CBNU
/
18_WDI_CNN
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases 4 Wiki Activity

업데이트 'readme.md'

Browse Source
main
pinb 2 years ago
parent 5a4956063d
commit d8ccdc87a7
1 changed files with 120 additions and 41 deletions
Show Stats Download Patch File Download Diff File

161
readme.md
Unescape Escape View File

@ -1,21 +1,34 @@
# 지능화 캡스톤 프로젝트 #1 - WDI-CNN # 지능화 캡스톤 프로젝트 #1 - WDI-CNN
### *Wafer Map 데이터를 9종류의 Class로 분류하는 CNN 모델 만들기* ### *Wafer Map 데이터를 9종류의 Class로 분류하는 CNN 모델 만들기*
-----
[PINBlog Gitea Repository](https://gitea.pinblog.codes/CBNU/03_WDI_CNN)
----- -----
### 논문 ### 논문
반도체 제조공정의 불균형 데이터셋에 대한 웨이퍼 불량 식별을 위한 심층 컨볼루션 신경망 반도체 제조공정의 불균형 데이터셋에 대한 웨이퍼 불량 식별을 위한 심층 컨볼루션 신경망
* 번역본 * [번역본](https://gitea.pinblog.codes/attachments/9b2424f7-7e7d-4ad1-a368-86a523d67504)
https://gitea.pinblog.codes/attachments/9b2424f7-7e7d-4ad1-a368-86a523d67504
* [원본](https://gitea.pinblog.codes/attachments/9a31bb80-bc0a-4d5a-83b1-4ef0557456ad)
* 원본 * 인용된 논문 리스트
https://gitea.pinblog.codes/attachments/9a31bb80-bc0a-4d5a-83b1-4ef0557456ad
| 번호 | 논문 제목 | 저자 | 출판사 및 링크 |
|------|-----------|------|----------------|
| 1 | Deep Residual Learning for Image Recognition | Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun | IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. [링크](https://arxiv.org/abs/1512.03385) |
| 2 | Very Deep Convolutional Networks for Large-Scale Image Recognition | Karen Simonyan, Andrew Zisserman | International Conference on Learning Representations (ICLR), 2015. [링크](https://arxiv.org/abs/1409.1556) |
| 3 | Going Deeper with Convolutions | Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, Andrew Rabinovich | IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015. [링크](https://arxiv.org/abs/1409.4842) |
| 4 | Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift | Sergey Ioffe, Christian Szegedy | International Conference on Machine Learning (ICML), 2015. [링크](https://arxiv.org/abs/1502.03167) |
| 5 | Dropout: A Simple Way to Prevent Neural Networks from Overfitting | Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, Ruslan Salakhutdinov | Journal of Machine Learning Research (JMLR), 2014. [링크](http://jmlr.org/papers/volume15/srivastava14a/srivastava14a.pdf) |
----- -----
### Dataset ### Dataset
[Kaggle - WDI Data](https://www.kaggle.com/qingyi/wm811k-wafer-map/code) [Kaggle - WDI Data](https://www.kaggle.com/qingyi/wm811k-wafer-map/code)
@ -25,19 +38,33 @@ https://gitea.pinblog.codes/attachments/9a31bb80-bc0a-4d5a-83b1-4ef0557456ad
----- -----
### 수행방법 ### 수행방법
* 위 논문을 참고하여 CNN 모델을 구현하고, * 위 논문을 참고하여 CNN 모델을 구현하고,
WDI Dataset을 학습하여 9개의 클래스로 분류한다. WDI Dataset을 학습하여 9개의 클래스로 분류한다.
(Center, Donut, Edge-Loc, Edge-Ring, Loc, Near-full, none, Random, Scratch) | 클래스 | 라벨 | Train 이미지 개수 | Validation 이미지 개수 | Test 이미지 개수 |
|--------|------|-------------------|------------------------|------------------|
https://gitea.pinblog.codes/CBNU/03_WDI_CNN/releases/tag/info | None | 0 | 117,431 | 15,000 | 15,000 |
| Center | 1 | 3,294 | 500 | 500 |
| Donut | 2 | 444 | 50 | 50 |
| Edge-Loc | 3 | 4,189 | 500 | 500 |
| Edge-Ring |4 |7,680 |1,000 |1,000 |
| Local |5 |2,794 |400 |400 |
| Random |6 |666 |100 |100 |
| Scratch |7 |894 |150 |150 |
| Near-full |8 |149 |- |- |
[프로젝트 관련 자료](https://gitea.pinblog.codes/CBNU/03_WDI_CNN/releases/tag/info)
