test_loader = DataLoader(VideoSortingPerFrameTrainDataSet(
frame_num=frame_num,
path_list=ucf101_test_path_load(args.dataset_path, args.test_label_path,
args.class_path),
interval_frame=frame_interval,
sorting_start_index=sorting_start_index,
sorting_end_index=sorting_end_index),
batch_size=batch_size,
shuffle=False)
train_iterate_len = len(train_loader)
test_iterate_len = len(test_loader)
# 初期設定
# resnet18を取得
Net = CNN_RNN(frame_num,
pretrained=args.use_pretrained_model,
bidirectional=args.use_bidirectional)
criterion = torch.nn.CrossEntropyLoss() # Loss関数を定義
optimizer = torch.optim.Adam(Net.parameters(),
lr=args.learning_rate) # 重み更新方法を定義
current_epoch = 0
# ログファイルの生成
if not args.no_reset_log_file:
with open(log_train_path, mode='w') as f:
f.write(
'epoch,loss,full_fit_accuracy,per_fit_accuracy,sorting_accuracy,time,learning_rate\n'
)
with open(log_test_path, mode='w') as f:
f.write(
'epoch,loss,full_fit_accuracy,per_fit_accuracy,sorting_accuracy,time,learning_rate\n'
frame_num=frame_num,
path_list=ucf101_train_path_load(args.dataset_path, args.train_label_path),
frame_interval=interval_frames),
batch_size=batch_size, shuffle=True)
test_loader = DataLoader(
VideoSortTrainDataSet(
frame_num=frame_num,
path_list=ucf101_test_path_load(args.dataset_path, args.test_label_path, args.class_path),
frame_interval=interval_frames),
batch_size=batch_size, shuffle=False)
train_iterate_len = len(train_loader)
test_iterate_len = len(test_loader)
# 初期設定
# resnet18を取得
Net = CNN_RNN(args.frame_num, pretrained=args.use_pretrained_model, bidirectional=args.use_bidirectional,
task='classification')
criterion = torch.nn.CrossEntropyLoss() # Loss関数を定義
optimizer = torch.optim.Adam(Net.parameters(), lr=args.learning_rate) # 重み更新方法を定義
current_epoch = 0
# ログファイルの生成
if not args.no_reset_log_file:
with open(log_train_path, mode='w') as f:
f.write('epoch,loss,full_fit_accuracy,per_fit_accuracy,time,learning_rate\n')
with open(log_test_path, mode='w') as f:
f.write('epoch,loss,full_fit_accuracy,per_fit_accuracy,time,learning_rate\n')
# CUDA環境の有無で処理を変更
if args.use_cuda:
criterion = criterion.cuda()
Net = torch.nn.DataParallel(Net.cuda())