def __init__(self, extention, args):
output_filename = utils.root_path() + _filename_pattern() + extention
self._args = [utils.exe_path('7za'), 'a', output_filename]
self._args += args
self._args.append('*')
self._work_dir = tempfile.mkdtemp(prefix='', dir=utils.temp_path())
try:
os.remove(output_filename)
except OSError:
pass
def __init__(self,
params_path,
build_model,
load_data=load_data,
train=True,
batch_size=200,
learning_rate=1e-2,
validate_rate=0.2,
epochs=10):
self.params_path = utils.root_path(params_path)
self.load_data = load_data
self.train = train
self.batch_size = batch_size
self.validate_rate = validate_rate
self.epochs = epochs
self.model = build_model()
self.compile_model(learning_rate)
# load train params if exists
self.load_weights()
# coding=utf-8
import os
import cv2
import utils
import imutils
import numpy as np
from PIL import Image
from v1.classifier import FaceClassifier
IMAGE_PATH = utils.root_path('doc/images/test2.jpg')
PREDICTION_POSTFIX = '_prediction'
SAVE_PREDICTION = True
PROB_THRESHOLD = 0.5
if __name__ == '__main__':
classifier = FaceClassifier()
target_W = utils.IM_WIDTH
target_H = utils.IM_HEIGHT
im = Image.open(IMAGE_PATH)
im = np.array(im)
faces = utils.detect_faces(im)
if len(faces):
xs, ls = [], []
for (x, y, w, h) in faces:
face = im[y: y + h, x: x + w, :]
face = imutils.resize(face, target_W, target_H)
if face.shape[0] == target_H and face.shape[1] == target_W:
xs.append(face)
ls.append((x, y + 5 if y < 20 else y - 5))
prediction = classifier.predict(np.array(xs))
def _configure():
# hard-coded seed helps to have predictable generated names
random.seed(31337)
# set current directory to project's root directory
# in order to make build process portable and self-contained,
# and to do not store anything in current or user's home/temp directories
os.chdir(utils.root_path())
try:
os.mkdir(utils.cache_path())
except OSError:
pass
rarfile.UNRAR_TOOL = utils.exe_path('unrar')
parser = argparse.ArgumentParser()
# Generic arguments
parser.add_argument('-v',
'--verbosity',
type=int,
choices=range(0, 4),
default=0,
help='set output verbosity level')
parser.add_argument('--disable-optimization',
help='disable WAD files optimization',
action='store_true')
parser.add_argument('-c',
'--compression',
type=str,
choices=['none', 'default', 'pk3', '7zpk3', 'pk7'],
help='set output file compression')
parser.add_argument('-p',
'--profiling',
help='enable Python performance profiling',
action='store_true')
# Operational mode arguments
parser.add_argument(
'--zdoomed-souls',
help='build ZDoomed Souls instead of An Awesome Awesomeness',
action='store_true')
parser.add_argument(
'--check-repo-update',
help='look for new assets in web repository instead of '
' building a package',
action='store_true')
parser.add_argument(
'--clean-asset-cache',
help='delete cached assets instead of building a package',
action='store_true')
parser.add_argument('-d',
'--dry-run',
help='do all build steps but do not write a package',
action='store_true')
parser.add_argument(
'-s',
'--single',
type=int,
default=0,
help='build only one repository asset specified by its index')
# Patching-related arguments
parser.add_argument('--allow-set-pitch',
help='allow A_SetPitch() calls in DECORATE',
action='store_true')
parser.add_argument('--allow-class-replacement',
help='allow class replacement in DECORATE',
action='store_true')
parser.add_argument(
'--allow-doomednum',
help='allow editor number (DoomEdNum) assignment in DECORATE',
action='store_true')
parser.add_argument('--png-sprites',
help='convert all sprites to PNG format',
action='store_true')
parser.add_argument('--png-sprites-compression',
type=int,
choices=[value for value in range(-1, 10)],
