faceGAN 156000
글
faceGAN by PyTorch
https://github.com/yunjey/pytorch-tutorial/tree/master/tutorials/03-advanced/deep_convolutional_gan
node.js upgrade by nvm
https://www.digitalocean.com/community/tutorials/how-to-install-node-js-on-ubuntu-16-04
R: tesseract OCR
install.packages ( "tesseract" ) img <- image_read ( "http://jeroen.github.io/images/testocr.png" ) print ( img ) # Extract text cat ( image_ocr ( img ) )
R install & Run in Ubuntu Terminal
https://www.digitalocean.com/community/tutorials/how-to-install-r-on-ubuntu-16-04-2
MFCC
Applications [ edit ] MFCCs are commonly used as features in speech recognition [6] systems, such as the systems which can automatically recognize numbers spoken into a telephone. MFCCs are also increasingly finding uses in music information retrieval applications such as genre classification, audio similarity measures, etc. [7] Noise sensitivity [ edit ] MFCC values are not very robust in the presence of additive noise, and so it is common to normalise their values in speech recognition systems to lessen the influence of noise. Some researchers propose modifications to the basic MFCC algorithm to improve robustness, such as by raising the log-mel-amplitudes to a suitable power (around 2 or 3) before taking the DCT, which reduces the influence of low-energy components. [8]
2d fft by imagemagick
convert image.png -fft +depth +adjoin fft_image_%d.png brew reinstall imagemagick --with-fftw
1D 2D 3D FFT by Pytorch
# Example that does a batch of three 2D transformations of size 4 by 5. import torch import pytorch_fft.fft as fft A, zeros = torch.randn( 3 , 4 , 5 ).cuda(), torch.zeros( 3 , 4 , 5 ).cuda() B_real, B_imag = fft.fft2(A_real, A_imag) fft.ifft2(B_real, B_imag) # equals (A, zeros) B_real, B_imag = fft.rfft2(A) # is a truncated version which omits # redundant entries reverse(torch.arange( 0 , 6 )) # outputs [5,4,3,2,1,0] reverse(torch.arange( 0 , 6 ), 2 ) # outputs [4,5,2,3,0,1] expand(B_real) # is equivalent to fft.fft2(A, zeros)[0] expand(B_imag, imag = True ) # is equivalent to fft.fft2(A, zeros)[1]
spectrogram vs scalogram
- 기저함수의 선택이 가능하다. - 파형보단 주파수 분포가 중요하다. - 파형 전체를 보여준다. - 속도가 느리다. -저주파로 갈수록 정보가 제한적이다. - - 가로축으로 파형도 어느정도 표현된다. - 파형을 잘라서 스텍트럼을 보여준다. - 기저함수의 선택이 불가능하다. - 속도가 빠르다. - 웨이블릿 기저 함수들. - 프랙탈 구조를 갖는 놈들도 보인다. - 이걸로 오그먼테이션도 가능할듯 싶다.
ofxTSNE code
// h----------------------------- #pragma once #include "ofMain.h" #include "ofxCcv.h" // Another OpenCV #include "ofxTSNE.h" // T-SNE Plot #include "ofxGui.h" class ofApp : public ofBaseApp{ public: void setup(); void update(); void draw(); void keyPressed(int key); void keyReleased(int key); void mouseMoved(int x, int y ); void mouseDragged(int x, int y, int button); void mousePressed(int x, int y, int button); void mouseReleased(int x, int y, int button); void mouseEntered(int x, int y); voi...
Confusion Matrix Multiclass using R
require( ggplot2 ) input <- read.delim( " conf_matrix.csv " , header = TRUE , sep = " , " ) input.matrix <- data.matrix( input ) input.matrix.normalized <- normalize( input.matrix ) colnames( input.matrix.normalized ) = c( " A " , " B " , " C " , " D " , " E " , " F " , " G " , " H " , " I " , " J " , " K " , " L " , " M " , " N " ) rownames( input.matrix.normalized ) = colnames( input.matrix.normalized ) confusion <- as.data.frame(as.table( input.matrix.normalized )) plot <- ggplot( confusion ) plot + geom_tile(aes( x = Var1 , y = Var2 , fill = Freq )) + sc...
