mem_size = 512
#model_state_dict = '/content/drive/My Drive/Lorenzo/ego-rnn-two-in-one/results/16frames_two_in_one/rgb_16frames_two_in_one_3/model_rgb_state_dict.pth' # Weights of the pre-trained model
model_state_dict = "/content/drive/My Drive/Lorenzo/ego-rnn-two-in-one/results/rgb_16frames_two_in_one_conv1_unfreeze_3stages_kl/model_rgb_state_dict.pth"
model = attentionModel(num_classes=num_classes, mem_size=mem_size)
model.load_state_dict(torch.load(model_state_dict), strict=True)
model.train(False)
for params in model.parameters():
params.requires_grad = False
model = model.to(DEVICE)
normalize = transforms.Normalize(
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]
)
preprocess = transforms.Compose([
transforms.Scale(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
normalize])
in_path = "/content/drive/My Drive/Lorenzo/ego-rnn-two-in-one/get_modulation/frames"
out_path = "/content/drive/My Drive/Lorenzo/ego-rnn-two-in-one/get_modulation/out2"
inputFlow = []
inputFrame = []
frames = []
args.scales, args.sample_size, crop_positions=['c'])
before_crop_duration = int(args.sample_duration * args.downsample)
if args.mode == 'train':
temporal_transform = TemporalSequentialCrop(before_crop_duration,
args.downsample)
if args.view == 'front_depth' or args.view == 'front_IR':
spatial_transform = spatial_transforms.Compose([
crop_method,
spatial_transforms.RandomRotate(),
spatial_transforms.SaltImage(),
spatial_transforms.Dropout(),
spatial_transforms.ToTensor(args.norm_value),
spatial_transforms.Normalize([0], [1])
])
elif args.view == 'top_depth' or args.view == 'top_IR':
spatial_transform = spatial_transforms.Compose([
spatial_transforms.RandomHorizontalFlip(),
spatial_transforms.Scale(args.sample_size),
spatial_transforms.CenterCrop(args.sample_size),
spatial_transforms.RandomRotate(),
spatial_transforms.SaltImage(),
spatial_transforms.Dropout(),
spatial_transforms.ToTensor(args.norm_value),
spatial_transforms.Normalize([0], [1])
])
print(
"=================================Loading Anormal-Driving Training Data!================================="
def main():
torch.manual_seed(args.seed)
os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu_devices
use_gpu = torch.cuda.is_available()
if not args.evaluate:
sys.stdout = Logger(osp.join(args.save_dir, 'log_train.txt'), mode='a')
else:
sys.stdout = Logger(osp.join(args.save_dir, 'log_test.txt'), mode='a')
print("==========\nArgs:{}\n==========".format(args))
if use_gpu:
print("Currently using GPU {}".format(args.gpu_devices))
cudnn.benchmark = True
torch.cuda.manual_seed_all(args.seed)
else:
print("Currently using CPU (GPU is highly recommended)")
print("Initializing dataset {}".format(args.dataset))
dataset = data_manager.init_imgreid_dataset(name=args.dataset,
dataset_dir=args.root,
fore_dir=args.fore_dir)
transform_train = ST.Compose([
ST.Scale((args.height, args.width), interpolation=3),
ST.RandomHorizontalFlip(),
ST.ToTensor(),
ST.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
ST.RandomErasing(0.5)
])
transform_test = ST.Compose([
ST.Scale((args.height, args.width), interpolation=3),
ST.ToTensor(),
ST.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])
pin_memory = True if use_gpu else False
trainloader = DataLoader(
ImageDataset_hardSplit_seg(dataset.train, transform=transform_train),
sampler=RandomIdentitySampler(dataset.train,
num_instances=args.num_instances),
batch_size=args.train_batch,
num_workers=args.workers,
pin_memory=pin_memory,
