def get_models_to_evaluate():
"""
Returns the models to evaluate based on what is in logdir and modeldir
specified as command line arguments. The matching pattern is specified by
the match argument.
Returns: [(log_dir, model_dir, source, target, model_name, method_name), ...]
"""
files = pathlib.Path(FLAGS.logdir).glob(FLAGS.match)
models_to_evaluate = []
for log_dir in files:
items = str(log_dir.stem).split("-")
assert len(items) >= 3 or len(items) <= 5, \
"name should be one of source-target-model-{-method{-num,},-num,}"
method_name = "none"
if len(items) == 3:
source, target, model_name = items
elif len(items) == 4 or len(items) == 5:
source, target, model_name, keyword = items[:4]
if keyword in methods:
method_name = keyword
else:
pass # probably a debug number, which we don't care about
model_dir = os.path.join(FLAGS.modeldir, log_dir.stem)
assert os.path.exists(model_dir), "Model does not exist "+str(model_dir)
assert model_name in models.names(), "Unknown model "+str(model_name)
assert source in load_datasets.names(), "Unknown source "+str(source)
assert target in [""]+load_datasets.names(), "Unknown target "+str(target)
models_to_evaluate.append((str(log_dir), model_dir, source, target,
model_name, method_name))
return models_to_evaluate
if __name__ == '__main__':
parser = argparse.ArgumentParser(description="Network evaluation example")
# data
parser.add_argument('-b', '--batch-size', type=int, default=64)
parser.add_argument('-j', '--workers', type=int, default=4)
parser.add_argument('--height', type=int, default=224)
parser.add_argument('--width', type=int, default=224)
parser.add_argument('--seed', type=int, default=1)
parser.add_argument('--topk', '-t', type=int, default=5)
# model
parser.add_argument('-a',
'--arch',
type=str,
default='resnet50',
choices=models.names())
parser.add_argument('--weights',
'-w',
type=str,
metavar='PATH',
help='path to the checkpoint')
parser.add_argument('--ann_file',
type=str,
metavar='PATH',
help="path to the annotation file")
parser.add_argument('--data_dir',
type=str,
metavar='PATH',
help="path to the data folder")
parser.add_argument('--print-freq',
'-p',
from absl import app
from absl import flags
from absl import logging
import models
import load_datasets
from metrics import Metrics
from checkpoints import CheckpointManager
from file_utils import last_modified_number, write_finished
from gpu_memory import set_gpu_memory
FLAGS = flags.FLAGS
flags.DEFINE_enum("model", None, models.names(), "What model type to use")
flags.DEFINE_string("modeldir", "models", "Directory for saving model files")
flags.DEFINE_string("logdir", "logs", "Directory for saving log files")
flags.DEFINE_enum("method", None, ["none", "dann", "pseudo", "instance"],
"What method of domain adaptation to perform (or none)")
flags.DEFINE_enum("source", None, load_datasets.names(),
"What dataset to use as the source")
flags.DEFINE_enum("target", "", [""] + load_datasets.names(),
"What dataset to use as the target")
flags.DEFINE_integer("steps", 80000, "Number of training steps to run")
flags.DEFINE_float("lr", 0.001, "Learning rate for training")
flags.DEFINE_float("lr_domain_mult", 1.0,
"Learning rate multiplier for training domain classifier")
flags.DEFINE_float("lr_target_mult", 0.5,
"Learning rate multiplier for training target classifier")
flags.DEFINE_float("gpumem", 3350,
