def mk_dataset():
face_dir = "/Users/zoushuai/Python/lab/datasets/original/face"
nonface_dir = "/Users/zoushuai/Python/lab/datasets/original/nonface"
face_list = resize_image(face_dir)
nonface_list = resize_image(nonface_dir)
train_set = face_list[0:250]
train_set.extend(
nonface_list[0:250]) # trainset contains 250 faces and 250 nonfaces
train_set = np.array(train_set)
validate_set = face_list[250:500]
validate_set.extend(nonface_list[250:500]
) # validateset contains 250 faces and 250 nonfaces
validate_set = np.array(validate_set)
train_img2feature_list = []
validate_img2feature_list = []
for i in range(500):
npdFeature_train = NPDFeature(train_set[i])
train_img2feature_list.append(npdFeature_train.extract())
npdFeature_validate = NPDFeature(validate_set[i])
validate_img2feature_list.append(npdFeature_validate.extract())
train_img2feature_list = np.array(train_img2feature_list)
validate_img2feature_list = np.array(validate_img2feature_list)
AdaBoostClassifier.save(train_img2feature_list, 'train')
AdaBoostClassifier.save(validate_img2feature_list, 'validate')
def train(X_train, y_train):
clf = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),
n_weakers_limit=20)
print("Training a AdaBoost Classifier.")
clf.fit(X_train, y_train)
# If model directories don't exist, create them
if not os.path.isdir(os.path.split(adb_model_path)[0]):
os.makedirs(os.path.split(adb_model_path)[0])
clf.save(clf, adb_model_path)
def organizeData(inputFileName, outputFileName):
'''
函数作用：读取图片，转化成灰度格式图片，抽取NPD特征，保存到本地
:param outputFileName:输出文件名
:param inputFileName:输入文件名
:return:
'''
ImageList = getAllImage(inputFileName)
grayList = convertImage(ImageList, 24)
featureList = getNPDFeature(grayList)
AdaBoostClassifier.save(featureList, outputFileName)
def get_features(path):
image_paths = [os.path.join(path, f) for f in os.listdir(path)]
features = []
# read the images and extract the features
for image_path in image_paths:
img = cv2.imread(image_path)
# convert into gray image
gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
img_reshape = cv2.resize(gray_img, (25, 25), interpolation=cv2.INTER_CUBIC)
image = feature.NPDFeature(img_reshape)
pre_features = feature.NPDFeature.extract(image)
AdaBoostClassifier.save(pre_features, "save.p")
face_feature = AdaBoostClassifier.load("save.p")
features.append(face_feature)
return features
def pre_process():
face_features = numpy.array([])
nonface_features = numpy.array([])
features = numpy.array([])
#转化图片
process_image(path_face_origin, path_face_purpose)
process_image(path_nonface_origin, path_nonface_purpose)
num_face = len(os.listdir('datasets/original/face'))
num_nonface = len(os.listdir('datasets/original/nonface'))
#获取特征
face_features = get_feature(path_face_purpose)
nonface_features = get_feature(path_nonface_purpose)
print(face_features.shape,nonface_features.shape)
print(num_face,num_nonface)
#改变特征形状
face_features.shape = num_face, 165600
nonface_features.shape = num_nonface, 165600
#准备加入label
face_y = numpy.ones(shape=(num_face, 1))
nonface_y = numpy.zeros(shape=(num_nonface, 1))
nonface_y -= numpy.ones(shape=(num_nonface, 1))
#加入label
face_features = numpy.concatenate([face_features, face_y], axis=1)
nonface_features = numpy.concatenate([nonface_features, nonface_y], axis=1)
#将所有数据合并
features = numpy.row_stack((face_features,nonface_features))
#写入缓存
AdaBoostClassifier.save(features,"feature.data")
return features
def Feature_extract():
#提取特征
face_path = '.\\datasets\\original\\face\\face_%03d.jpg'
faces_path = []
for i in range(500):
faces_path.append(face_path % i)
nonface_path = '.\\datasets\\original\\nonface\\nonface_%03d.jpg'
nonfaces_path = []
for i in range(500):
nonfaces_path.append(nonface_path % i)
train = np.zeros((1000, 165600))
for i in range(500):
img = Image.open(faces_path[i])