drop_last=True,
)
queryloader = DataLoader(
ImageDataset(dataset.query, transform=transform_test),
batch_size=args.test_batch,
shuffle=False,
num_workers=args.workers,
pin_memory=pin_memory,
drop_last=False,
)
galleryloader = DataLoader(
ImageDataset(dataset.gallery, transform=transform_test),
batch_size=args.test_batch,
shuffle=False,
num_workers=args.workers,
pin_memory=pin_memory,
drop_last=False,
)
print("Initializing model: {}".format(args.arch))
model = models.init_model(name=args.arch,
num_classes=dataset.num_train_pids)
print(model)
criterion_xent = CrossEntropyLabelSmooth(use_gpu=use_gpu)
criterion_htri = TripletLoss()
criterion_mask = MaskLoss()
criterion_split = HardSplitLoss()
criterion_cluster = ClusterLoss()
optimizer = init_optim(args.optim, model.parameters(), args.lr,
args.weight_decay)
scheduler = lr_scheduler.MultiStepLR(optimizer,
milestones=args.stepsize,
gamma=args.gamma)
if args.resume:
if check_isfile(args.resume):
checkpoint = torch.load(args.resume)
model.load_state_dict(checkpoint['state_dict'])
args.start_epoch = checkpoint['epoch']
rank1 = checkpoint['rank1']
print("Loaded checkpoint from '{}'".format(args.resume))
print("- start_epoch: {}\n- rank1: {}".format(
args.start_epoch, rank1))
if use_gpu:
model = nn.DataParallel(model).cuda()
if args.evaluate:
print("Evaluate only")
test(model, queryloader, galleryloader, use_gpu)
return
start_time = time.time()
train_time = 0
best_rank1 = -np.inf
best_epoch = 0
print("==> Start training")
for epoch in range(args.start_epoch, args.max_epoch):
start_train_time = time.time()
train(epoch, model, criterion_xent, criterion_htri, criterion_mask,
criterion_split, criterion_cluster, optimizer, trainloader,
use_gpu)
train_time += round(time.time() - start_train_time)
scheduler.step()
if (epoch + 1) > args.start_eval and (
epoch + 1) % args.eval_step == 0 or epoch == 0:
print("==> Test")
rank1 = test(model, queryloader, galleryloader, use_gpu)
is_best = rank1 > best_rank1
if is_best:
best_rank1 = rank1
best_epoch = epoch + 1
if use_gpu:
state_dict = model.module.state_dict()
else:
state_dict = model.state_dict()
save_checkpoint(
{
'state_dict': state_dict,
'rank1': rank1,
'epoch': epoch,
}, is_best,
osp.join(args.save_dir,
'checkpoint_ep' + str(epoch + 1) + '.pth.tar'))
print("==> Best Rank-1 {:.1%}, achieved at epoch {}".format(
best_rank1, best_epoch))
elapsed = round(time.time() - start_time)
elapsed = str(datetime.timedelta(seconds=elapsed))
train_time = str(datetime.timedelta(seconds=train_time))
print(
"Finished. Total elapsed time (h:m:s): {}. Training time (h:m:s): {}.".
format(elapsed, train_time))
print("==========\nArgs:{}\n==========".format(args))
from tqdm import tqdm
import os
import random
# In[2]:
norm_value = 1
# ds_mean = [114.7748 / norm_value, 107.7354 / norm_value, 99.4750 / norm_value]
ds_mean = [90.0, 98.0, 102.0]
ds_std = [38.7568578 / norm_value, 37.88248729 / norm_value,40.02898126 / norm_value]
st = spatial_transforms.Compose([
spatial_transforms.Scale(112),
spatial_transforms.CornerCrop(112, 'c'),
spatial_transforms.ToTensor(1),
spatial_transforms.Normalize(ds_mean, [1,1,1]) #Hara et. al. normalised by mean only
# spatial_transforms.Normalize(ds_mean, ds_std)
])
tt = temporal_transforms.LoopPadding(16)
ds = ucf101.UCF101(
root_path="/media/datasets/Video/UCF-101/jpg",
annotation_path="/media/datasets/Video/UCF-101/ucf101_01.json",
subset="validation",
n_samples_for_each_video=0,
spatial_transform=st,
temporal_transform=tt
)
# In[3]: