분류 전체보기
-
이번에는 한국컴퓨터정보학회 2022 동계학술대회 논문집 제30권 1호에 발표된 논문인 Target Detection Method using Lightweight Mean Shift Segmentation and Shape Features를 읽고, 이전과 다르게 세세한 리뷰 보다는, 요약 및 논문에 대한 내 생각을 위주로 정리해보고자 합니다.Index1. Summary2. My Opinion1. Summary본 논문은 하드웨어 리소스가 제한된 장치에서 영상 분할 방법을 이용하여 small size object의detection을 하는 방법의 개선 방안에 대해 제안한다.영상 분할 방법은 region 기반 방법과, edge 기반 방법으로 나누어지며 대표적으로 FCM, Quickshift, Felzenszwal..
[리뷰] Target Detection Method using Lightweight Mean Shift Segmentation and Shape Features이번에는 한국컴퓨터정보학회 2022 동계학술대회 논문집 제30권 1호에 발표된 논문인 Target Detection Method using Lightweight Mean Shift Segmentation and Shape Features를 읽고, 이전과 다르게 세세한 리뷰 보다는, 요약 및 논문에 대한 내 생각을 위주로 정리해보고자 합니다.Index1. Summary2. My Opinion1. Summary본 논문은 하드웨어 리소스가 제한된 장치에서 영상 분할 방법을 이용하여 small size object의detection을 하는 방법의 개선 방안에 대해 제안한다.영상 분할 방법은 region 기반 방법과, edge 기반 방법으로 나누어지며 대표적으로 FCM, Quickshift, Felzenszwal..
2024.09.10 -
Import Library¶In [1]: import FinanceDataReader as fdrfrom sklearn.preprocessing import MinMaxScalerimport torchimport timeimport matplotlib.pyplot as plt%matplotlib inline Define LSTM model¶In [2]: class LSTM(torch.nn.Module) : def __init__(self, num_classes, input_size, hidden_size, num_layers, seq_length, device) : super(LSTM, self).__init__() self.num_classes = n..
LSTM을 이용한 주식 가격 예측Import Library¶In [1]: import FinanceDataReader as fdrfrom sklearn.preprocessing import MinMaxScalerimport torchimport timeimport matplotlib.pyplot as plt%matplotlib inline Define LSTM model¶In [2]: class LSTM(torch.nn.Module) : def __init__(self, num_classes, input_size, hidden_size, num_layers, seq_length, device) : super(LSTM, self).__init__() self.num_classes = n..
2024.09.10 -
Python with numpy¶ Variables and Literals¶= means assign.a=5 means assign the numeric literal 5 to the variable a.a="Five" means assign the string literal "Five" to the variable a. In [1]:a = 5print("a =", a)a = "Five"print("a =", a) a = 5a = Five Operators¶ In [2]:x = 14y = 4# Add two operandsprint('x + y =', x+y) # Output: x + y = 18# Subtract right operand from the leftprint('x - y =', x-y..
Numpy in PythonPython with numpy¶ Variables and Literals¶= means assign.a=5 means assign the numeric literal 5 to the variable a.a="Five" means assign the string literal "Five" to the variable a. In [1]:a = 5print("a =", a)a = "Five"print("a =", a) a = 5a = Five Operators¶ In [2]:x = 14y = 4# Add two operandsprint('x + y =', x+y) # Output: x + y = 18# Subtract right operand from the leftprint('x - y =', x-y..
2024.09.10 -
탐색적 데이터 분석과 데이터 시각화 실습¶In [1]: !pip install -q --upgrade matplotlib /bin/bash: pip: command not found들어가기¶‘통계학자’ 나이팅게일의 ‘로즈 다이어그램’In [2]: from packaging import versionimport matplotlib as mplimport matplotlib.pyplot as pltimport matplotlib_inline.backend_inlineimport numpy as npimport pandas as pdassert version.parse(mpl.__version__) >= version.Version("3.5"), ( "에러가 난다면 첫..