img = img.convert('L').resize((24, 24))
nf = NPDFeature(np.array(img))
train[i * 2] = nf.extract()
img = Image.open(nonfaces_path[i])
img = img.convert('L').resize((24, 24))
nf = NPDFeature(np.array(img))
train[i * 2 + 1] = nf.extract()
AdaBoostClassifier.save(train, 'train.txt')
os.listdir('datasets/original/face'))))
pathes = np.append(
pathes,
np.array(
list(
map(lambda s: 'datasets/original/nonface/' + s,
os.listdir('datasets/original/nonface')))))
for index, path in enumerate(pathes):
with Image.open(path) as image:
print(index, path)
image = image.convert('L')
image = image.resize((24, 24))
imageData = np.array(image)
npd = feature.NPDFeature(imageData)
features.append(npd.extract())
AdaBoostClassifier.save(features, 'features.dump')
features = np.array(features)
print(features.shape)
X_train, X_val, y_train, y_val = train_test_split(features,
y,
test_size=0.25)
classifier = AdaBoostClassifier(DecisionTreeClassifier, 5)
classifier.fit(X_train, y_train)
score = classifier.predict_scores(X_val, y_val)
predict = classifier.predict(X_val)
y_val = np.array(list(map(lambda x: int(x), y_val.reshape(1, -1)[0])))
def split_dataset(dataset, train_ratio=0.8):
"""
:return: X_train, y_train, X_valid, y_valid
"""
pivot = int(2 * SAMPLES_N * train_ratio)
train_set = dataset[0][:pivot], dataset[1][:pivot]
valid_set = dataset[0][pivot:], dataset[1][pivot:]
return train_set + valid_set
if __name__ == "__main__":
X_train, y_train, X_valid, y_valid = split_dataset(load_dataset())
adaBoost = AdaBoostClassifier(DecisionTreeClassifier, WEAKERS_LIMIT)
accs = adaBoost.fit(X_train, y_train, X_valid, y_valid)
plt.figure(figsize=[8, 5])
plt.title('Accuracy')
plt.xlabel('Num of weak classifiers')
plt.ylabel('Accuracy')
plt.plot(accs[0], '--', c='b', linewidth=3, label='train')
plt.plot(accs[1], c='r', linewidth=3, label='valid')
plt.legend()
plt.grid()
plt.savefig('AdaBoost-accuracy.png')
plt.show()
AdaBoostClassifier.save(adaBoost, 'AdaBoost-Model.pkl')
img_nonface=mpimg.imread(currentpath2+"{:0>3d}".format(i)+".jpg")
img_nonface_=rgb2gray(img_nonface)
f=NPDFeature(img_nonface_)
feature_=f.extract()
feature.append(feature_)
label.append(-1)
if __name__ == "__main__":
# write your code here
readimg()
train_feature,validation_feature,train_label,validation_label=train_test_split(feature,label,test_size=0.3)
#adaboost
adaboostClassifier=AdaBoostClassifier(DecisionTreeClassifier(max_depth = 1, random_state = 1),20)
adaboostClassifier.save(train_feature,'train_feature')
adaboostClassifier.save(train_label,'train_label')
adaboostClassifier.save(validation_feature,'validation_feature')
adaboostClassifier.save(validation_label,'validation_label')
adaboostClassifier.fit(train_feature,train_label)
adaboostClassifier.drawPic()
'''
#debug
adaboostClassifier=AdaBoostClassifier(DecisionTreeClassifier(max_depth = 1, random_state = 1),20)
train_feature=adaboostClassifier.load("train_feature")
train_label=adaboostClassifier.load("train_label")
adaboostClassifier.fit(train_feature,train_label)
adaboostClassifier.drawPic()
img = Image.open(nonfaces_path[i])
img = img.convert('L').resize((24, 24))
nf = NPDFeature(np.array(img))
train[i * 2 + 1] = nf.extract()
AdaBoostClassifier.save(train, 'train.txt')
try:
X = AdaBoostClassifier.load("train.txt")
except IOError:
Feature_extract()
X = AdaBoostClassifier.load("train.txt")
Y = np.zeros((1000, 1))
for i in range(1000):
Y[i] = (i + 1) % 2
Y = np.where(Y > 0, 1, -1)
X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2)
booster = AdaBoostClassifier(DecisionTreeClassifier, 15)
booster.fit(X_train, Y_train)
predict = booster.predict(X_test)
wrong_count = 0
for j in range(predict.shape[0]):
if predict[j] != Y_test[j]:
wrong_count += 1
AdaBoostClassifier.save(classification_report(Y_test, predict),
"classifier_report.txt")
pass
