alfredplpl

library(randomForest)

#「フィッシャーのあやめ」データセットを読み出す
data(iris)

#データセットを特徴量とラベルに分割
features<-iris[1:4]
labels<-iris[5]

#ラベルを因子化

alfredplpl

                setosa versicolor virginica MeanDecreaseAccuracy MeanDecreaseGini
Sepal.Length  7.277437  7.0994514  8.087167            10.842432         9.553025
Sepal.Width   5.029336 -0.4402422  2.381594             3.438372         2.528796
Petal.Length 22.909121 32.9889332 30.209000            35.201262        43.785432
Petal.Width  21.741775 30.5797963 31.309059            32.724397        43.400080

alfredplpl

import cv2  
import numpy as np

class BagOfFeatures:
    """This is a class of Bag-of-Features by K-means for OpenCV"""
    codebookSize=0
    classifier=None
    def __init__(self, codebookSize):
        self.codebookSize=codebookSize
        self.classifier=cv2.KNearest()

alfredplpl

#see also: http://scikit-learn.org/stable/modules/generated/sklearn.mixture.GMM.html
import numpy as np
from sklearn import mixture,preprocessing

class BagOfFeaturesGMM:
    """This is a class of Bag-of-Features by GMM """
    codebookSize=0
    classifier=None
    
    def __init__(self, codebookSize):

alfredplpl

cv::Mat image=cv::imread("./image.png",-1);

cv::Rect rect(startColumn,startRow,image.cols,image.rows);
cv::Mat subdst=dst(rect);
image.copyTo(subdst);

alfredplpl

cv::Mat image=cv::imread("./image.png",-1);
cv::Mat trans=cv::imread("./trans.png",0);

cv::Rect rect1(startColumn1,startRow1,trans.cols,trans.rows);
cv::Mat subdst1=dst(rect1);
trans.convertTo(subdst1,subdst1.type(),alpha);//double alphaは透過率[0.0, 1.0]

cv::Rect rect2(startColumn2,startRow2,image.cols,image.rows);
cv::Mat subdst2=dst(rect2);
image.copyTo(subdst2);

alfredplpl

cv::Mat image=cv::imread("./image.png",-1);
cv::Mat image_mask=cv::imread("./mask.png",0);
 
cv::Rect rect(startColumn,startRow,image.cols,image.rows);
cv::Mat subdst=dst(rect);
image.copyTo(subdst,image_mask);

alfredplpl

# -*- coding: utf-8 -*-
#BoF(SIFT)により一般物体認識を行うサンプルコード
#今回はイルカと象を識別
#評価用のデータセットとしてCaltech 101を使用
#see also: http://www.vision.caltech.edu/Image_Datasets/Caltech101/

import cv2
import numpy as np
from sklearn import svm
from sklearn import cross_validation

alfredplpl

Ave. score(BagOfFeatures):70.492308[%]
Ave. score(BagOfFeaturesGMM):71.323077[%]

alfredplpl

#Its behavior is like hash (associative array).
hash.test=list("square"=function(x){x**2},
		   "pi"=3.1415926535)
hash.test[["square"]](2)
hash.test[["pi"]]

#probability that a random selected number is a prime number.
6/hash.test[["square"]](hash.test[["pi"]])