Exploratory data analysis and visualization탐색적 데이터 분석과 데이터 시각화 실습¶In [1]: !pip install -q --upgrade matplotlib /bin/bash: pip: command not found들어가기¶‘통계학자’ 나이팅게일의 ‘로즈 다이어그램’In [2]: from packaging import versionimport matplotlib as mplimport matplotlib.pyplot as pltimport matplotlib_inline.backend_inlineimport numpy as npimport pandas as pdassert version.parse(mpl.__version__) >= version.Version("3.5"), ( "에러가 난다면 첫..
2024.09.10 -
hr_data_binary_classification¶IBM HR data¶https://www.kaggle.com/datasets/pavansubhasht/ibm-hr-analytics-attrition-dataset구글 드라이브 연동¶In [ ]: from google.colab import drivedrive.mount('/content/drive') Mounted at /content/drive1. 문제 파악 및 목표 설정¶2. 데이터 수집 및 전처리¶In [ ]: import numpy as npimport pandas as pdimport matplotlib.pyplot as pltimport seaborn as snsimport missingno as ms..
IBM HR data Binary Classificationhr_data_binary_classification¶IBM HR data¶https://www.kaggle.com/datasets/pavansubhasht/ibm-hr-analytics-attrition-dataset구글 드라이브 연동¶In [ ]: from google.colab import drivedrive.mount('/content/drive') Mounted at /content/drive1. 문제 파악 및 목표 설정¶2. 데이터 수집 및 전처리¶In [ ]: import numpy as npimport pandas as pdimport matplotlib.pyplot as pltimport seaborn as snsimport missingno as ms..
2024.09.10 -
Implement CNN Model¶Import Library¶In [1]: import scipy.io as sioimport matplotlib.pyplot as pltimport tensorflow as tfimport numpy as np%matplotlib inline 2022-12-07 17:52:14.633420: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To tur..
Face RecognitionImplement CNN Model¶Import Library¶In [1]: import scipy.io as sioimport matplotlib.pyplot as pltimport tensorflow as tfimport numpy as np%matplotlib inline 2022-12-07 17:52:14.633420: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To tur..
2024.09.10 -
Import Library¶In [1]: import tensorflow as tfimport numpy as npimport matplotlib.pyplot as plt%matplotlib inline 2022-11-23 17:10:00.971077: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `..
Image PredictionImport Library¶In [1]: import tensorflow as tfimport numpy as npimport matplotlib.pyplot as plt%matplotlib inline 2022-11-23 17:10:00.971077: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `..
2024.09.10 -
In [1]: import numpy as npimport tensorflow as tfimport timenp.random.seed(101) 2022-10-12 17:13:46.461742: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.build you..
Linear Classifier 02In [1]: import numpy as npimport tensorflow as tfimport timenp.random.seed(101) 2022-10-12 17:13:46.461742: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.build you..
2024.09.10 -
In [1]: import numpy as npnp.random.seed(101) build your own data¶x_train: [12, 4] y_train: [12, ] x_test: [4, 4] y_test: [4, ][1, 0, 0, 0] --> 0 [0, 1, 0, 0] --> 0 [0, 0, 1, 0] --> 1 [0, 0, 0, 1] --> 1In [2]: x_train = np.zeros(shape=[12, 4], dtype=np.float32)y_train = np.random.randint(0, 4, [12, ])print(x_train)print(y_train) [[0. 0. 0. 0.] [0. 0. 0. 0.] [0. 0. 0. 0.] ..
Linear Classifier 01In [1]: import numpy as npnp.random.seed(101) build your own data¶x_train: [12, 4] y_train: [12, ] x_test: [4, 4] y_test: [4, ][1, 0, 0, 0] --> 0 [0, 1, 0, 0] --> 0 [0, 0, 1, 0] --> 1 [0, 0, 0, 1] --> 1In [2]: x_train = np.zeros(shape=[12, 4], dtype=np.float32)y_train = np.random.randint(0, 4, [12, ])print(x_train)print(y_train) [[0. 0. 0. 0.] [0. 0. 0. 0.] [0. 0. 0. 0.] ..
2024.09.10 -
CS231n Python Tutorial With Google Colab¶ This tutorial was originally written by Justin Johnson for cs231n. It was adapted as a Jupyter notebook for cs228 by Volodymyr Kuleshov and Isaac Caswell.This version has been adapted for Colab by Kevin Zakka for the Spring 2020 edition of cs231n. It runs Python3 by default. Introduction¶ Python is a great general-purpose programming language on its o..