default=-1,
help='set compression level for sprites in PNG format')
args = parser.parse_args()
patching.verbosity_level = args.verbosity
patching.allow_set_pitch = args.allow_set_pitch
patching.allow_class_replacement = args.allow_class_replacement
patching.allow_doomednum = args.allow_doomednum
patching.enable_optimization = not args.disable_optimization
patching.png_sprites = args.png_sprites
patching.png_sprites_compression = args.png_sprites_compression
if args.zdoomed_souls:
utils.set_mode(utils.MODE_ZDS)
return args
# all modules are in the lib directory
sys.path[0] = os.path.dirname(os.path.abspath(__file__)) + '/../lib'
import doomwad
from case_insensitive import CaseInsensitiveDict
from iwad_lumps import *
from iwad_actors import ACTORS_ALL
from iwad_sndinfo import LOGICAL_SOUNDS_ALL
from iwad_brightmaps import BRIGHTMAPS_ALL
from patching import (actor_stateful_pattern, actor_stateless_pattern,
actor_header_regex)
import utils
pk3_filename = utils.root_path() + 'realm667-aaa.pk3'
excluded_lump_names = [
# lumps
'ANIMDEFS',
'DECALDEF',
'DECORATE',
'GLDEFS',
'KEYCONF',
'LANGUAGE',
'LOADACS',
'LOCKDEFS',
'SNDINFO',
'TEXTURES',
# GLDEFS aliases
train_generator,
steps_per_epoch=steps_per_epoch,
callbacks=[ModelCheckpoint(self.weights_path)],
epochs=self.epoch)
# self.model.fit(x, y, batch_size=32, epochs=self.epoch)
def evaluate(self, x, y):
return self.model.evaluate(x, y, batch_size=32)
def predict(self, x, standard=True):
if standard:
x = x / 255.
return self.model.predict(x)
PATH_TRAIN = utils.root_path('data/love/roles')
PATH_VAL = utils.root_path('data/love/roles')
# 是否训练
TRAIN = True
# 是否验证
VALIDATE = False
if __name__ == '__main__':
print('Init model.')
classifier = FaceClassifier(lr=1e-2, epoch=50)
if TRAIN:
print('Train model.')
classifier.train(PATH_TRAIN, classes=utils.NAMES)
if VALIDATE:
names, index2name, name2index = utils.parse_name()
class FaceClassifier:
weights_path = utils.root_path('v1/weights/weights.h5')
def __init__(self, weight_path=None, lr=1e-2, epoch=30):
if weight_path is not None:
self.weights_path = weight_path
self.epoch = epoch
self.model = self.build_model()
self.compile_model(lr)
def build_model(self):
model = Sequential()
model.add(
Conv2D(8, (3, 3),
activation='relu',
input_shape=(utils.IM_HEIGHT, utils.IM_WIDTH, 3)))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(16, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(256, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(8, activation='softmax'))
return model
def compile_model(self, lr):
sgd = SGD(lr=lr, decay=1e-6, momentum=0.9, nesterov=True)
self.model.compile(loss='categorical_crossentropy', optimizer=sgd)
if os.path.exists(self.weights_path):
self.model.load_weights(self.weights_path)
print('Load weights.h5 successfully.')
else:
print('Model params not found.')
def train(self, train_dir, classes=None, batch_size=32):
file_num = utils.calculate_file_num(train_dir)
steps_per_epoch = file_num // batch_size
print('steps number is %d every epoch.' % steps_per_epoch)
train_data_gen = ImageDataGenerator(rescale=1. / 255,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True)
train_generator = train_data_gen.flow_from_directory(
train_dir,
classes=classes,
target_size=(utils.IM_WIDTH, utils.IM_HEIGHT),
batch_size=batch_size,
class_mode='categorical')
utils.ensure_dir(os.path.dirname(self.weights_path))
self.model.fit_generator(
train_generator,
steps_per_epoch=steps_per_epoch,
callbacks=[ModelCheckpoint(self.weights_path)],
epochs=self.epoch)
# self.model.fit(x, y, batch_size=32, epochs=self.epoch)
def evaluate(self, x, y):
return self.model.evaluate(x, y, batch_size=32)
def predict(self, x, standard=True):
if standard:
x = x / 255.