Confusion Matrix using R
actual, predicted A, B B, B, C, C B, A --------------------- require( caret ) results <- read.delim( " classifier-results.csv " , sep = " , " , header = FALSE ) names( results ) <- c( " actual " , " predicted " ) results.matrix <- confusionMatrix( results $ predicted , results $ actual )
랜덤 셔플 파일 이동
$ mv $(ls | gshuf -n30) ../result/ 이미지가 있는 디렉토리에서 .. 30개 만큼 리절트 폴더로 mv $ brew install coreutils
[python] csv file to binary
#!/usr/bin/env python import array with open ( 'input.csv' , 'rt' ) as f : text = f . read () entries = text . split ( ',' ) values = [ int ( x ) for x in entries ] # do a scalar here: if your input goes from [-100, 100] then # you may need to translate/scale into [0, 2^16-1] for # 16-bit PCM # e.g.: # values = [(val * scale) for val in values] with open ( 'output.pcm' , 'wb' ) as out : pcm_vals = array . array ( 'h' , values ) # 16-bit signed pcm_vals . tofile ( out ) You could also use Python's wave module instead of just writing raw PCM. Here's how the example above works: $ echo 1 , 2 , 3 , 4 , 5 , 6 , 7 > input . csv $ ./ so_pcm . py $ xxd output . pcm 0000000 : 0100 0200 0300 0400 0500 0600 0700 ..............
ofx SVM Classification
// h----------------------------- #pragma once #include "ofMain.h" #include "ofxLearn.h" // 딥러닝 패키지 class ofApp : public ofBaseApp{ public: void setup(); void update(); void draw(); void keyPressed(int key); void keyReleased(int key); void mouseMoved(int x, int y); void mouseDragged(int x, int y, int button); void mousePressed(int x, int y, int button); void mouseReleased(int x, int y, int button); void windowResized(int w, int h); void dragEvent(ofDragInfo dragInfo); void gotMessage(ofMessage msg); ofxLearnSVM classifier; // 써포트 벡터 머신 vector<vector<double> > trainingEx...
makeROCData.py
#!/usr/bin/python import os import re from scipy import ndimage, misc import torch import torch.nn as nn #from __future__ import print_function import argparse from PIL import Image import torchvision.models as models import skimage.io from torch.autograd import Variable as V from torch.nn import functional as f from torchvision import transforms as trn # define image transformation centre_crop = trn.Compose([ trn.ToPILImage(), trn.Scale(256), trn.CenterCrop(224), trn.ToTensor(), trn.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) ]) images = [] for root, dirnames, filenames in os.walk("/home/kerb/Documents/data_bm_0913/test/benign/"): for filename in filenames: # for all files ...
[PyTorch] testAccuacy in CPU
import torch import torch.nn as nn #from __future__ import print_function import argparse from PIL import Image import torchvision.models as models import skimage.io from torch.autograd import Variable as V from torch.nn import functional as f from torchvision import transforms as trn # define image transformation centre_crop = trn.Compose([ trn.ToPILImage(), trn.Scale(256), trn.CenterCrop(224), trn.ToTensor(), trn.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) ]) filename=r'/Users/dti/Desktop/PyTorch4testAccuracy/hymenoptera_data/test/bees/1.jpg' img = skimage.io.imread(filename) x = V(centre_crop(img).unsqueeze(0), volatile=True) model = models.__dict__['resnet18'] model = torch.load('model5.pth') logit = model(x) print(logit) h_x = f...
[python] redirecting to file
f = open ( 'out.txt' , 'w' ) print >> f , 'Filename:' , filename # or f.write('...\n') f . close ()