CS231n Python TutorialCS231n Python Tutorial With Google Colab¶ This tutorial was originally written by Justin Johnson for cs231n. It was adapted as a Jupyter notebook for cs228 by Volodymyr Kuleshov and Isaac Caswell.This version has been adapted for Colab by Kevin Zakka for the Spring 2020 edition of cs231n. It runs Python3 by default. Introduction¶ Python is a great general-purpose programming language on its o..
2024.09.10 -
1. 개념어떠한 데이터 입력이 들어왔을 때, 해당 입력을 압축 시켜 latent vector로 만든 후, 해당 embedding vector를 다시 본래의 입력 데이터 형태로 복원 하는 신경망 2. 구조인코더와 디코더로 이루어짐인코더는 입력 데이터를 압축하여 latent vector로 만듬, recognition network 라고도 불림디코더는 latent vector를이용하여 원래의 입력 형태로 만듬, generative network 라고도 불림 3. 활용데이터에 대한 차원 축소, 이상 탐지, 노이즈 제거, 인식, 생성 등에 활용됨 4. VAE(; Variational AutoEncoder)기본적인 형태의 오토인코더와 다름확률적 오토인코더로, 학습이 끝난 이후에도 출력이 부분적으로 우연에 의해 결..
Auto Encoder 요약1. 개념어떠한 데이터 입력이 들어왔을 때, 해당 입력을 압축 시켜 latent vector로 만든 후, 해당 embedding vector를 다시 본래의 입력 데이터 형태로 복원 하는 신경망 2. 구조인코더와 디코더로 이루어짐인코더는 입력 데이터를 압축하여 latent vector로 만듬, recognition network 라고도 불림디코더는 latent vector를이용하여 원래의 입력 형태로 만듬, generative network 라고도 불림 3. 활용데이터에 대한 차원 축소, 이상 탐지, 노이즈 제거, 인식, 생성 등에 활용됨 4. VAE(; Variational AutoEncoder)기본적인 형태의 오토인코더와 다름확률적 오토인코더로, 학습이 끝난 이후에도 출력이 부분적으로 우연에 의해 결..
2024.09.05 -
0. 시작하기에 앞서본 게시물은 아래 참고링크의 게시글을 매우 적극적으로 활용하였으므로, 원 저작자에게 감사의 뜻을 남깁니다. 1. 개념U-Net은 오토인코더(autoencoder)와 같이 데이터의 차원을 축소했다가 다시 확장하는 방식의 모델로, Semantic Segmentation을 수행할 수 있다.그러나 오토인코더는 인코딩 단계에서 차원 축소를 거치면서 이미지 객체에 대한 자세한 위치 정보를 잃게 되고, 디코딩 단계에서도 저차원의 정보만을 이용하기 때문에 위치 정보 손실을 회복하지 못하는 단점이 존재한다.이러한 단점을 극복하기 위해 U-Net은 고차원 정보도 함께 이용하여 이미지의 특징을 추출함과 동시에 정확한 위치 파악을 하기위해 인코딩 단계의 각 레이어에서 얻은 특징을 디코딩 단계의 각 레이어..
U-Net: Convolutional Networks for Biomedical Image Segmentation0. 시작하기에 앞서본 게시물은 아래 참고링크의 게시글을 매우 적극적으로 활용하였으므로, 원 저작자에게 감사의 뜻을 남깁니다. 1. 개념U-Net은 오토인코더(autoencoder)와 같이 데이터의 차원을 축소했다가 다시 확장하는 방식의 모델로, Semantic Segmentation을 수행할 수 있다.그러나 오토인코더는 인코딩 단계에서 차원 축소를 거치면서 이미지 객체에 대한 자세한 위치 정보를 잃게 되고, 디코딩 단계에서도 저차원의 정보만을 이용하기 때문에 위치 정보 손실을 회복하지 못하는 단점이 존재한다.이러한 단점을 극복하기 위해 U-Net은 고차원 정보도 함께 이용하여 이미지의 특징을 추출함과 동시에 정확한 위치 파악을 하기위해 인코딩 단계의 각 레이어에서 얻은 특징을 디코딩 단계의 각 레이어..
2024.09.02