return self.model.predict(x)
# coding=utf-8
import cv2
import utils
import imutils
import numpy as np
from v1.classifier import FaceClassifier
# VIDEO_PATH = u'E:/Youku Files/transcode/爱情公寓 第一季 06_超清.mp4'
# SEEK = 2.5985e6
VIDEO_PATH = u'/Users/zijiao/Desktop/love1_3.mp4'
PATH_WRITE = utils.root_path('result.mp4')
SEEK = 2.6005e6
PROB_THRESHOLD = 0.5
CLASSIFY = True
SHOW_VIDEO = True
WRITE_VIDEO = False
if __name__ == '__main__':
# cap = cv2.VideoCapture(0)
cap = cv2.VideoCapture(VIDEO_PATH)
cap.set(0, SEEK)
width = int(cap.get(3))
height = int(cap.get(4))
fps = 24 # 视频帧率
# 指定写视频的格式, I420-avi, MJPG-mp4
def sign_up(total_count, needed_count):
dict_path = os.path.join(utils.root_path(), 'english.txt')
seed = utils.get_seed(dict_path, 12)
keys = shamir.crypto_keys(needed_count, total_count, seed)
WalletManager.create_wallet(seed.encode())
return keys
continue
index = np.argmax(prediction[0])
subdir = utils.NAMES_EN[index]
save_path = os.path.join(CLASSIFY_PATH, subdir)
utils.ensure_dir(save_path)
final_path = os.path.join(save_path, file_name)
cv2.imwrite(final_path, face)
# print('File %s saved.' % file_name)
return im
# 读取视频路径
# PATH_VIDEO = u'E:/Youku Files/transcode/爱情公寓 第一季 10_超清.mp4'
PATH_VIDEO = u'/Users/zijiao/Desktop/love1_3.mp4'
# 普通模式保存路径
PATH_SAVE = utils.root_path('data/love/images')
# 分类模式保存路径
CLASSIFY_PATH = utils.root_path('data/love/predict')
# seek位置
SEEK = 2.6005e6
# 识别阈值(0.0 ~ 1.0)
FACE_MIN = 0.0
FACE_MAX = 1.0
# 是否保存图片
SAVE_IMAGE = True
# 是否自动分类
CLASSIFY = True
# 每隔多少帧进行采样
STRIDES = 1
# 是否显示画面
SHOW_CONTENT = False
# Standard Library
from sys import stdout
import argparse
import logging
from os.path import abspath, basename, dirname, relpath
from typing import List
from time import strftime
# My Code
from markov_chars import generate as genc
from markov_chunks import generate as genw
from utils import root_path
output_file: str = root_path(
'generated',
f"{strftime('%H:%M:%S#%y-%m-%d')}.txt",
)
parser = argparse.ArgumentParser(description='Generate prose.',
prog=basename(dirname(abspath(__file__))))
parser.add_argument('--lookbehind',
'-n',
metavar='INT',
type=int,
nargs='?',
default=6,
choices=tuple(range(1, 20)),
required=False,
help='max length of lookbehind ngrams')
parser.add_argument('--length',
from PyQt5.QtWidgets import QApplication
import sys
import os
import utils
from app import Application
if __name__ == "__main__":
icon_path = os.path.join(utils.root_path(), 'icons', 'appicon.png')
app = QApplication(sys.argv)
window = Application('DCG', utils.qml_path(), icon_path)
window.setSource('main.qml')
window.setup_gui_and_connect_base_signals()
window.load_state()
window.show()
app.exec()
window.save_state()