train_net(args.network,
args.train_path,
args.num_class,
args.batch_size,
args.data_shape, [args.mean_r, args.mean_g, args.mean_b],
args.resume,
args.finetune,
args.pretrained,
args.epoch,
args.prefix,
ctx,
args.begin_epoch,
args.end_epoch,
args.frequent,
args.learning_rate,
args.momentum,
args.weight_decay,
args.lr_refactor_step,
args.lr_refactor_ratio,
val_path=args.val_path,
num_example=args.num_example,
class_names=class_names,
label_pad_width=args.label_width,
freeze_layer_pattern=args.freeze_pattern,
iter_monitor=args.monitor,
monitor_pattern=args.monitor_pattern,
log_file=args.log_file,
nms_thresh=args.nms_thresh,
force_nms=args.force_nms,
ovp_thresh=args.overlap_thresh,
use_difficult=args.use_difficult,
voc07_metric=args.use_voc07_metric)
type=float,
default=0.9,
help='ratio to refactor learning rate')
parser.add_argument('--log',
dest='log_file',
type=str,
default="train.log",
help='save training log to file')
parser.add_argument(
'--monitor',
dest='monitor',
type=int,
default=0,
help='log network parameters every N iters if larger than 0')
args = parser.parse_args()
return args
if __name__ == '__main__':
args = parse_args()
ctx = [mx.gpu(int(i)) for i in args.gpus.split(',')]
ctx = mx.cpu() if not ctx else ctx
train_net(args.network, args.dataset, args.image_set, args.year,
args.devkit_path, args.batch_size, args.data_shape,
[args.mean_r, args.mean_g, args.mean_b], args.resume,
args.finetune, args.pretrained, args.epoch, args.prefix, ctx,
args.begin_epoch, args.end_epoch, args.frequent,
args.learning_rate, args.momentum, args.weight_decay,
args.val_image_set, args.val_year, args.lr_refactor_epoch,
args.lr_refactor_ratio, args.monitor, args.log_file)
help='weight decay')
parser.add_argument('--mean-r', dest='mean_r', type=float, default=123,
help='red mean value')
parser.add_argument('--mean-g', dest='mean_g', type=float, default=117,
help='green mean value')
parser.add_argument('--mean-b', dest='mean_b', type=float, default=104,
help='blue mean value')
parser.add_argument('--lr-epoch', dest='lr_refactor_epoch', type=int, default=50,
help='refactor learning rate every N epoch')
parser.add_argument('--lr-ratio', dest='lr_refactor_ratio', type=float, default=0.9,
help='ratio to refactor learning rate')
parser.add_argument('--log', dest='log_file', type=str, default="train.log",
help='save training log to file')
parser.add_argument('--monitor', dest='monitor', type=int, default=0,
help='log network parameters every N iters if larger than 0')
args = parser.parse_args()
return args
if __name__ == '__main__':
args = parse_args()
ctx = [mx.gpu(int(i)) for i in args.gpus.split(',')]
ctx = mx.cpu() if not ctx else ctx
train_net(args.network, args.dataset, args.image_set, args.year,
args.devkit_path, args.batch_size,
args.data_shape, [args.mean_r, args.mean_g, args.mean_b],
args.resume, args.finetune, args.pretrained,
args.epoch, args.prefix, ctx, args.begin_epoch, args.end_epoch,
args.frequent, args.learning_rate, args.momentum, args.weight_decay,
args.val_image_set, args.val_year, args.lr_refactor_epoch,
args.lr_refactor_ratio, args.monitor, args.log_file)
return class_names
if __name__ == '__main__':
args = parse_args()
# context list
ctx = [mx.gpu(int(i)) for i in args.gpus.split(',') if i.strip()]
ctx = [mx.cpu()] if not ctx else ctx
# class names if applicable
class_names = parse_class_names(args)
# start training
train_net(args.network, args.train_path,
args.num_class, args.batch_size,
args.data_shape, [args.mean_r, args.mean_g, args.mean_b],
args.resume, args.finetune, args.pretrained,
args.epoch, args.prefix, ctx, args.begin_epoch, args.end_epoch,
args.frequent, args.learning_rate, args.momentum, args.weight_decay,
args.lr_refactor_step, args.lr_refactor_ratio,
val_path=args.val_path,
num_example=args.num_example,
class_names=class_names,
label_pad_width=args.label_width,
freeze_layer_pattern=args.freeze_pattern,
iter_monitor=args.monitor,
monitor_pattern=args.monitor_pattern,
log_file=args.log_file,
nms_thresh=args.nms_thresh,
force_nms=args.force_nms,
ovp_thresh=args.overlap_thresh,
use_difficult=args.use_difficult,
voc07_metric=args.use_voc07_metric)
from train.train_net import train_net
#import tools.find_mxnet
import mxnet as mx
import os
import sys
import opt
if __name__ == '__main__':
train_net()
