNeck).to(device)
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)

Training Loop

writer = SummaryWriter("./resnet/tensorboard") 

def train(model, device, train_loader, optimizer, epoch):
    model.train()
    for batch_idx, (data,target) in enumerate(train_loader):
        target = target.type(torch.LongTensor)
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        output = model(data)
        loss = criterion(output, target)
        loss.backward()
        optimizer.step()
        if (batch_idx + 1) % 30 == 0:
            print(f"{batch_idx*len(data)}/{len(train_loader.dataset)}")

def test(model, device, test_loader):
    model.eval()
    test_loss = 0
    correct = 0
    with torch.no_grad():
        for data, target in test_loader:
            data, target = data.to(device), target.to(device)
            output = model(data)
            test_loss += criterion(output, target).item()
            writer.add_scalar("Test Loss", test_loss, epoch)
            pred = output.argmax(1)
            correct += float((pred == target).sum())
            writer.add_scalar("Test Accuracy", correct, epoch)
            
        test_loss /= len(test_loader.dataset)
        correct /= len(test_loader.dataset)
        return test_loss, correct
        writer.close()

Per-Epoch Activity

for epoch in tqdm(range(1, num_epochs + 1)):
    train(model, device, trainloader, optimizer, epoch)
    test_loss, test_accuracy = test(model, device, testloader)
    writer.add_scalar("Test Loss", test_loss, epoch)
    writer.add_scalar("Test Accuracy", test_accuracy, epoch)
    print(f"Processing Result = Epoch : {epoch}   Loss : {test_loss}   Accuracy : {test_accuracy}")
    writer.close()

Result

print(f" Result of ResNet = Epoch : {epoch}   Loss : {test_loss}   Accuracy : {test_accuracy}")

Visualization

%load_ext tensorboard
%tensorboard --logdir=./resnet/tensorboard --port=6006

Visualize result of Accuracy and Loss by Tensorboard

728x90
반응형
저작자표시

'📂 프로젝트 > ◾ PAPERS' 카테고리의 다른 글

[GoogleNet] 논문 리뷰 및 구현 (코드 설명 포함)  (0) 2023.02.08
[AlexNet] 논문 리뷰 및 구현 (코드 설명 포함)  (2) 2023.02.08

관련글

댓글

티스토리툴바