|
|
|
|
|
"""
|
|
|
|
|
|
Programmer : EOF
|
|
|
|
|
|
File : config.py
|
|
|
|
|
|
Date : 2016.01.06
|
|
|
|
|
|
E-mail : jasonleaster@163.com
|
|
|
|
|
|
License : MIT License
|
|
|
|
|
|
|
|
|
|
|
|
Description :
|
|
|
|
|
|
This is a configure file for this project.
|
|
|
|
|
|
|
|
|
|
|
|
"""
|
|
|
|
|
|
import os
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def get_project_rootpath():
|
|
|
|
|
|
"""
|
|
|
|
|
|
获取项目根目录。此函数的能力体现在，不论当前module被import到任何位置，都可以正确获取项目根目录
|
|
|
|
|
|
:return:
|
|
|
|
|
|
"""
|
|
|
|
|
|
path = os.path.realpath(os.curdir)
|
|
|
|
|
|
while True:
|
|
|
|
|
|
# PyCharm项目中，'.idea'是必然存在的，且名称唯一
|
|
|
|
|
|
if '.idea' in os.listdir(path):
|
|
|
|
|
|
path = str(path).replace('\\', '/')
|
|
|
|
|
|
path += '/'
|
|
|
|
|
|
return path
|
|
|
|
|
|
path = os.path.dirname(path)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rpath = get_project_rootpath()
|
|
|
|
|
|
DEBUG_MODEL = True
|
|
|
|
|
|
USING_CASCADE = False
|
|
|
|
|
|
|
|
|
|
|
|
# training set directory for face and non-face images
|
|
|
|
|
|
# TRAINING_FACE = os.path.normpath(os.path.join(rpath, "FaceDetection/TrainingImages/FACES/"))
|
|
|
|
|
|
# TRAINING_NONFACE = os.path.normpath(os.path.join(rpath, "FaceDetection/TrainingImages/NFACES/"))
|
|
|
|
|
|
TRAINING_FACE = "./TrainingImages/FACES/"
|
|
|
|
|
|
TRAINING_NONFACE = "./TrainingImages/NFACES/"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# test set directory for face and non-face images
|
|
|
|
|
|
TEST_FACE = "./TrainingImages/FACES/"
|
|
|
|
|
|
TEST_NONFACE = "./TrainingImages/NFACES/"
|
|
|
|
|
|
|
|
|
|
|
|
# single image for testing
|
|
|
|
|
|
# TEST_IMG = "./Test/soccer.gif"
|
|
|
|
|
|
TEST_IMG = "./Test/nens.png"
|
|
|
|
|
|
|
|
|
|
|
|
FEATURE_FILE_TRAINING = "./features/features_train.cache"
|
|
|
|
|
|
FEATURE_FILE_TESTING = "./features/features_test.cache"
|
|
|
|
|
|
|
|
|
|
|
|
FEATURE_FILE_SUBSET = "./features/features_train_subset"
|
|
|
|
|
|
FEATURE_FILE_SUBSET_0 = "./features/features_train_subset0.cache"
|
|
|
|
|
|
FEATURE_FILE_SUBSET_1 = "./features/features_train_subset1.cache"
|
|
|
|
|
|
|
|
|
|
|
|
# For parallel
|
|
|
|
|
|
PROCESS_NUM = 2
|
|
|
|
|
|
|
|
|
|
|
|
ADABOOST_CACHE_FILE = "./model/adaboost_classifier.cache"
|
|
|
|
|
|
ROC_FILE = "./model/roc.cache"
|
|
|
|
|
|
|
|
|
|
|
|
FIGURES = "./figure/"
|
|
|
|
|
|
|
|
|
|
|
|
# image size in the training set 19 * 19
|
|
|
|
|
|
TRAINING_IMG_HEIGHT = 19
|
|
|
|
|
|
TRAINING_IMG_WIDTH = 19
|
|
|
|
|
|
|
|
|
|
|
|
# How many different types of Haar-feature
|
|
|
|
|
|
FEATURE_TYPE_NUM = 5
|
|
|
|
|
|
# How many number of features that a single training image have
|
|
|
|
|
|
FEATURE_NUM = 37862
|
|
|
|
|
|
# FEATURE_NUM = 16373
|
|
|
|
|
|
# FEATURE_NUM = 49608
|
|
|
|
|
|
|
|
|
|
|
|
# number of positive and negative sample will be used in the training process
|
|
|
|
|
|
# POSITIVE_SAMPLE = 4800
|
|
|
|
|
|
# NEGATIVE_SAMPLE = 9000
|
|
|
|
|
|
POSITIVE_SAMPLE = 240
|
|
|
|
|
|
NEGATIVE_SAMPLE = 450
|
|
|
|
|
|
|
|
|
|
|
|
SAMPLE_NUM = POSITIVE_SAMPLE + NEGATIVE_SAMPLE
|
|
|
|
|
|
|
|
|
|
|
|
TESTING_POSITIVE_SAMPLE = 20
|
|
|
|
|
|
TESTING_NEGATIVE_SAMPLE = 20
|
|
|
|
|
|
|
|
|
|
|
|
TESTING_SAMPLE_NUM = TESTING_NEGATIVE_SAMPLE + TESTING_POSITIVE_SAMPLE
|
|
|
|
|
|
|
|
|
|
|
|
LABEL_POSITIVE = +1
|
|
|
|
|
|
LABEL_NEGATIVE = -1
|
|
|
|
|
|
|
|
|
|
|
|
WHITE = 255
|
|
|
|
|
|
BLACK = 0
|
|
|
|
|
|
|
|
|
|
|
|
EXPECTED_TPR = 0.999
|
|
|
|
|
|
EXPECTED_FPR = 0.0005
|
|
|
|
|
|
|
|
|
|
|
|
# for CASCADE
|
|
|
|
|
|
EXPECTED_FPR_PRE_LAYYER = 0.1
|
|
|
|
|
|
EXPECTED_TPR_PRE_LAYYER = 0.999
|
|
|
|
|
|
|
|
|
|
|
|
# the threshold range of adaboost. (from -inf to +inf)
|
|
|
|
|
|
AB_TH_MIN = -15
|
|
|
|
|
|
AB_TH_MAX = +15
|
|
|
|
|
|
|
|
|
|
|
|
HAAR_FEATURE_TYPE_I = "I"
|
|
|
|
|
|
HAAR_FEATURE_TYPE_II = "II"
|
|
|
|
|
|
HAAR_FEATURE_TYPE_III = "III"
|
|
|
|
|
|
HAAR_FEATURE_TYPE_IV = "IV"
|
|
|
|
|
|
HAAR_FEATURE_TYPE_V = "V"
|
|
|
|
|
|
|
|
|
|
|
|
AB_TH = -3.
|
|
|
|
|
|
OVER_LAP_TH = 0.1
|
|
|
|
|
|
|
|
|
|
|
|
MAX_WEAK_NUM = 12
|
|
|
|
|
|
|
|
|
|
|
|
CASACADE_LIMIT = 3
|
|
|
|
|
|
ADABOOST_LIMIT = 150
|
|
|
|
|
|
|
|
|
|
|
|
SEARCH_WIN_STEP = 4
|
|
|
|
|
|
|
|
|
|
|
|
DETECT_START = 1.
|
|
|
|
|
|
DETECT_END = 2.
|
|
|
|
|
|
DETECT_STEP = 0.2
|
