if __name__ == '__main__':
#################
# ArgumentParser
#################
parser = argparse.ArgumentParser(description="YOLO-V3 training procedure.")
parser.add_argument("--config_project",
type=str,
default="./data/config/config_project.yaml",
help="项目配置文件.")
parser.add_argument("--config_common",
type=str,
default="./data/config/config_common.yaml",
help="默认配置文件.")
args = parser.parse_args()
config_common = get_config(args.config_common)
config_project = get_config(args.config_project).as_dict()
config_common.update(config_project)
#创建部分存储路径
if not os.path.exists(config_common.data_set["log_save_path"]):
try:
os.makedirs(config_common.data_set["log_save_path"])
print("make log save dir:%s" %
config_common.data_set["log_save_path"])
except Exception as ex:
print("make log save dir:%s, failed, cause of %s" %
(config_common.data_set["log_save_path"], ex))
sys.exit(1)
#创建日志文件
logger = Logger(config_common.data_set["log_save_path"], "detection_train")
def demo_cnn(root_path, audio_path):
# Extract feature data
def main(cfgs):
global logger
logger = setup_logging(cfgs.CNN.LOG_PATH)
if cfgs.CNN.DISABLE_LOG == True:
logger.disabled = False;
write_config_to_log(cfgs, logger)
train(cfgs)
def write_config_to_log(cfgs, logger):
logger.info("{} CONFIGS {}".format('_'*30, "_"*30))
logger.info("{} Split dataset train - val : {}-{}".format('\t', 1- cfgs.CNN.SPLIT, cfgs.CNN.SPLIT ))
logger.info("{} ROOT: ".format('\t', cfgs.CNN.ROOT ))
logger.info("{} DATASET_NAME name: {}".format('\t', cfgs.CNN.DATASET_NAME ))
logger.info("{} FEATURES: {}".format('\t', cfgs.CNN.FEATURES ))
logger.info("{} OUTPUT_PATH: {}".format('\t', cfgs.CNN.OUTPUT_PATH ))
logger.info("{} WEIGHTS_PATH: {}".format('\t', cfgs.CNN.WEIGHTS_PATH ))
logger.info("{} ARCHITECTURE_MODEL_PATH: {}".format('\t', cfgs.CNN.ARCHITECTURE_MODEL_PATH ))
logger.info("{} VISUALIZE_LOSS_PATH: {}".format('\t', cfgs.CNN.VISUALIZE_LOSS_PATH ))
logger.info("{} DISABLE LOG: {}".format('\t', cfgs.CNN.DISABLE_LOG))
logger.info("{} LOG_PATH: {}".format('\t', cfgs.CNN.LOG_PATH))
logger.info("{} MODEL: ".format('\t' ))
logger.info("{} INPUT_SIZE: {}".format('\t\t', cfgs.CNN.MODEL.INPUT_SIZE))
logger.info("{} OUTPUT_SIZE: {}".format('\t\t', cfgs.CNN.MODEL.OUTPUT_SIZE))
logger.info("{} ACTIVATION: {}".format('\t\t', cfgs.CNN.MODEL.ACTIVATION))
logger.info("{} PADDING: {}".format('\t\t', cfgs.CNN.MODEL.PADDING))
logger.info("{} LEARNING_RATE: {}".format('\t\t', cfgs.CNN.MODEL.LEARNING_RATE))
logger.info("{} DE_CAY: {}".format('\t\t', cfgs.CNN.MODEL.DE_CAY))
logger.info("{} LOSS_FUNCTION: {}".format('\t\t', cfgs.CNN.MODEL.LOSS_FUNCTION))
logger.info("{} METRIC: {}".format('\t\t', cfgs.CNN.MODEL.METRIC))
logger.info("{} BATCH_SIZE: {}".format('\t\t', cfgs.CNN.MODEL.BATCH_SIZE))
logger.info("{} EPOCHS: {}".format('\t\t', cfgs.CNN.MODEL.EPOCHS))
print("{}\n".format('_'*100))
def arg_parser():
parser = argparse.ArgumentParser(description='Argument of extract data')
parser.add_argument('--root', '-r', required=True,
type=str, help='The path consists all type extracted features')
parser.add_argument('--data_n', '-n', default='all',
type=str, help='name of dataset you want to train')
# parser.add_argument('--features', '-f', default="all",
# type=str, help='The type method that you want to extract')
parser.add_argument('--argu', '-a', action='store_true',
help='Augment data')
parser.add_argument('--output', '-o', required=True,
type=str, help='The path will save model')
parser.add_argument('--architecture_model', '-ar', action='store_true',
help='Save architecture model')
parser.add_argument('--visualize_loss', '-v', action='store_true',
help='Allow visualize loss')
parser.add_argument('--dis_log', '-dl', action='store_true',
help='Disable write log')
parser.add_argument('--configs_file', '-c', required=True,
type=str, help='The path config file for model')
parser.add_argument(
"--opts",
help="Argument of Dict guided method, modify config options using the command-line 'KEY VALUE' pairs",
default=[],
nargs=argparse.REMAINDER,
)
return parser.parse_args()
def print_args(args):
print("{} PARSER {}".format('_'*30, "_"*30))
print("{} The path root of exacted features: \n\t{}".format('#'*3, args.root))
print("{} Dataset name to train: \n\t{}".format('#'*3, args.data_n))
# print("{} The methods exacted feature: \n\t{}".format('#'*3, args.features))
print("{} Augment data: \n\t{}".format('#'*3, args.argu))
print("{} Dataset name to train: \n\t{}".format('#'*3, args.data_n))
print("{} The path of output : \n\t{}".format('#'*3, args.output))
print("{} Allow save architecture model : \n\t{}".format('#'*3, args.architecture_model))
print("{} Allow visualize loss log : \n\t{}".format('#'*3, args.visualize_loss))
print("{} Disable write log file : \n\t{}".format('#'*3, args.dis_log))
print("{} The path of configs file : \n\t{}".format('#'*3, args.configs_file))
def print_cfgs(cfgs):
print("{} CONFIGS {}".format('_'*30, "_"*30))
print("{} Split dataset train - val : {}-{}".format('\t', 1- cfgs.CNN.SPLIT, cfgs.CNN.SPLIT ))
print("{} ROOT: ".format('\t', cfgs.CNN.ROOT ))
print("{} DATASET_NAME name: {}".format('\t', cfgs.CNN.DATASET_NAME ))
print("{} FEATURES: {}".format('\t', cfgs.CNN.FEATURES ))
print("{} OUTPUT_PATH: {}".format('\t', cfgs.CNN.OUTPUT_PATH ))
print("{} WEIGHTS_PATH: {}".format('\t', cfgs.CNN.WEIGHTS_PATH ))
print("{} ARCHITECTURE_MODEL_PATH: {}".format('\t', cfgs.CNN.ARCHITECTURE_MODEL_PATH ))
print("{} VISUALIZE_LOSS_PATH: {}".format('\t', cfgs.CNN.VISUALIZE_LOSS_PATH ))
print("{} DISABLE_LOG: {}".format('\t', cfgs.CNN.DISABLE_LOG ))
print("{} MODEL: ".format('\t' ))
print("{} INPUT_SIZE: {}".format('\t\t', cfgs.CNN.MODEL.INPUT_SIZE))
print("{} OUTPUT_SIZE: {}".format('\t\t', cfgs.CNN.MODEL.OUTPUT_SIZE))
print("{} ACTIVATION: {}".format('\t\t', cfgs.CNN.MODEL.ACTIVATION))
print("{} PADDING: {}".format('\t\t', cfgs.CNN.MODEL.PADDING))