# Model # Model
<details>
```python <summary>Code View</summary>
<div markdown="1">
````python
import torch import torch
import torch.nn as nn import torch.nn as nn
import torch.nn.functional as F import torch.nn.functional as F
@ -99,12 +126,17 @@ class CNN_WDI(nn.Module):
return F.softmax(x, dim=1) return F.softmax(x, dim=1)
cnn_wdi = CNN_WDI(class_num=9) cnn_wdi = CNN_WDI(class_num=9)
````
``` </div>
</details>
# Load Data # Load Data
<details>
```python <summary>Code View</summary>
<div markdown="1">
````python
from torchvision import transforms, datasets from torchvision import transforms, datasets
# 데이터 전처리 # 데이터 전처리
@ -124,11 +156,17 @@ data_transforms = transforms.Compose([
train_dataset = datasets.ImageFolder(root='E:/wm_images/train/', transform=data_transforms) train_dataset = datasets.ImageFolder(root='E:/wm_images/train/', transform=data_transforms)
val_dataset = datasets.ImageFolder(root='E:/wm_images/val/', transform=data_transforms) val_dataset = datasets.ImageFolder(root='E:/wm_images/val/', transform=data_transforms)
test_dataset = datasets.ImageFolder(root='E:/wm_images/test/', transform=data_transforms) test_dataset = datasets.ImageFolder(root='E:/wm_images/test/', transform=data_transforms)
``` ````
# Settings </div>
</details>
```python # Settings
<details>
<summary>Code View</summary>
<div markdown="1">
````python
import torch.optim as optim import torch.optim as optim
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
@ -149,12 +187,17 @@ num_epochs = 100 #* 192
# Random sample size # Random sample size
train_max_images = 95 train_max_images = 95
val_max_images = 25 val_max_images = 25
````
``` </div>
</details>
# Train Function # Train Function
<details>
```python <summary>Code View</summary>
<div markdown="1">
````python
# 학습 함수 정의 # 학습 함수 정의
def train(model, dataloader, criterion, optimizer, device): def train(model, dataloader, criterion, optimizer, device):
model.train() model.train()
@ -181,11 +224,17 @@ def train(model, dataloader, criterion, optimizer, device):
epoch_acc = running_corrects.double() / len(dataloader.dataset) epoch_acc = running_corrects.double() / len(dataloader.dataset)
return epoch_loss, epoch_acc return epoch_loss, epoch_acc
``` ````
# Evaluate Function </div>
</details>
```python # Evaluate Function
<details>
<summary>Code View</summary>
<div markdown="1">
````python
# 평가 함수 정의 # 평가 함수 정의
def evaluate(model, dataloader, criterion, device): def evaluate(model, dataloader, criterion, device):
model.eval() model.eval()
@ -208,12 +257,17 @@ def evaluate(model, dataloader, criterion, device):
epoch_acc = running_corrects.double() / len(dataloader.dataset) epoch_acc = running_corrects.double() / len(dataloader.dataset)
return epoch_loss, epoch_acc return epoch_loss, epoch_acc
``` ````
# Train </div>
</details>
# Train
```python <details>
<summary>Code View</summary>
<div markdown="1">
````python
# Train & Validation의 Loss, Acc 기록 파일 # Train & Validation의 Loss, Acc 기록 파일
s_title = 'Epoch,\tTrain Loss,\tTrain Acc,\tVal Loss,\tVal Acc\n' s_title = 'Epoch,\tTrain Loss,\tTrain Acc,\tVal Loss,\tVal Acc\n'
with open('output.txt', 'a') as file: with open('output.txt', 'a') as file:
@ -243,19 +297,24 @@ for epoch in range(num_epochs + 1):
if epoch % 10 == 0: if epoch % 10 == 0:
# 모델 저장 # 모델 저장
torch.save(cnn_wdi.state_dict(), 'CNN_WDI_' + str(epoch) + 'epoch.pth') torch.save(cnn_wdi.state_dict(), 'CNN_WDI_' + str(epoch) + 'epoch.pth')
``` ````
</div>
</details>
----- -----
### 평가방법 ### 평가방법
* 모델의 성능지표(Precision, Recall, Accuracy, F1-Score)를 혼동행렬(Confusion Metrix)로 구현한다. * 모델의 성능지표(Precision, Recall, Accuracy, F1-Score)를 혼동행렬(Confusion Metrix)로 구현한다.
# Confusion Metrix # Confusion Metrix
<details>
```python <summary>Code View</summary>
<div markdown="1">
````python
import numpy as np import numpy as np
import matplotlib.pyplot as plt import matplotlib.pyplot as plt
import seaborn as sns import seaborn as sns
@ -335,12 +394,17 @@ def predict_and_plot_metrics(title, model, dataloader, criterion, device):
plot_metrics(title, class_names, precisions, recalls, f1_scores, epoch_acc.item()) plot_metrics(title, class_names, precisions, recalls, f1_scores, epoch_acc.item())
return epoch_loss, epoch_acc, report return epoch_loss, epoch_acc, report
````
``` </div>
</details>
# Evaluate # Evaluate
<details>
```python <summary>Code View</summary>
<div markdown="1">
````python
import os import os
import re import re
@ -363,11 +427,17 @@ for model in sorted_models:
# Call the predict_and_plot_metrics function with the appropriate arguments # Call the predict_and_plot_metrics function with the appropriate arguments
epoch_loss, epoch_acc, report = predict_and_plot_metrics(model, cnn_wdi, test_loader, criterion, device) epoch_loss, epoch_acc, report = predict_and_plot_metrics(model, cnn_wdi, test_loader, criterion, device)
# print(f'Model: {model} Test Loss: {test_loss:.4f} Acc: {test_acc:.4f}') # print(f'Model: {model} Test Loss: {test_loss:.4f} Acc: {test_acc:.4f}')
``` ````
# Loss Graph </div>
</details>
```python # Loss Graph
<details>
<summary>Code View</summary>
<div markdown="1">
````python
import matplotlib.pyplot as plt import matplotlib.pyplot as plt
# 파일에서 데이터를 읽어들입니다. # 파일에서 데이터를 읽어들입니다.
@ -405,12 +475,17 @@ plt.ylabel('Values')
plt.title('Training and Validation Loss and Accuracy') plt.title('Training and Validation Loss and Accuracy')
plt.legend() plt.legend()
plt.show() plt.show()
````
``` </div>
</details>
# Print Selecting Test Model Result # Print Selecting Test Model Result
<details>
```python <summary>Code View</summary>
<div markdown="1">
````python
def output(model, dataloader, criterion, device): def output(model, dataloader, criterion, device):
model.eval() model.eval()
running_loss = 0.0 running_loss = 0.0
@ -460,10 +535,14 @@ def output(model, dataloader, criterion, device):
selected_model = 'CNN_WDI_20epoch.pth' selected_model = 'CNN_WDI_20epoch.pth'
cnn_wdi.load_state_dict(torch.load(selected_model)) cnn_wdi.load_state_dict(torch.load(selected_model))
output(cnn_wdi, test_loader, criterion, device) output(cnn_wdi, test_loader, criterion, device)
``` ````
</div>
</details>
----- -----
### 테스트 결과 ### 테스트 결과
[1차 테스트](https://gitea.pinblog.codes/CBNU/03_WDI_CNN/wiki/1%EC%B0%A8-%ED%85%8C%EC%8A%A4%ED%8A%B8_%EC%9B%90%EB%B3%B8-%EB%8D%B0%EC%9D%B4%ED%84%B0-%ED%95%99%EC%8A%B5) [1차 테스트](https://gitea.pinblog.codes/CBNU/03_WDI_CNN/wiki/1%EC%B0%A8-%ED%85%8C%EC%8A%A4%ED%8A%B8_%EC%9B%90%EB%B3%B8-%EB%8D%B0%EC%9D%B4%ED%84%B0-%ED%95%99%EC%8A%B5)

Write Preview
Loading…
Cancel
Save