def main_merge():
global args, best_corr
args.store_name = '{}_merged'.format(args.model)
args.store_name = args.store_name + datetime.now().strftime('_%m-%d_%H-%M')
args.start_epoch = 0
check_rootfolders(args)
model = Baseline(args.img_feat_size, args.au_feat_size)
model = torch.nn.DataParallel(model).cuda()
optimizer = torch.optim.Adam(model.parameters(), lr=args.learning_rate)
if args.use_multistep:
scheduler = torch.optim.lr_scheduler.MultiStepLR(
optimizer, args.step_milestones, args.step_decay)
# ckpt structure {epoch, state_dict, optimizer, best_corr}
if args.resume and os.path.isfile(args.resume):
print('Load checkpoint:', args.resume)
ckpt = torch.load(args.resume)
args.start_epoch = ckpt['epoch']
best_corr = ckpt['best_corr']
model.load_state_dict(ckpt['state_dict'])
optimizer.load_state_dict(ckpt['optimizer'])
print('Loaded ckpt at epoch:', args.start_epoch)
# initialize datasets
train_loader = torch.utils.data.DataLoader(dataset=EEV_Dataset(
csv_path=[args.train_csv, args.val_csv],
vidmap_path=[args.train_vidmap, args.val_vidmap],
image_feat_path=args.image_features,
audio_feat_path=args.audio_features,
mode='merge'),
batch_size=args.batch_size,
shuffle=True,
num_workers=args.workers,
pin_memory=False,
drop_last=True)
log_training = open(
os.path.join(args.root_log, args.store_name, 'log.csv'), 'w')
with open(os.path.join(args.root_log, args.store_name, 'args.txt'),
'w') as f:
f.write(str(args))
tb_writer = SummaryWriter(
log_dir=os.path.join(args.root_log, args.store_name))
for epoch in range(args.start_epoch, args.epochs):
train(train_loader, model, optimizer, epoch, log_training, tb_writer)
save_checkpoint(
{
'epoch': epoch + 1,
'state_dict': model.state_dict(),
'optimizer': optimizer.state_dict(),
'best_corr': 0.0,
}, False)
if args.use_multistep:
scheduler.step()
def main():
net = Baseline(num_classes=culane.num_classes, deep_base=args['deep_base']).cuda()
print('load checkpoint \'%s.pth\' for evaluation' % args['checkpoint'])
pretrained_dict = torch.load(os.path.join(ckpt_path, exp_name, args['checkpoint'] + '_checkpoint.pth'))
pretrained_dict = {k[7:]: v for k, v in pretrained_dict.items()}
net.load_state_dict(pretrained_dict)
net.eval()
save_dir = os.path.join(ckpt_path, exp_name, 'vis_%s_test' % args['checkpoint'])
check_mkdir(save_dir)
log_path = os.path.join(save_dir, str(datetime.datetime.now()) + '.log')
data_list = [l.strip('\n') for l in open(os.path.join(culane.root, culane.list, 'test_gt.txt'), 'r')]
loss_record = AverageMeter()
gt_all, prediction_all=[], []
for idx in range(len(data_list)):
print('evaluating %d / %d' % (idx + 1, len(data_list)))
img = Image.open(culane.root + data_list[idx].split(' ')[0]).convert('RGB')
gt = Image.open(culane.root + data_list[idx].split(' ')[1])
img, gt = val_joint_transform(img, gt)
with torch.no_grad():
img_var = Variable(img_transform(img).unsqueeze(0)).cuda()
gt_var = Variable(mask_transform(gt).unsqueeze(0)).cuda()
prediction = net(img_var)[0]
loss = criterion(prediction, gt_var)
loss_record.update(loss.data, 1)
scoremap = F.softmax(prediction, dim=1).data.squeeze().cpu().numpy()
prediction = prediction.data.max(1)[1].squeeze().cpu().numpy().astype(np.uint8)
prediction_all.append(prediction)
gt_all.append(np.array(gt))
if args['save_results']:
check_mkdir(save_dir + data_list[idx].split(' ')[0][:-10])
out_file = open(os.path.join(save_dir, data_list[idx].split(' ')[0][1:-4] + '.lines.txt'), 'w')
prob2lines(scoremap, out_file)
acc, acc_cls, mean_iu, fwavacc = evaluation(prediction_all, gt_all, culane.num_classes)
log = 'val results: loss %.5f acc %.5f acc_cls %.5f mean_iu %.5f fwavacc %.5f' % \
(loss_record.avg, acc, acc_cls, mean_iu, fwavacc)
print(log)
open(log_path, 'w').write(log + '\n')
def test():
# Prepare env
env = create_env()
h, w, c = env.observation_space.shape
# Load 5 best models
device = torch.device("cpu")
model_dir = "./policy_grad"
model_fns = {}
for fn in os.listdir(model_dir):
if fn.endswith('.pth'):
score = fn.split("_")[-1][:-4]
model_fns[fn] = float(score)
top_5 = heapq.nlargest(3, model_fns, key=model_fns.get)
models = []
for fn in top_5:
path = os.path.join(model_dir, fn)
model = Baseline(h, w).to(device)
model.load_state_dict(torch.load(path, map_location='cpu'))
model.eval()
models.append(model)
# Watch race car perform
state = env.reset().transpose((2, 0, 1))
state = torch.tensor([state], dtype=torch.float, device=device)
total_reward = 0
for t in count():
# Select and perform an action
votes = []
for model in models:
pi, _ = model(state)
votes.append(pi.argmax().item())
action_idx = Counter(votes).most_common(1)[0][0]
action = index_to_action(action_idx)
state, reward, done, _ = env.step(action)
env.render()
# Update
state = state.transpose((2, 0, 1))
state = torch.tensor([state], dtype=torch.float, device=device)
total_reward += reward
if done:
break
print("Total reward: {}".format(total_reward))
def main():
net = Baseline(num_classes=culane.num_classes,
deep_base=args['deep_base']).cuda().train()
net = DataParallelWithCallback(net)
optimizer = optim.SGD([{
'params': [
param
for name, param in net.named_parameters() if name[-4:] == 'bias'
],
'lr':
2 * args['base_lr']
}, {
'params': [
param
for name, param in net.named_parameters() if name[-4:] != 'bias'
],
'lr':
args['base_lr']
}],
momentum=args['momentum'])
if len(args['checkpoint']) > 0:
print('training resumes from \'%s\'' % args['checkpoint'])
net.load_state_dict(
torch.load(
os.path.join(ckpt_path, exp_name,
args['checkpoint'] + '_checkpoint.pth')))
optimizer.load_state_dict(
torch.load(
os.path.join(ckpt_path, exp_name,
args['checkpoint'] + '_checkpoint_optim.pth')))
optimizer.param_groups[0]['lr'] = 2 * args['base_lr']
optimizer.param_groups[1]['lr'] = args['base_lr']
check_mkdir(os.path.join(ckpt_path, exp_name))
open(log_path, 'w').write(str(args) + '\n\n')
train(net, optimizer)
def main_test():
print('Running test...')
torch.multiprocessing.set_sharing_strategy('file_system')
model = Baseline()
if args.use_swa:
model = torch.optim.swa_utils.AveragedModel(model)
model = torch.nn.DataParallel(model).cuda()
# ckpt structure {epoch, state_dict, optimizer, best_corr}
if args.resume and os.path.isfile(args.resume):
print('Load checkpoint:', args.resume)
ckpt = torch.load(args.resume)
args.start_epoch = ckpt['epoch']
best_corr = ckpt['best_corr']
model.load_state_dict(ckpt['state_dict'])
print('Loaded ckpt at epoch:', args.start_epoch)
else:
print('No model given. Abort!')
exit(1)
test_loader = torch.utils.data.DataLoader(
dataset=EEV_Dataset(
csv_path=None,
vidmap_path=args.test_vidmap,
image_feat_path=args.image_features,
audio_feat_path=args.audio_features,
mode='test',
test_freq=args.test_freq
),
batch_size=None, shuffle=False,
num_workers=args.workers, pin_memory=False
)
model.eval()
batch_time = AverageMeter()
t_start = time.time()
outputs = []
with torch.no_grad():
for i, (img_feat, au_feat, frame_count, vid) in enumerate(test_loader):
img_feat = torch.stack(img_feat).cuda()
au_feat = torch.stack(au_feat).cuda()
assert len(au_feat.size()) == 3, 'bad auf %s' % (vid)
output = model(img_feat, au_feat) # [Clip S 15]
# rearrange and remove extra padding in the end
output = rearrange(output, 'Clip S C -> (Clip S) C')
output = torch.cat([output, output[-1:]]) # repeat the last frame to avoid missing
if args.train_freq < args.test_freq:
# print('interpolating:', output.size()[0], frame_count)
output = interpolate_output(output, args.train_freq, 6)
# print('Interpolated:', output.size()[0], frame_count)
# truncate extra frames
assert output.size(0) >= frame_count, '{}/{}'.format(output.size(0), frame_count)
output = output[:frame_count]
outputs.append((vid, frame_count, output.cpu().detach().numpy()))
# update statistics
batch_time.update(time.time() - t_start)
t_start = time.time()
if i % args.print_freq == 0:
output = ('Test: [{0}/{1}]\t'
'Time: {batch_time.val:.3f} ({batch_time.avg:.3f})'.format(
i, len(test_loader), batch_time=batch_time))
print(output)
time_stamps = [0, 166666, 333333, 500000, 666666, 833333]
time_step = 1000000 # time starts at 0
header = 'Video ID,Timestamp (milliseconds),amusement,anger,awe,concentration,confusion,contempt,contentment,disappointment,doubt,elation,interest,pain,sadness,surprise,triumph\n'
final_res = {}
for vid, frame_count, out in outputs:# videos
video_time = frame_count // 6 + 1
# print('video', vid, video_time)
entry_count = 0
for t in range(video_time): # seconds
for i in range(6): # frames
timestamp = time_step * t + time_stamps[i]
fcc = t * 6 + i
if fcc >= frame_count:
continue
# print('Frame count', frame_count)
frame_output = out[fcc]
frame_output = [str(x) for x in frame_output]
temp = '{vid},{timestamp},'.format(vid=vid,timestamp=timestamp) + ','.join(frame_output) + '\n'
# file.write(temp)
if vid in final_res:
final_res[vid].append(temp)
else:
final_res[vid] = [temp]
entry_count += 1
assert entry_count == frame_count
# fixed for now
missing = [('WKXrnB7alT8', 2919), ('o0ooW14pIa4', 3733), ('GufMoL_MuNE',2038), ('Uee0Tv1rTz8', 1316), ('ScvvOWtb04Q', 152), ('R9kJlLungmo', 3609),('QMW3GuohzzE', 822), ('fjJYTW2n6rk', 4108), ('rbTIMt0VcLw', 1084),('L9cdaj74kLo', 3678), ('l-ka23gU4NA', 1759)]
for vid, length in missing:
video_time = length // 6 + 1
# print('video', vid, video_time)
for t in range(video_time): # seconds
for i in range(6): # frames
timestamp = time_step * t + time_stamps[i]
fcc = t * 6 + i
if fcc >= length:
continue
frame_output = ',0'*15
temp = '{vid},{timestamp}'.format(vid=vid, timestamp=timestamp) + frame_output + '\n'
# file.write(temp)
if vid in final_res:
final_res[vid].append(temp)
else:
final_res[vid] = [temp]
print('Write test outputs...')
with open('test_output.csv', 'w') as file:
file.write(header)
temp_vidmap = [x.strip().split(' ') for x in open(args.test_vidmap)]
temp_vidmap = [x[0] for x in temp_vidmap]
for vid in tqdm(temp_vidmap):
for entry in final_res[vid]:
file.write(entry)
def main_train(config, checkpoint_dir=None):
global args, best_corr
best_corr = 0.0
args.store_name = '{}'.format(args.model)
args.store_name = args.store_name + datetime.now().strftime('_%m-%d_%H-%M-%S')
args.start_epoch = 0
# check_rootfolders(args)
if args.model == 'Baseline':
model = Baseline()
elif args.model == 'TCFPN':
model = TCFPN(layers=[48, 64, 96], in_channels=(2048 + 128), num_classes=15, kernel_size=11)
model = torch.nn.DataParallel(model).cuda()
if config['optimizer'] == 'adam':
optimizer = torch.optim.Adam(model.parameters(), lr=args.learning_rate)
elif config['optimizer'] == 'adamw':
optimizer = torch.optim.AdamW(model.parameters(), lr=config['lr'])
# custom optimizer
if args.use_sam:
base_optim = torch.optim.Adam
optimizer = SAM(model.parameters(), base_optim, lr=config['lr'])
# custom lr scheduler
if args.use_cos_wr:
scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer, T_0=args.cos_wr_t0,T_mult=args.cos_wr_t_mult)
elif args.use_cos:
scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optimizer, args.cos_t_max)
# SWA
if args.use_swa:
swa_model = torch.optim.swa_utils.AveragedModel(model)
swa_scheduler = torch.optim.swa_utils.SWALR(optimizer, swa_lr=config['lr'])
# ckpt structure {epoch, state_dict, optimizer, best_corr}
# if args.resume and os.path.isfile(args.resume):
# print('Load checkpoint:', args.resume)
# ckpt = torch.load(args.resume)
# args.start_epoch = ckpt['epoch']
# best_corr = ckpt['best_corr']
# model.load_state_dict(ckpt['state_dict'])
# optimizer.load_state_dict(ckpt['optimizer'])
# print('Loaded ckpt at epoch:', args.start_epoch)
if checkpoint_dir:
model_state, optimizer_state = torch.load(
os.path.join(checkpoint_dir, "checkpoint"))
model.load_state_dict(model_state)
optimizer.load_state_dict(optimizer_state)
# initialize datasets
train_loader = torch.utils.data.DataLoader(
dataset=EEV_Dataset(
csv_path=args.train_csv,
vidmap_path=args.train_vidmap,
image_feat_path=args.image_features,
audio_feat_path=args.audio_features,
mode='train', lpfilter=args.lp_filter
),
batch_size=config['batch_size'], shuffle=True,
num_workers=args.workers, pin_memory=False,
drop_last=True
)
val_loader = torch.utils.data.DataLoader(
dataset=EEV_Dataset(
csv_path=args.val_csv,
vidmap_path=args.val_vidmap,
image_feat_path=args.image_features,
audio_feat_path=args.audio_features,
mode='val'
),
batch_size=None, shuffle=False,
num_workers=args.workers, pin_memory=False
)
accuracy = correlation
# with open(os.path.join(args.root_log, args.store_name, 'args.txt'), 'w') as f:
# f.write(str(args))
# tb_writer = SummaryWriter(log_dir=os.path.join(args.root_log, args.store_name))
for epoch in range(args.start_epoch, args.epochs):
# train
train(train_loader, model, optimizer, epoch, None, None)
# do lr scheduling after epoch
if args.use_swa and epoch >= args.swa_start:
print('swa stepping...')
swa_model.update_parameters(model)
swa_scheduler.step()
elif args.use_cos_wr:
print('cos warm restart (T0:{} Tm:{}) stepping...'.format(args.cos_wr_t0, args.cos_wr_t_mult))
scheduler.step()
elif args.use_cos:
print('cos (Tmax:{}) stepping...'.format(args.cos_t_max))
scheduler.step()
# validate
if args.use_swa and epoch >= args.swa_start:
# validate use swa model
corr, loss = validate(val_loader, swa_model, accuracy, epoch, None, None)
else:
corr, loss = validate(val_loader, model, accuracy, epoch, None, None)
is_best = corr > best_corr
best_corr = max(corr, best_corr)
# tb_writer.add_scalar('acc/validate_corr_best', best_corr, epoch)
# output_best = 'Best corr: %.4f\n' % (best_corr)
# print(output_best)
# save_checkpoint({
# 'epoch': epoch + 1,
# 'state_dict': model.state_dict(),
# 'optimizer': optimizer.state_dict(),
# 'best_corr': best_corr,
# }, is_best)
with tune.checkpoint_dir(epoch) as checkpoint_dir:
path = os.path.join(checkpoint_dir, "checkpoint")
if is_best:
path = os.path.join(checkpoint_dir, "checkpoint_best")
torch.save((model.state_dict(), optimizer.state_dict()), path)
tune.report(loss=loss, accuracy=corr, best_corr=best_corr)
def main_train():
global args, best_corr
args.store_name = '{}'.format(args.model)
args.store_name = args.store_name + datetime.now().strftime(
'_%m-%d_%H-%M-%S')
args.start_epoch = 0
if not args.val_only:
check_rootfolders(args)
if args.model == 'Baseline':
if args.cls_indices:
model = Baseline(args.img_feat_size,
args.au_feat_size,
num_classes=len(args.cls_indices))
else:
print('Feature size:', args.img_feat_size, args.au_feat_size)
model = Baseline(args.img_feat_size, args.au_feat_size)
elif args.model == 'TCFPN':
model = TCFPN(layers=[48, 64, 96],
in_channels=(128),
num_classes=15,
kernel_size=11)
elif args.model == 'BaseAu':
model = Baseline_Au(args.au_feat_size)
elif args.model == 'BaseImg':
model = Baseline_Img(args.img_feat_size)
elif args.model == 'EmoBase':
model = EmoBase()
model = torch.nn.DataParallel(model).cuda()
optimizer = torch.optim.Adam(model.parameters(),
lr=args.learning_rate,
weight_decay=args.weight_decay)
# optimizer = torch.optim.AdamW(model.parameters(), lr=args.learning_rate)
# custom optimizer
if args.use_sam:
base_optim = torch.optim.Adam
optimizer = SAM(model.parameters(), base_optim, lr=args.learning_rate)
# custom lr scheduler
if args.use_cos_wr:
scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(
optimizer, T_0=args.cos_wr_t0, T_mult=args.cos_wr_t_mult)
elif args.use_cos:
scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(
optimizer, args.cos_t_max)
elif args.use_multistep:
scheduler = torch.optim.lr_scheduler.MultiStepLR(
optimizer, args.step_milestones, args.step_decay)
# SWA
if args.use_swa:
swa_model = torch.optim.swa_utils.AveragedModel(model)
swa_scheduler = torch.optim.swa_utils.SWALR(optimizer,
swa_lr=args.learning_rate)
# ckpt structure {epoch, state_dict, optimizer, best_corr}
if args.resume and os.path.isfile(args.resume):
print('Load checkpoint:', args.resume)
ckpt = torch.load(args.resume)
args.start_epoch = ckpt['epoch']
best_corr = ckpt['best_corr']
model.load_state_dict(ckpt['state_dict'])
optimizer.load_state_dict(ckpt['optimizer'])
print('Loaded ckpt at epoch:', args.start_epoch)
# initialize datasets
train_loader = torch.utils.data.DataLoader(dataset=EEV_Dataset(
csv_path=args.train_csv,
vidmap_path=args.train_vidmap,
image_feat_path=args.image_features,
audio_feat_path=args.audio_features,
mode='train',
lpfilter=args.lp_filter,
train_freq=args.train_freq,
val_freq=args.val_freq,
cls_indices=args.cls_indices),
batch_size=args.batch_size,
shuffle=True,
num_workers=args.workers,
pin_memory=False,
drop_last=True)
val_loader = torch.utils.data.DataLoader(dataset=EEV_Dataset(
csv_path=args.val_csv,
vidmap_path=args.val_vidmap,
image_feat_path=args.image_features,
audio_feat_path=args.audio_features,
mode='val',
train_freq=args.train_freq,
val_freq=args.val_freq,
cls_indices=args.cls_indices,
repeat_sample=args.repeat_sample),
batch_size=None,
shuffle=False,
num_workers=args.workers,
pin_memory=False)
accuracy = correlation
if args.val_only:
print('Run validation ...')
print('start epoch:', args.start_epoch, 'model:', args.resume)
validate(val_loader, model, accuracy, args.start_epoch, None, None)
return
log_training = open(
os.path.join(args.root_log, args.store_name, 'log.csv'), 'w')
with open(os.path.join(args.root_log, args.store_name, 'args.txt'),
'w') as f:
f.write(str(args))
tb_writer = SummaryWriter(
log_dir=os.path.join(args.root_log, args.store_name))
for epoch in range(args.start_epoch, args.epochs):
train(train_loader, model, optimizer, epoch, log_training, tb_writer)
# do lr scheduling after epoch
if args.use_swa and epoch >= args.swa_start:
print('swa stepping...')
swa_model.update_parameters(model)
swa_scheduler.step()
elif args.use_cos_wr or args.use_cos or args.use_multistep:
scheduler.step()
if (epoch + 1) > 2 and ((epoch + 1) % args.eval_freq == 0 or
(epoch + 1) == args.epochs):
# validate
if args.use_swa and epoch >= args.swa_start:
# validate use swa model
corr = validate(val_loader, swa_model, accuracy, epoch,
log_training, tb_writer)
else:
corr = validate(val_loader, model, accuracy, epoch,
log_training, tb_writer)
is_best = corr > best_corr
best_corr = max(corr, best_corr)
tb_writer.add_scalar('acc/validate_corr_best', best_corr, epoch)
output_best = 'Best corr: %.4f\n' % (best_corr)
print(output_best)
log_training.write(output_best + '\n')
log_training.flush()
save_checkpoint(
{
'epoch': epoch + 1,
'state_dict': model.state_dict(),
'optimizer': optimizer.state_dict(),
'best_corr': best_corr,
}, is_best)
def prepare(args):
resume_from_checkpoint = args.resume_from_checkpoint
prepare_start_time = time.time()
logger.info('global', 'Start preparing.')
check_config_dir()
logger.info('setting', config_info(), time_report=False)
model = Baseline(num_classes=Config.nr_class)
logger.info('setting', model_summary(model), time_report=False)
logger.info('setting', str(model), time_report=False)
train_transforms = transforms.Compose([
transforms.Resize(Config.input_shape),
transforms.RandomApply([
transforms.ColorJitter(
brightness=0.3, contrast=0.3, saturation=0.3, hue=0)
],
p=0.5),
transforms.RandomHorizontalFlip(),
transforms.Pad(10),
transforms.RandomCrop(Config.input_shape),
transforms.ToTensor(),
transforms.RandomErasing(),
transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
])
test_transforms = transforms.Compose([
transforms.Resize(Config.input_shape),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
])
trainset = Veri776_train(transforms=train_transforms, need_attr=True)
testset = Veri776_test(transforms=test_transforms, need_attr=True)
pksampler = PKSampler(trainset, p=Config.P, k=Config.K)
train_loader = torch.utils.data.DataLoader(trainset,
batch_size=Config.batch_size,
sampler=pksampler,
num_workers=Config.nr_worker,
pin_memory=True)
test_loader = torch.utils.data.DataLoader(
testset,
batch_size=Config.batch_size,
sampler=torch.utils.data.SequentialSampler(testset),
num_workers=Config.nr_worker,
pin_memory=True)
weight_decay_setting = parm_list_with_Wdecay(model)
optimizer = torch.optim.Adam(weight_decay_setting, lr=Config.lr)
scheduler = torch.optim.lr_scheduler.LambdaLR(optimizer,
lr_lambda=lr_multi_func)
losses = {}
losses['cross_entropy_loss'] = torch.nn.CrossEntropyLoss()
losses['type_ce_loss'] = torch.nn.CrossEntropyLoss()
losses['color_ce_loss'] = torch.nn.CrossEntropyLoss()
losses['triplet_hard_loss'] = triplet_hard_loss(
margin=Config.triplet_margin)
for k in losses.keys():
losses[k] = losses[k].cuda()
start_epoch = 0
if resume_from_checkpoint and os.path.exists(Config.checkpoint_path):
checkpoint = load_checkpoint()
start_epoch = checkpoint['epoch']
model.load_state_dict(checkpoint['model'])
optimizer.load_state_dict(checkpoint['optimizer'])
scheduler.load_state_dict(checkpoint['scheduler'])
# continue training for next the epoch of the checkpoint, or simply start from 1
start_epoch += 1
ret = {
'start_epoch': start_epoch,
'model': model,
'train_loader': train_loader,
'test_loader': test_loader,
'optimizer': optimizer,
'scheduler': scheduler,
'losses': losses
}
prepare_end_time = time.time()
time_spent = sec2min_sec(prepare_start_time, prepare_end_time)
logger.info(
'global', 'Finish preparing, time spend: {}mins {}s.'.format(
time_spent[0], time_spent[1]))
return ret