print("{} LEARNING_RATE: {}".format('\t\t', cfgs.CNN.MODEL.LEARNING_RATE))
print("{} DE_CAY: {}".format('\t\t', cfgs.CNN.MODEL.DE_CAY))
print("{} LOSS_FUNCTION: {}".format('\t\t', cfgs.CNN.MODEL.LOSS_FUNCTION))
print("{} METRIC: {}".format('\t\t', cfgs.CNN.MODEL.METRIC))
print("{} BATCH_SIZE: {}".format('\t\t', cfgs.CNN.MODEL.BATCH_SIZE))
print("{} EPOCHS: {}".format('\t\t', cfgs.CNN.MODEL.EPOCHS))
print("{}\n".format('_'*100))
def setup_config(cfgs, args):
cfgs.CNN.ROOT = args.root
cfgs.CNN.DATASET_NAME = args.data_n
cfgs.CNN.ARGUMENT = args.argu
cfgs.CNN.OUTPUT_PATH = args.output
cfgs.CNN.WEIGHTS_PATH = os.path.join(args.output, "model.h5")
if args.architecture_model == True:
cfgs.CNN.ARCHITECTURE_MODEL_PATH = os.path.join(args.output, "architecture_model.json")
if args.visualize_loss == True:
cfgs.CNN.VISUALIZE_LOSS_PATH = os.path.join(args.output, "loss.png")
cfgs.CNN.DISABLE_LOG = args.dis_log
cfgs.CNN.LOG_PATH = os.path.join(args.output, "log.log")
return cfgs
if __name__ == "__main__":
args = arg_parser()
cfgs = get_config()
cfgs.merge_from_file('./configs/CNN/CNN.yaml')
if not os.path.isfile(args.configs_file):
print("File configs not exits!!")
# old_stdout = sys.stdout
# log_file = open(os.path.join(args.output, 'log.log'),"w")
# sys.stdout = log_file
cfgs.merge_from_file(args.configs_file)
print_args(args)
cfgs = setup_config(cfgs, args)
print_cfgs(cfgs)
main(cfgs)
'#' * 3, args.features))
print("{} Augment data: \n\t{}".format('#' * 3, args.argu))
print("{} Dataset name to train: \n\t{}".format('#' * 3, args.data_n))
print("{} The path of model ouput : \n\t{}".format('#' * 3, args.output))
print("{} The path of configs file : \n\t{}".format(
'#' * 3, args.configs_file))
print("{} CONFIGS {}".format('_' * 30, "_" * 30))
print("{} Split dataset train - val : {}-{}".format(
'#' * 3, 1 - cfgs.DS.SPLIT, cfgs.DS.SPLIT))
print("{}\n".format('_' * 100))
if __name__ == "__main__":
args = arg_parser()
cfgs = get_config()
cfgs.merge_from_file('./configs/BASE/Base.yaml')
if not os.path.isfile(args.configs_file):
print("File configs not exits!!")
# old_stdout = sys.stdout
# log_file = open(os.path.join(args.output, 'log.log'),"w")
# sys.stdout = log_file
cfgs.merge_from_file(args.configs_file)
print_args_cfg(args, cfgs)
main(args, cfgs)
# sys.stdout = old_stdout
# log_file.close()
'''
-r E:\Courses\Recognition\Final_Project\Pattern_Recognition_Final_Project\feature_data \
-n TESS \
import os
import yaml
from easydict import EasyDict as edict
from utils.config_parse import get_config
cfg = get_config()
cfg.merge_from_file(
"E:\Courses\Recognition\Final_Project\Pattern_Recognition_Final_Project\configs\methodA\A.yaml"
)
cfg.merge_from_file(
"E:\Courses\Recognition\Final_Project\Pattern_Recognition_Final_Project\configs\methodA\AA.yaml"
)
dic = {'TRI': "2", 'NGON': "4"}
cfg.merge_from_dict(dic)
print('config A: ', cfg.A.NAME)
print("config B: ", cfg.AA.INFO)
print('config dic: ', cfg.TRI)