def get_data(name, meta_dir, gpu_nums):
isTrain = name == 'train'
ds = Cityscapes(meta_dir, name, shuffle=True)
if isTrain: #special augmentation
shape_aug = [
RandomResize(xrange=(0.7, 1.5),
yrange=(0.7, 1.5),
aspect_ratio_thres=0.15),
RandomCropWithPadding(args.crop_size, IGNORE_LABEL),
Flip(horiz=True),
]
else:
shape_aug = []
ds = AugmentImageComponents(ds, shape_aug, (0, 1), copy=False)
def f(ds):
image, label = ds
m = np.array([104, 116, 122])
const_arr = np.resize(m, (1, 1, 3)) # NCHW
image = image - const_arr
return image, label
ds = MapData(ds, f)
if isTrain:
ds = BatchData(ds, args.batch_size * gpu_nums)
ds = PrefetchDataZMQ(ds, 3)
else:
ds = BatchData(ds, 1)
return ds
def train_net(args, ctx):
logger.auto_set_dir()
from symbols.symbol_resnet_deeplabv2 import resnet101_deeplab_new
sym_instance = resnet101_deeplab_new()
sym = sym_instance.get_symbol(NUM_CLASSES,
is_train=True,
use_global_stats=True)
eval_sym_instance = resnet101_deeplab_new()
eval_sym = eval_sym_instance.get_symbol(NUM_CLASSES,
is_train=False,
use_global_stats=True)
# setup multi-gpu
gpu_nums = len(ctx)
input_batch_size = args.batch_size * gpu_nums
train_data = get_data("train", LIST_DIR, len(ctx))
test_data = get_data("val", LIST_DIR, len(ctx))
# infer shape
data_shape_dict = {
'data': (args.batch_size, 3, args.crop_size[0], args.crop_size[1]),
'label': (args.batch_size, 1, args.crop_size[0], args.crop_size[1])
}
pprint.pprint(data_shape_dict)
sym_instance.infer_shape(data_shape_dict)
# load and initialize params
epoch_string = args.load.rsplit("-", 2)[1]
begin_epoch = 1
if not args.scratch:
begin_epoch = int(epoch_string)
logger.info('continue training from {}'.format(begin_epoch))
arg_params, aux_params = load_init_param(args.load, convert=True)
else:
logger.info(args.load)
arg_params, aux_params = load_init_param(args.load, convert=True)
sym_instance.init_weights(arg_params, aux_params)
# check parameter shapes
sym_instance.check_parameter_shapes(arg_params, aux_params,
data_shape_dict)
data_names = ['data']
label_names = ['label']
mod = MutableModule(sym,
data_names=data_names,
label_names=label_names,
context=ctx,
fixed_param_prefix=fixed_param_prefix)
# decide training params
# metric
fcn_loss_metric = metric.FCNLogLossMetric(args.frequent,
Cityscapes.class_num())
eval_metrics = mx.metric.CompositeEvalMetric()
# callback
batch_end_callbacks = [
callback.Speedometer(input_batch_size, frequent=args.frequent)
]
#batch_end_callbacks = [mx.callback.ProgressBar(total=train_data.size/train_data.batch_size)]
epoch_end_callbacks = \
[mx.callback.module_checkpoint(mod, os.path.join(logger.get_logger_dir(),"mxnetgo"), period=1, save_optimizer_states=True),
]
lr_scheduler = StepScheduler(train_data.size() * EPOCH_SCALE, lr_step_list)
# optimizer
optimizer_params = {
'momentum': 0.9,
'wd': 0.0005,
'learning_rate': 2.5e-4,
'lr_scheduler': lr_scheduler,
'rescale_grad': 1.0,
'clip_gradient': None
}
logger.info("epoch scale = {}".format(EPOCH_SCALE))
mod.fit(train_data=train_data,
args=args,
eval_sym=eval_sym,
eval_sym_instance=eval_sym_instance,
eval_data=test_data,
eval_metric=eval_metrics,
epoch_end_callback=epoch_end_callbacks,
batch_end_callback=batch_end_callbacks,
kvstore=kvstore,
optimizer='sgd',
optimizer_params=optimizer_params,
arg_params=arg_params,
aux_params=aux_params,
begin_epoch=begin_epoch,
num_epoch=end_epoch,
epoch_scale=EPOCH_SCALE,
validation_on_last=validation_on_last)
os.environ['PYTHONUNBUFFERED'] = '1'
os.environ['MXNET_CUDNN_AUTOTUNE_DEFAULT'] = '0'
os.environ['MXNET_ENABLE_GPU_P2P'] = '0'
IGNORE_LABEL = 255
CROP_HEIGHT = 672
CROP_WIDTH = 672
tile_height = 1024
tile_width = 1024
batch_size = 7
EPOCH_SCALE = 4
end_epoch = 10
lr_step_list = [(6, 1e-3), (10, 1e-4)]
NUM_CLASSES = Cityscapes.class_num()
validation_on_last = 2
kvstore = "device"
fixed_param_prefix = ["conv1", "bn_conv1", "res2", "bn2", "gamma", "beta"]
symbol_str = "resnet_v1_101_deeplab"
def parse_args():
parser = argparse.ArgumentParser(description='Train deeplab network')
parser.add_argument("--gpu", default="5")
parser.add_argument('--frequent',
help='frequency of logging',
default=800,
type=int)
parser.add_argument('--view', action='store_true')
parser.add_argument("--validation", action="store_true")
def get_data(name, data_dir, meta_dir, config):
ds = Cityscapes(data_dir, meta_dir, name, shuffle=False)
ds = BatchData(ds, 1)
return ds
