if __name__ == '__main__':
x_train, y_train = load_cifar10_data(dataset='train')
x_test, y_test = load_cifar10_data(dataset='test')
print("Dense SIFT feature extraction")
x_train_feature = [extract_DenseSift_descriptors(img) for img in x_train]
x_test_feature = [extract_DenseSift_descriptors(img) for img in x_test]
x_train_kp, x_train_des = zip(*x_train_feature)
x_test_kp, x_test_des = zip(*x_test_feature)
print("Train/Test split: {:d}/{:d}".format(len(y_train), len(y_test)))
print("Codebook Size: {:d}".format(VOC_SIZE))
print("Pyramid level: {:d}".format(PYRAMID_LEVEL))
print("Building the codebook, it will take some time")
codebook = build_codebook(x_train_des, VOC_SIZE)
# import cPickle
#
# with open('./spm_lv1_codebook.pkl', 'w') as f:
# cPickle.dump(codebook, f)
print("Spatial Pyramid Matching encoding")
x_train = [
spatial_pyramid_matching(x_train[i],
x_train_des[i],
codebook,
level=PYRAMID_LEVEL)
for i in range(len(x_train))
]
x_test = [
print "抽取 SIFT 特征..."
x_train = [extract_sift_descriptors(img) for img in x_train]
x_test = [extract_sift_descriptors(img) for img in x_test]
# 移除没有抽取出sift特征的图片
x_train = [each for each in zip(x_train, y_train) if each[0] is not None]
x_train, y_train = zip(*x_train)
x_test = [each for each in zip(x_test, y_test) if each[0] is not None]
x_test, y_test = zip(*x_test)
print "Train/Test split: {:d}/{:d}".format(len(y_train), len(y_test))
print "Vocab Size: {:d}".format(VOC_SIZE)
# 使用Kmeans聚类算法对抽取出来的特征进行聚类，VOC_SIZE表示k-means中的k
# 这一步可能需要很长时间...
print "构建词典..."
codebook = build_codebook(x_train, voc_size=VOC_SIZE)
with open('./bow_codebook.pkl', 'w') as f:
cPickle.dump(codebook, f)
# 将图片转成它的向量化表示
print "图片向量化表示..."
x_train = [input_vector_encoder(x, codebook) for x in x_train]
x_train = np.asarray(x_train)
x_test = [input_vector_encoder(each, codebook) for each in x_test]
x_test = np.asarray(x_test)
# 输入到分类器进行分类
print("训练与测试分类器...")
svm_classifier(x_train, y_train, x_test, y_test)
x_train, y_train = load_cifar10_data(dataset='train')
x_test, y_test = load_cifar10_data(dataset='test')
print "SIFT feature extraction"
x_train = [extract_sift_descriptors(img) for img in x_train]
x_test = [extract_sift_descriptors(img) for img in x_test]
# Remove None in SIFT extraction
x_train = [each for each in zip(x_train, y_train) if each[0] != None]
x_train, y_train = zip(*x_train)
x_test = [each for each in zip(x_test, y_test) if each[0] != None]
x_test, y_test = zip(*x_test)
print "Train/Test split: {:d}/{:d}".format(len(y_train), len(y_test))
print "Codebook Size: {:d}".format(VOC_SIZE)
print "Building the codebook, it will take some time"
codebook = build_codebook(x_train, voc_size=VOC_SIZE)
import cPickle
with open('./bow_codebook.pkl','w') as f:
cPickle.dump(codebook, f)
print "Bag of words encoding"
x_train = [input_vector_encoder(x, codebook) for x in x_train]
x_train = np.asarray(x_train)
x_test = [input_vector_encoder(each, codebook) for each in x_test]
x_test = np.asarray(x_test)
svm_classifier(x_train, y_train, x_test, y_test)
print "Training data load from {}".format(args.train)
print "Test data load from {}".format(args.test)
X, y = load_my_data(args.train)
# to save time...
X = X[:200]
y = y[:200]
print "Codebook Size: {:d}".format(spm.VOC_SIZE)
print "Pyramid level: {:d}".format(spm.PYRAMID_LEVEL)
print "Dense SIFT feature extraction"
x_feature = [extract_DenseSift_descriptors(img) for img in X]
x_kp, x_des = zip(*x_feature)
print "Building the codebook, it will take some time"
codebook = build_codebook(x_des, spm.VOC_SIZE)
#import cPickle
#with open('./data/codebook_spm.pkl','w') as f:
# cPickle.dump(codebook, f)
print "Spatial Pyramid Matching encoding"
X = [spm.spatial_pyramid_matching(X[i],
x_des[i],
codebook,
level=spm.PYRAMID_LEVEL)
for i in xrange(len(x_des))]
X = np.asarray(X)
svm_classifier(X, y)
if __name__ == '__main__':
x_train, y_train = load_cifar10_data(dataset='train')
x_test, y_test = load_cifar10_data(dataset='test')
print "Dense SIFT feature extraction"
x_train_feature = [extract_DenseSift_descriptors(img) for img in x_train]
x_test_feature = [extract_DenseSift_descriptors(img) for img in x_test]
x_train_kp, x_train_des = zip(*x_train_feature)
x_test_kp, x_test_des = zip(*x_test_feature)
print "Train/Test split: {:d}/{:d}".format(len(y_train), len(y_test))
print "Codebook Size: {:d}".format(VOC_SIZE)
print "Pyramid level: {:d}".format(PYRAMID_LEVEL)
print "Building the codebook, it will take some time"
codebook = build_codebook(x_train_des, VOC_SIZE)
import cPickle
with open('./spm_lv1_codebook.pkl','w') as f:
cPickle.dump(codebook, f)
print "Spatial Pyramid Matching encoding"
x_train = [spatial_pyramid_matching(x_train[i],
x_train_des[i],
codebook,
level=PYRAMID_LEVEL)
for i in xrange(len(x_train))]
x_test = [spatial_pyramid_matching(x_test[i],
x_test_des[i],
codebook,
level=PYRAMID_LEVEL) for i in xrange(len(x_test))]