In [1]:

    

from numpy import *
from PIL import *
import pickle
from pylab import *
import os


    

In [2]:

    

import sift
import dsift
dsift = reload(dsift)
import imtools
imtools = reload(imtools)


    

In [3]:

    

def read_gesture_features_labels(path):
    # make a list of the files with .dsift at the end
    featlist = [os.path.join(path, f) for f in os.listdir(path)
               if f.endswith('.dsift')]
    
    # read features
    features = []
    for featfile in featlist:
        l, d = sift.read_features_from_file(featfile)
        features.append(d.flatten())
    features = array(features)
    
    # generate labels
    labels = [featfile.split('/')[-1][0] for featfile in featlist]
    
    return features, array(labels)


    

In [4]:

    

features, labels = read_gesture_features_labels('train2/')
test_features, test_labels = read_gesture_features_labels('test2/')
classnames = unique(labels)


    

In [5]:

    

# the first letter of the file name is the label
print labels


    

    




['V' 'A' 'A' 'B' 'P' 'B' 'P' 'C' 'F' 'A' 'P' 'V' 'F' 'F' 'P' 'B' 'P' 'F'
 'F' 'B' 'P' 'F' 'F' 'A' 'F' 'P' 'A' 'A' 'F' 'V' 'A' 'P' 'V' 'V' 'P' 'A'
 'P' 'C' 'B' 'B' 'C' 'P' 'C' 'V' 'P' 'V' 'V' 'C' 'B' 'C' 'V' 'P' 'P' 'B'
 'B' 'C' 'A' 'C' 'V' 'B' 'V' 'B' 'F' 'P' 'A' 'P' 'B' 'P' 'B' 'V' 'F' 'B'
 'C' 'A' 'C' 'C' 'F' 'F' 'V' 'P' 'C' 'F' 'P' 'A' 'A' 'F' 'P' 'B' 'A' 'V'
 'F' 'V' 'C' 'C' 'B' 'C' 'A' 'F' 'F' 'A' 'V' 'C' 'V' 'V' 'A' 'B' 'A' 'C'
 'B' 'C' 'B' 'B' 'V' 'A' 'C' 'V' 'C' 'F' 'C' 'F' 'F' 'P' 'A' 'B' 'P' 'P'
 'B' 'B' 'A' 'F' 'B' 'A' 'C' 'C' 'V' 'A' 'C' 'P' 'P' 'V' 'F' 'V' 'P' 'A'
 'F' 'A' 'V' 'F' 'F' 'C' 'C' 'B' 'V' 'P' 'B' 'V' 'P' 'V' 'A' 'C' 'V' 'A'
 'V' 'A' 'F' 'F' 'B' 'A' 'B' 'F' 'C' 'B' 'C' 'P']

In [6]:

    

import knn


    

In [7]:

    

# k neighbors
k = 1
knn_classifier = knn.KnnClassifier(labels, features)


    

In [8]:

    

res = array([knn_classifier.classify(test_features[i], k) for i in
            range(len(test_labels))])

# accuracy
acc = sum(1.0*(res==test_labels))/len(test_labels)
print 'Accuracy:', acc


    

    




Accuracy: 0.242718446602

In [9]:

    

def print_confusion(res, test_labels, classnames):
    
    n = len(classnames)
    
    class_ind = dict([(classnames[i], i) for i in range(n)])
    
    confuse = zeros((n, n))
    for i in range(len(test_labels)):
        confuse[class_ind[res[i]], class_ind[test_labels[i]]] += 1
    
    print 'Confusion matrix for'
    print classnames
    print confuse


    

In [10]:

    

print_confusion(res, test_labels, classnames)


    

    




Confusion matrix for
['A' 'B' 'C' 'F' 'P' 'V']
[[  1.   0.   2.   2.   0.   0.]
 [  2.   5.   0.  10.   7.   5.]
 [  2.   0.   7.   1.   1.   0.]
 [  1.   3.   1.   6.   0.   0.]
 [  0.   0.   4.   0.   2.   0.]
 [  4.   4.   4.  10.  15.   4.]]

In [11]:

    

import pca


    

In [74]:

    

V, S, m = pca.pca(array(features))
V = V[:75]


    

In [75]:

    

features2 = [dot(V, f-m) for f in features]
test_features2 = [dot(V, f-m) for f in test_features]


    

In [76]:

    

knn_classifier2 = knn.KnnClassifier(labels, features2)


    

In [77]:

    

res2 = array([knn_classifier2.classify(test_features2[i], k) for i in
            range(len(test_labels))])

# accuracy
acc2 = sum(1.0*(res2==test_labels))/len(test_labels)
print 'Accuracy:', acc2


    

    




Accuracy: 0.271844660194

In [78]:

    

print_confusion(res2, test_labels, classnames)


    

    




Confusion matrix for
['A' 'B' 'C' 'F' 'P' 'V']
[[  1.   0.   1.   2.   0.   0.]
 [  2.   6.   0.  10.   6.   4.]
 [  2.   0.   8.   0.   0.   0.]
 [  1.   3.   1.   6.   0.   0.]
 [  0.   0.   4.   0.   2.   0.]
 [  4.   3.   4.  11.  17.   5.]]

In [ ]: