def train(hyp, opt, device, tb_writer=None):
logger.info(
colorstr('hyperparameters: ') + ', '.join(f'{k}={v}'
for k, v in hyp.items()))
save_dir, epochs, batch_size, total_batch_size, weights, rank = \
Path(opt.save_dir), opt.epochs, opt.batch_size, opt.total_batch_size, opt.weights, opt.global_rank
# Directories
wdir = save_dir / 'weights'
wdir.mkdir(parents=True, exist_ok=True) # make dir
last = wdir / 'last.pt'
best = wdir / 'best.pt'
results_file = save_dir / 'results.txt'
# Save run settings
with open(save_dir / 'hyp.yaml', 'w') as f:
yaml.safe_dump(hyp, f, sort_keys=False)
with open(save_dir / 'opt.yaml', 'w') as f:
yaml.safe_dump(vars(opt), f, sort_keys=False)
# Configure
plots = False # create plots
cuda = device.type == 'cuda'
init_seeds(2 + rank)
with open(opt.data) as f:
data_dict = yaml.safe_load(f) # data dict
# Logging- Doing this before checking the dataset. Might update data_dict
loggers = {'wandb': None} # loggers dict
if rank in [-1, 0]:
opt.hyp = hyp # add hyperparameters
run_id = torch.load(weights).get('wandb_id') if weights.endswith(
'.pt') and os.path.isfile(weights) else None
wandb_logger = WandbLogger(opt, save_dir.stem, run_id, data_dict)
loggers['wandb'] = wandb_logger.wandb
data_dict = wandb_logger.data_dict
if wandb_logger.wandb:
weights, epochs, hyp = opt.weights, opt.epochs, opt.hyp # WandbLogger might update weights, epochs if resuming
nc = 1 if opt.single_cls else int(data_dict['nc']) # number of classes
names = ['item'] if opt.single_cls and len(
data_dict['names']) != 1 else data_dict['names'] # class names
assert len(names) == nc, '%g names found for nc=%g dataset in %s' % (
len(names), nc, opt.data) # check
is_coco = opt.data.endswith('coco.yaml') and nc == 80 # COCO dataset
# Model
pretrained = weights.endswith('.pt')
if pretrained:
with torch_distributed_zero_first(rank):
weights = attempt_download(
weights) # download if not found locally
map_device = 'cpu' if device.type == 'dml' else device
ckpt = torch.load(weights, map_location=map_device) # load checkpoint
model = Model(opt.cfg or ckpt['model'].yaml,
ch=3,
nc=nc,
anchors=hyp.get('anchors')).to(device) # create
exclude = [
'anchor'
] if (opt.cfg or hyp.get('anchors')) and not opt.resume else [
] # exclude keys
state_dict = ckpt['model'].float().state_dict() # to FP32
state_dict = intersect_dicts(state_dict,
model.state_dict(),
exclude=exclude) # intersect
model.load_state_dict(state_dict, strict=False) # load
logger.info(
'Transferred %g/%g items from %s' %
(len(state_dict), len(model.state_dict()), weights)) # report
else:
model = Model(opt.cfg, ch=3, nc=nc,
anchors=hyp.get('anchors')).to(device) # create
with torch_distributed_zero_first(rank):
check_dataset(data_dict) # check
train_path = data_dict['train']
test_path = data_dict['val']
# Freeze
freeze = [] # parameter names to freeze (full or partial)
for k, v in model.named_parameters():
v.requires_grad = True # train all layers
if any(x in k for x in freeze):
print('freezing %s' % k)
v.requires_grad = False
# Optimizer
nbs = 64 # nominal batch size
accumulate = max(round(nbs / total_batch_size),
1) # accumulate loss before optimizing
hyp['weight_decay'] *= total_batch_size * accumulate / nbs # scale weight_decay
logger.info(f"Scaled weight_decay = {hyp['weight_decay']}")
pg0, pg1, pg2 = [], [], [] # optimizer parameter groups
for k, v in model.named_modules():
if hasattr(v, 'bias') and isinstance(v.bias, nn.Parameter):
pg2.append(v.bias) # biases
if isinstance(v, nn.BatchNorm2d):
pg0.append(v.weight) # no decay
elif hasattr(v, 'weight') and isinstance(v.weight, nn.Parameter):
pg1.append(v.weight) # apply decay
if opt.adam:
optimizer = optim.Adam(pg0,
lr=hyp['lr0'],
betas=(hyp['momentum'],
0.999)) # adjust beta1 to momentum
else:
optimizer = optim.SGD(pg0,
lr=hyp['lr0'],
momentum=hyp['momentum'],
nesterov=True)
optimizer.add_param_group({
'params': pg1,
'weight_decay': hyp['weight_decay']
}) # add pg1 with weight_decay
optimizer.add_param_group({'params': pg2}) # add pg2 (biases)
logger.info('Optimizer groups: %g .bias, %g conv.weight, %g other' %
(len(pg2), len(pg1), len(pg0)))
del pg0, pg1, pg2
# Scheduler https://arxiv.org/pdf/1812.01187.pdf
# https://pytorch.org/docs/stable/_modules/torch/optim/lr_scheduler.html#OneCycleLR
if opt.linear_lr:
lf = lambda x: (1 - x / (epochs - 1)) * (1.0 - hyp['lrf']) + hyp[
'lrf'] # linear
else:
lf = one_cycle(1, hyp['lrf'], epochs) # cosine 1->hyp['lrf']
scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lf)
# plot_lr_scheduler(optimizer, scheduler, epochs)
# EMA
ema = ModelEMA(model) if rank in [-1, 0] else None
# Resume
start_epoch, best_fitness = 0, 0.0
if pretrained:
# Optimizer
if ckpt['optimizer'] is not None:
optimizer.load_state_dict(ckpt['optimizer'])
best_fitness = ckpt['best_fitness']
# EMA
if ema and ckpt.get('ema'):
ema.ema.load_state_dict(ckpt['ema'].float().state_dict())
ema.updates = ckpt['updates']
# Results
if ckpt.get('training_results') is not None:
results_file.write_text(
ckpt['training_results']) # write results.txt
# Epochs
start_epoch = ckpt['epoch'] + 1
if opt.resume:
assert start_epoch > 0, '%s training to %g epochs is finished, nothing to resume.' % (
weights, epochs)
if epochs < start_epoch:
logger.info(
'%s has been trained for %g epochs. Fine-tuning for %g additional epochs.'
% (weights, ckpt['epoch'], epochs))
epochs += ckpt['epoch'] # finetune additional epochs
del ckpt, state_dict
# Image sizes
gs = max(int(model.stride.max()), 32) # grid size (max stride)
nl = model.model[
-1].nl # number of detection layers (used for scaling hyp['obj'])
imgsz, imgsz_test = [check_img_size(x, gs) for x in opt.img_size
] # verify imgsz are gs-multiples
# DP mode
if cuda and rank == -1 and torch.cuda.device_count() > 1:
model = torch.nn.DataParallel(model)
# SyncBatchNorm
if opt.sync_bn and cuda and rank != -1:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model).to(device)
logger.info('Using SyncBatchNorm()')
# Trainloader
dataloader, dataset = create_dataloader(train_path,
imgsz,
batch_size,
gs,
opt,
hyp=hyp,
augment=True,
cache=opt.cache_images,
rect=opt.rect,
rank=rank,
world_size=opt.world_size,
workers=opt.workers,
image_weights=opt.image_weights,
quad=opt.quad,
prefix=colorstr('train: '))
mlc = np.concatenate(dataset.labels, 0)[:, 0].max() # max label class
nb = len(dataloader) # number of batches
assert mlc < nc, 'Label class %g exceeds nc=%g in %s. Possible class labels are 0-%g' % (
mlc, nc, opt.data, nc - 1)
# Process 0
if rank in [-1, 0]:
testloader = create_dataloader(
test_path,
imgsz_test,
batch_size * 2,
gs,
opt, # testloader
hyp=hyp,
cache=opt.cache_images and not opt.notest,
rect=True,
rank=-1,
world_size=opt.world_size,
workers=opt.workers,
pad=0.5,
prefix=colorstr('val: '))[0]
if not opt.resume:
labels = np.concatenate(dataset.labels, 0)
c = torch.tensor(labels[:, 0]) # classes
# cf = torch.bincount(c.long(), minlength=nc) + 1. # frequency
# model._initialize_biases(cf.to(device))
if plots:
plot_labels(labels, names, save_dir, loggers)
if tb_writer:
tb_writer.add_histogram('classes', c, 0)
# Anchors
if not opt.noautoanchor:
check_anchors(dataset,
model=model,
thr=hyp['anchor_t'],
imgsz=imgsz)
if cuda:
model.half().float() # pre-reduce anchor precision
# DDP mode
if cuda and rank != -1:
model = DDP(
model,
device_ids=[opt.local_rank],
output_device=opt.local_rank,
# nn.MultiheadAttention incompatibility with DDP https://github.com/pytorch/pytorch/issues/26698
find_unused_parameters=any(
isinstance(layer, nn.MultiheadAttention)
for layer in model.modules()))
# Model parameters
hyp['box'] *= 3. / nl # scale to layers
hyp['cls'] *= nc / 80. * 3. / nl # scale to classes and layers
hyp['obj'] *= (imgsz / 640)**2 * 3. / nl # scale to image size and layers
hyp['label_smoothing'] = opt.label_smoothing
model.nc = nc # attach number of classes to model
model.hyp = hyp # attach hyperparameters to model
model.gr = 1.0 # iou loss ratio (obj_loss = 1.0 or iou)
model.class_weights = labels_to_class_weights(
dataset.labels, nc).to(device) * nc # attach class weights
model.names = names
# Start training
t0 = time.time()
nw = max(round(hyp['warmup_epochs'] * nb),
1000) # number of warmup iterations, max(3 epochs, 1k iterations)
# nw = min(nw, (epochs - start_epoch) / 2 * nb) # limit warmup to < 1/2 of training
maps = np.zeros(nc) # mAP per class
results = (0, 0, 0, 0, 0, 0, 0
) # P, R, [email protected], [email protected], val_loss(box, obj, cls)
scheduler.last_epoch = start_epoch - 1 # do not move
scaler = amp.GradScaler(enabled=cuda)
compute_loss = ComputeLoss(model) # init loss class
logger.info(f'Image sizes {imgsz} train, {imgsz_test} test\n'
f'Using {dataloader.num_workers} dataloader workers\n'
f'Logging results to {save_dir}\n'
f'Starting training for {epochs} epochs...')
for epoch in range(
start_epoch, epochs
): # epoch ------------------------------------------------------------------
model.train()
# Update image weights (optional)
if opt.image_weights:
# Generate indices
if rank in [-1, 0]:
cw = model.class_weights.cpu().numpy() * (
1 - maps)**2 / nc # class weights
iw = labels_to_image_weights(dataset.labels,
nc=nc,
class_weights=cw) # image weights
dataset.indices = random.choices(
range(dataset.n), weights=iw,
k=dataset.n) # rand weighted idx
# Broadcast if DDP
if rank != -1:
indices = (torch.tensor(dataset.indices)
if rank == 0 else torch.zeros(dataset.n)).int()
dist.broadcast(indices, 0)
if rank != 0:
dataset.indices = indices.cpu().numpy()
# Update mosaic border
# b = int(random.uniform(0.25 * imgsz, 0.75 * imgsz + gs) // gs * gs)
# dataset.mosaic_border = [b - imgsz, -b] # height, width borders
mloss = torch.zeros(4, device=device) # mean losses
if rank != -1:
dataloader.sampler.set_epoch(epoch)
pbar = enumerate(dataloader)
logger.info(
('\n' + '%10s' * 8) % ('Epoch', 'gpu_mem', 'box', 'obj', 'cls',
'total', 'labels', 'img_size'))
if rank in [-1, 0]:
pbar = tqdm(pbar, total=nb) # progress bar
optimizer.zero_grad()
for i, (
imgs, targets, paths, _
) in pbar: # batch -------------------------------------------------------------
ni = i + nb * epoch # number integrated batches (since train start)
imgs = imgs.to(device, non_blocking=True).float(
) / 255.0 # uint8 to float32, 0-255 to 0.0-1.0
# Warmup
if ni <= nw:
xi = [0, nw] # x interp
# model.gr = np.interp(ni, xi, [0.0, 1.0]) # iou loss ratio (obj_loss = 1.0 or iou)
accumulate = max(
1,
np.interp(ni, xi, [1, nbs / total_batch_size]).round())
for j, x in enumerate(optimizer.param_groups):
# bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
x['lr'] = np.interp(ni, xi, [
hyp['warmup_bias_lr'] if j == 2 else 0.0,
x['initial_lr'] * lf(epoch)
])
if 'momentum' in x:
x['momentum'] = np.interp(
ni, xi, [hyp['warmup_momentum'], hyp['momentum']])
# Multi-scale
if opt.multi_scale:
sz = random.randrange(imgsz * 0.5,
imgsz * 1.5 + gs) // gs * gs # size
sf = sz / max(imgs.shape[2:]) # scale factor
if sf != 1:
ns = [math.ceil(x * sf / gs) * gs for x in imgs.shape[2:]
] # new shape (stretched to gs-multiple)
imgs = F.interpolate(imgs,
size=ns,
mode='bilinear',
align_corners=False)
# Forward
with amp.autocast(enabled=cuda):
pred = model(imgs) # forward
loss, loss_items = compute_loss(
[pre.to("cpu") for pre in pred],
targets.to("cpu")) # loss scaled by batch_size
if rank != -1:
loss *= opt.world_size # gradient averaged between devices in DDP mode
if opt.quad:
loss *= 4.
# Backward
scaler.scale(loss).backward()
# Optimize
if ni % accumulate == 0:
scaler.step(optimizer) # optimizer.step
scaler.update()
optimizer.zero_grad()
if ema:
ema.update(model)
# Print
if rank in [-1, 0]:
mloss = (mloss * i + loss_items) / (i + 1
) # update mean losses
mem = '%.3gG' % (torch.cuda.memory_reserved() / 1E9
if torch.cuda.is_available() else 0) # (GB)
s = ('%10s' * 2 +
'%10.4g' * 6) % ('%g/%g' % (epoch, epochs - 1), mem,
*mloss, targets.shape[0], imgs.shape[-1])
pbar.set_description(s)
# Plot
if plots and ni < 3:
f = save_dir / f'train_batch{ni}.jpg' # filename
Thread(target=plot_images,
args=(imgs, targets, paths, f),
daemon=True).start()
if tb_writer:
tb_writer.add_graph(
torch.jit.trace(de_parallel(model),
imgs,
strict=False), []) # model graph
# tb_writer.add_image(f, result, dataformats='HWC', global_step=epoch)
elif plots and ni == 10 and wandb_logger.wandb:
wandb_logger.log({
"Mosaics": [
wandb_logger.wandb.Image(str(x), caption=x.name)
for x in save_dir.glob('train*.jpg') if x.exists()
]
})
# end batch ------------------------------------------------------------------------------------------------
# end epoch ----------------------------------------------------------------------------------------------------
# Scheduler
lr = [x['lr'] for x in optimizer.param_groups] # for tensorboard
scheduler.step()
# DDP process 0 or single-GPU
if rank in [-1, 0]:
# mAP
ema.update_attr(model,
include=[
'yaml', 'nc', 'hyp', 'gr', 'names', 'stride',
'class_weights'
])
final_epoch = epoch + 1 == epochs
if not opt.notest or final_epoch: # Calculate mAP
wandb_logger.current_epoch = epoch + 1
results, maps, times = test.test(data_dict,
batch_size=batch_size * 2,
imgsz=imgsz_test,
model=ema.ema,
single_cls=opt.single_cls,
dataloader=testloader,
save_dir=save_dir,
save_json=is_coco
and final_epoch,
verbose=nc < 50
and final_epoch,
plots=plots and final_epoch,
wandb_logger=wandb_logger,
compute_loss=None,
is_coco=is_coco)
# Write
with open(results_file, 'a') as f:
f.write(s + '%10.4g' * 7 % results +
'\n') # append metrics, val_loss
# Log
tags = [
'train/box_loss',
'train/obj_loss',
'train/cls_loss', # train loss
'metrics/precision',
'metrics/recall',
'metrics/mAP_0.5',
'metrics/mAP_0.5:0.95',
'val/box_loss',
'val/obj_loss',
'val/cls_loss', # val loss
'x/lr0',
'x/lr1',
'x/lr2'
] # params
for x, tag in zip(list(mloss[:-1]) + list(results) + lr, tags):
if tb_writer:
tb_writer.add_scalar(tag, x, epoch) # tensorboard
if wandb_logger.wandb:
wandb_logger.log({tag: x}) # W&B
# Update best mAP
fi = fitness(np.array(results).reshape(
1, -1)) # weighted combination of [P, R, [email protected], [email protected]]
if fi > best_fitness:
best_fitness = fi
wandb_logger.end_epoch(best_result=best_fitness == fi)
# Save model
if (not opt.nosave) or (final_epoch and not opt.evolve): # if save
ckpt = {
'epoch':
epoch,
'best_fitness':
best_fitness,
'training_results':
results_file.read_text(),
'model':
deepcopy(de_parallel(model)).half(),
'ema':
deepcopy(ema.ema).half(),
'updates':
ema.updates,
'optimizer':
optimizer.state_dict(),
'wandb_id':
wandb_logger.wandb_run.id if wandb_logger.wandb else None
}
# Save last, best and delete
# torch.save(ckpt, last)
# if best_fitness == fi:
# torch.save(ckpt, best)
if wandb_logger.wandb:
if ((epoch + 1) % opt.save_period == 0
and not final_epoch) and opt.save_period != -1:
wandb_logger.log_model(last.parent,
opt,
epoch,
fi,
best_model=best_fitness == fi)
del ckpt
# end epoch ----------------------------------------------------------------------------------------------------
# end training
if rank in [-1, 0]:
logger.info(
f'{epoch - start_epoch + 1} epochs completed in {(time.time() - t0) / 3600:.3f} hours.\n'
)
if plots:
plot_results(save_dir=save_dir) # save as results.png
if wandb_logger.wandb:
files = [
'results.png', 'confusion_matrix.png',
*[f'{x}_curve.png' for x in ('F1', 'PR', 'P', 'R')]
]
wandb_logger.log({
"Results": [
wandb_logger.wandb.Image(str(save_dir / f), caption=f)
for f in files if (save_dir / f).exists()
]
})
if not opt.evolve:
if is_coco: # COCO dataset
for m in [last, best
] if best.exists() else [last]: # speed, mAP tests
results, _, _ = test.test(opt.data,
batch_size=batch_size * 2,
imgsz=imgsz_test,
conf_thres=0.001,
iou_thres=0.7,
model=attempt_load(
m, device).half(),
single_cls=opt.single_cls,
dataloader=testloader,
save_dir=save_dir,
save_json=True,
plots=False,
is_coco=is_coco)
# Strip optimizers
for f in last, best:
if f.exists():
strip_optimizer(f) # strip optimizers
if wandb_logger.wandb: # Log the stripped model
wandb_logger.wandb.log_artifact(
str(best if best.exists() else last),
type='model',
name='run_' + wandb_logger.wandb_run.id + '_model',
aliases=['latest', 'best', 'stripped'])
wandb_logger.finish_run()
else:
dist.destroy_process_group()
torch.cuda.empty_cache()
return results
import argparse
from utils.plots import plot_results
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--save_path',
type=str,
default='runs/train/exp3',
help='training results path')
opt = parser.parse_args()
plot_results(save_dir=opt.save_path)
def train(hyp, opt, device, tb_writer=None, wandb=None):
logger.info(f'Hyperparameters {hyp}')
save_dir, epochs, batch_size, total_batch_size, weights, rank = \
Path(opt.save_dir), opt.epochs, opt.batch_size, opt.total_batch_size, opt.weights, opt.global_rank
# Directories
wdir = save_dir / 'weights'
wdir.mkdir(parents=True, exist_ok=True) # make dir
last = wdir / 'last.pt'
best = wdir / 'best.pt'
results_file = save_dir / 'results.txt'
# Save run settings
with open(save_dir / 'hyp.yaml', 'w') as f:
yaml.dump(hyp, f, sort_keys=False)
with open(save_dir / 'opt.yaml', 'w') as f:
yaml.dump(vars(opt), f, sort_keys=False)
# Configure
plots = not opt.evolve # create plots
cuda = device.type != 'cpu'
init_seeds(2 + rank)
with open(opt.data) as f:
data_dict = yaml.load(f, Loader=yaml.FullLoader) # data dict
with torch_distributed_zero_first(rank):
check_dataset(data_dict) # check
train_path = data_dict['train']
test_path = data_dict['val']
nc, names = (1, ['item']) if opt.single_cls else (int(
data_dict['nc']), data_dict['names']) # number classes, names
assert len(names) == nc, '%g names found for nc=%g dataset in %s' % (
len(names), nc, opt.data) # check
# Model
pretrained = weights.endswith('.pt')
if pretrained:
with torch_distributed_zero_first(rank):
attempt_download(weights) # download if not found locally
ckpt = torch.load(weights, map_location=device) # load checkpoint
if hyp.get('anchors'):
ckpt['model'].yaml['anchors'] = round(
hyp['anchors']) # force autoanchor
model = Model(opt.cfg or ckpt['model'].yaml, ch=3,
nc=nc).to(device) # create
exclude = ['anchor'] if opt.cfg or hyp.get('anchors') else [
] # exclude keys
state_dict = ckpt['model'].float().state_dict() # to FP32
state_dict = intersect_dicts(state_dict,
model.state_dict(),
exclude=exclude) # intersect
model.load_state_dict(state_dict, strict=False) # load
logger.info(
'Transferred %g/%g items from %s' %
(len(state_dict), len(model.state_dict()), weights)) # report
else:
model = Model(opt.cfg, ch=3, nc=nc).to(device) # create
# Freeze
freeze = [] # parameter names to freeze (full or partial)
for k, v in model.named_parameters():
v.requires_grad = True # train all layers
if any(x in k for x in freeze):
print('freezing %s' % k)
v.requires_grad = False
# Optimizer
nbs = 64 # nominal batch size
accumulate = max(round(nbs / total_batch_size),
1) # accumulate loss before optimizing
hyp['weight_decay'] *= total_batch_size * accumulate / nbs # scale weight_decay
pg0, pg1, pg2 = [], [], [] # optimizer parameter groups
for k, v in model.named_modules():
if hasattr(v, 'bias') and isinstance(v.bias, nn.Parameter):
pg2.append(v.bias) # biases
if isinstance(v, nn.BatchNorm2d):
pg0.append(v.weight) # no decay
elif hasattr(v, 'weight') and isinstance(v.weight, nn.Parameter):
pg1.append(v.weight) # apply decay
if opt.adam:
optimizer = optim.Adam(pg0,
lr=hyp['lr0'],
betas=(hyp['momentum'],
0.999)) # adjust beta1 to momentum
else:
optimizer = optim.SGD(pg0,
lr=hyp['lr0'],
momentum=hyp['momentum'],
nesterov=True)
optimizer.add_param_group({
'params': pg1,
'weight_decay': hyp['weight_decay']
}) # add pg1 with weight_decay
optimizer.add_param_group({'params': pg2}) # add pg2 (biases)
logger.info('Optimizer groups: %g .bias, %g conv.weight, %g other' %
(len(pg2), len(pg1), len(pg0)))
del pg0, pg1, pg2
# Scheduler https://arxiv.org/pdf/1812.01187.pdf
# https://pytorch.org/docs/stable/_modules/torch/optim/lr_scheduler.html#OneCycleLR
lf = lambda x: ((1 + math.cos(x * math.pi / epochs)) / 2) * (1 - hyp[
'lrf']) + hyp['lrf'] # cosine
scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lf)
# plot_lr_scheduler(optimizer, scheduler, epochs)
# Logging
if wandb and wandb.run is None:
opt.hyp = hyp # add hyperparameters
wandb_run = wandb.init(
config=opt,
resume="allow",
project='YOLOv5'
if opt.project == 'runs/train' else Path(opt.project).stem,
name=save_dir.stem,
id=ckpt.get('wandb_id') if 'ckpt' in locals() else None)
loggers = {'wandb': wandb} # loggers dict
# Resume
start_epoch, best_fitness = 0, 0.0
if pretrained:
# Optimizer
if ckpt['optimizer'] is not None:
optimizer.load_state_dict(ckpt['optimizer'])
best_fitness = ckpt['best_fitness']
# Results
if ckpt.get('training_results') is not None:
with open(results_file, 'w') as file:
file.write(ckpt['training_results']) # write results.txt
# Epochs
start_epoch = ckpt['epoch'] + 1
if opt.resume:
assert start_epoch > 0, '%s training to %g epochs is finished, nothing to resume.' % (
weights, epochs)
if epochs < start_epoch:
logger.info(
'%s has been trained for %g epochs. Fine-tuning for %g additional epochs.'
% (weights, ckpt['epoch'], epochs))
epochs += ckpt['epoch'] # finetune additional epochs
del ckpt, state_dict
# Image sizes
gs = int(max(model.stride)) # grid size (max stride)
imgsz, imgsz_test = [check_img_size(x, gs) for x in opt.img_size
] # verify imgsz are gs-multiples
# DP mode
if cuda and rank == -1 and torch.cuda.device_count() > 1:
model = torch.nn.DataParallel(model)
# SyncBatchNorm
if opt.sync_bn and cuda and rank != -1:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model).to(device)
logger.info('Using SyncBatchNorm()')
# EMA
ema = ModelEMA(model) if rank in [-1, 0] else None
# DDP mode
if cuda and rank != -1:
model = DDP(model,
device_ids=[opt.local_rank],
output_device=opt.local_rank)
# Trainloader
dataloader, dataset = create_dataloader(train_path,
imgsz,
batch_size,
gs,
opt,
hyp=hyp,
augment=True,
cache=opt.cache_images,
rect=opt.rect,
rank=rank,
world_size=opt.world_size,
workers=opt.workers,
image_weights=opt.image_weights)
mlc = np.concatenate(dataset.labels, 0)[:, 0].max() # max label class
nb = len(dataloader) # number of batches
assert mlc < nc, 'Label class %g exceeds nc=%g in %s. Possible class labels are 0-%g' % (
mlc, nc, opt.data, nc - 1)
# Process 0
if rank in [-1, 0]:
ema.updates = start_epoch * nb // accumulate # set EMA updates
testloader = create_dataloader(
test_path,
imgsz_test,
total_batch_size,
gs,
opt, # testloader
hyp=hyp,
cache=opt.cache_images and not opt.notest,
rect=True,
rank=-1,
world_size=opt.world_size,
workers=opt.workers,
pad=0.5)[0]
if not opt.resume:
labels = np.concatenate(dataset.labels, 0)
c = torch.tensor(labels[:, 0]) # classes
# cf = torch.bincount(c.long(), minlength=nc) + 1. # frequency
# model._initialize_biases(cf.to(device))
if plots:
Thread(target=plot_labels,
args=(labels, save_dir, loggers),
daemon=True).start()
if tb_writer:
tb_writer.add_histogram('classes', c, 0)
# Anchors
if not opt.noautoanchor:
check_anchors(dataset,
model=model,
thr=hyp['anchor_t'],
imgsz=imgsz)
# Model parameters
hyp['cls'] *= nc / 80. # scale coco-tuned hyp['cls'] to current dataset
model.nc = nc # attach number of classes to model
model.hyp = hyp # attach hyperparameters to model
model.gr = 1.0 # iou loss ratio (obj_loss = 1.0 or iou)
model.class_weights = labels_to_class_weights(dataset.labels, nc).to(
device) # attach class weights
model.names = names
# Start training
t0 = time.time()
nw = max(round(hyp['warmup_epochs'] * nb),
1000) # number of warmup iterations, max(3 epochs, 1k iterations)
# nw = min(nw, (epochs - start_epoch) / 2 * nb) # limit warmup to < 1/2 of training
maps = np.zeros(nc) # mAP per class
results = (0, 0, 0, 0, 0, 0, 0
) # P, R, [email protected], [email protected], val_loss(box, obj, cls)
scheduler.last_epoch = start_epoch - 1 # do not move
scaler = amp.GradScaler(enabled=cuda)
logger.info('Image sizes %g train, %g test\n'
'Using %g dataloader workers\nLogging results to %s\n'
'Starting training for %g epochs...' %
(imgsz, imgsz_test, dataloader.num_workers, save_dir, epochs))
for epoch in range(
start_epoch, epochs
): # epoch ------------------------------------------------------------------
model.train()
# Update image weights (optional)
if opt.image_weights:
# Generate indices
if rank in [-1, 0]:
cw = model.class_weights.cpu().numpy() * (
1 - maps)**2 # class weights
iw = labels_to_image_weights(dataset.labels,
nc=nc,
class_weights=cw) # image weights
dataset.indices = random.choices(
range(dataset.n), weights=iw,
k=dataset.n) # rand weighted idx
# Broadcast if DDP
if rank != -1:
indices = (torch.tensor(dataset.indices)
if rank == 0 else torch.zeros(dataset.n)).int()
dist.broadcast(indices, 0)
if rank != 0:
dataset.indices = indices.cpu().numpy()
# Update mosaic border
# b = int(random.uniform(0.25 * imgsz, 0.75 * imgsz + gs) // gs * gs)
# dataset.mosaic_border = [b - imgsz, -b] # height, width borders
mloss = torch.zeros(4, device=device) # mean losses
if rank != -1:
dataloader.sampler.set_epoch(epoch)
pbar = enumerate(dataloader)
logger.info(
('\n' + '%10s' * 8) % ('Epoch', 'gpu_mem', 'box', 'obj', 'cls',
'total', 'targets', 'img_size'))
if rank in [-1, 0]:
pbar = tqdm(pbar, total=nb) # progress bar
optimizer.zero_grad()
for i, (
imgs, targets, paths, _
) in pbar: # batch -------------------------------------------------------------
ni = i + nb * epoch # number integrated batches (since train start)
imgs = imgs.to(device, non_blocking=True).float(
) / 255.0 # uint8 to float32, 0-255 to 0.0-1.0
# Warmup
if ni <= nw:
xi = [0, nw] # x interp
# model.gr = np.interp(ni, xi, [0.0, 1.0]) # iou loss ratio (obj_loss = 1.0 or iou)
accumulate = max(
1,
np.interp(ni, xi, [1, nbs / total_batch_size]).round())
for j, x in enumerate(optimizer.param_groups):
# bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
x['lr'] = np.interp(ni, xi, [
hyp['warmup_bias_lr'] if j == 2 else 0.0,
x['initial_lr'] * lf(epoch)
])
if 'momentum' in x:
x['momentum'] = np.interp(
ni, xi, [hyp['warmup_momentum'], hyp['momentum']])
# Multi-scale
if opt.multi_scale:
sz = random.randrange(imgsz * 0.5,
imgsz * 1.5 + gs) // gs * gs # size
sf = sz / max(imgs.shape[2:]) # scale factor
if sf != 1:
ns = [math.ceil(x * sf / gs) * gs for x in imgs.shape[2:]
] # new shape (stretched to gs-multiple)
imgs = F.interpolate(imgs,
size=ns,
mode='bilinear',
align_corners=False)
# Forward
with amp.autocast(enabled=cuda):
pred = model(imgs) # forward
loss, loss_items = compute_loss(
pred, targets.to(device),
model) # loss scaled by batch_size
if rank != -1:
loss *= opt.world_size # gradient averaged between devices in DDP mode
# Backward
scaler.scale(loss).backward()
# Optimize
if ni % accumulate == 0:
scaler.step(optimizer) # optimizer.step
scaler.update()
optimizer.zero_grad()
if ema:
ema.update(model)
# Print
if rank in [-1, 0]:
mloss = (mloss * i + loss_items) / (i + 1
) # update mean losses
mem = '%.3gG' % (torch.cuda.memory_reserved() / 1E9
if torch.cuda.is_available() else 0) # (GB)
s = ('%10s' * 2 +
'%10.4g' * 6) % ('%g/%g' % (epoch, epochs - 1), mem,
*mloss, targets.shape[0], imgs.shape[-1])
pbar.set_description(s)
# Plot
if plots and ni < 3:
f = save_dir / f'train_batch{ni}.jpg' # filename
Thread(target=plot_images,
args=(imgs, targets, paths, f),
daemon=True).start()
# if tb_writer:
# tb_writer.add_image(f, result, dataformats='HWC', global_step=epoch)
# tb_writer.add_graph(model, imgs) # add model to tensorboard
elif plots and ni == 3 and wandb:
wandb.log({
"Mosaics": [
wandb.Image(str(x), caption=x.name)
for x in save_dir.glob('train*.jpg')
]
})
# end batch ------------------------------------------------------------------------------------------------
# end epoch ----------------------------------------------------------------------------------------------------
# Scheduler
lr = [x['lr'] for x in optimizer.param_groups] # for tensorboard
scheduler.step()
# DDP process 0 or single-GPU
if rank in [-1, 0]:
# mAP
if ema:
ema.update_attr(
model,
include=['yaml', 'nc', 'hyp', 'gr', 'names', 'stride'])
final_epoch = epoch + 1 == epochs
if not opt.notest or final_epoch: # Calculate mAP
results, maps, times = test.test(
opt.data,
batch_size=total_batch_size,
imgsz=imgsz_test,
model=ema.ema,
single_cls=opt.single_cls,
dataloader=testloader,
save_dir=save_dir,
plots=plots and final_epoch,
log_imgs=opt.log_imgs if wandb else 0)
# Write
with open(results_file, 'a') as f:
f.write(
s + '%10.4g' * 7 % results +
'\n') # P, R, [email protected], [email protected], val_loss(box, obj, cls)
if len(opt.name) and opt.bucket:
os.system('gsutil cp %s gs://%s/results/results%s.txt' %
(results_file, opt.bucket, opt.name))
# Log
tags = [
'train/box_loss',
'train/obj_loss',
'train/cls_loss', # train loss
'metrics/precision',
'metrics/recall',
'metrics/mAP_0.5',
'metrics/mAP_0.5:0.95',
'val/box_loss',
'val/obj_loss',
'val/cls_loss', # val loss
'x/lr0',
'x/lr1',
'x/lr2'
] # params
for x, tag in zip(list(mloss[:-1]) + list(results) + lr, tags):
if tb_writer:
tb_writer.add_scalar(tag, x, epoch) # tensorboard
if wandb:
wandb.log({tag: x}) # W&B
# Update best mAP
fi = fitness(np.array(results).reshape(
1, -1)) # weighted combination of [P, R, [email protected], [email protected]]
if fi > best_fitness:
best_fitness = fi
# Save model
save = (not opt.nosave) or (final_epoch and not opt.evolve)
if save:
with open(results_file, 'r') as f: # create checkpoint
ckpt = {
'epoch':
epoch,
'best_fitness':
best_fitness,
'training_results':
f.read(),
'model':
ema.ema,
'optimizer':
None if final_epoch else optimizer.state_dict(),
'wandb_id':
wandb_run.id if wandb else None
}
# Save last, best and delete
torch.save(ckpt, last)
if best_fitness == fi:
torch.save(ckpt, best)
del ckpt
# end epoch ----------------------------------------------------------------------------------------------------
# end training
if rank in [-1, 0]:
# Strip optimizers
for f in [last, best]:
if f.exists(): # is *.pt
strip_optimizer(f) # strip optimizer
os.system('gsutil cp %s gs://%s/weights' %
(f, opt.bucket)) if opt.bucket else None # upload
# Plots
if plots:
plot_results(save_dir=save_dir) # save as results.png
if wandb:
files = [
'results.png', 'precision_recall_curve.png',
'confusion_matrix.png'
]
wandb.log({
"Results": [
wandb.Image(str(save_dir / f), caption=f)
for f in files if (save_dir / f).exists()
]
})
logger.info('%g epochs completed in %.3f hours.\n' %
(epoch - start_epoch + 1, (time.time() - t0) / 3600))
# Test best.pt
if opt.data.endswith('coco.yaml') and nc == 80: # if COCO
for conf, iou, save_json in ([0.25, 0.45,
False], [0.001, 0.65,
True]): # speed, mAP tests
results, _, _ = test.test(opt.data,
batch_size=total_batch_size,
imgsz=imgsz_test,
conf_thres=conf,
iou_thres=iou,
model=attempt_load(
best if best.exists() else last,
device).half(),
single_cls=opt.single_cls,
dataloader=testloader,
save_dir=save_dir,
save_json=save_json,
plots=False)
else:
dist.destroy_process_group()
wandb.run.finish() if wandb and wandb.run else None
torch.cuda.empty_cache()
return results
def train(self, hyp, opt, device):
opt = self.opt
hyp = self.hyp
nbs = 64 # nominal batch size
logger.info(
colorstr('hyperparameters: ') + ', '.join(f'{k}={v}'
for k, v in hyp.items()))
save_dir, epochs, batch_size, weights = \
Path(opt.save_dir), opt.epochs, opt.batch_size, opt.weights
total_batch_size = batch_size
plots = True # create plots
# Directories
wdir = save_dir / 'weights'
wdir.mkdir(parents=True, exist_ok=True) # make dir
last = wdir / 'last.pt'
best = wdir / 'best.pt'
results_file = save_dir / 'results.txt'
# Save run settings
with open(save_dir / 'hyp.yaml', 'w') as f:
yaml.dump(hyp, f, sort_keys=False)
with open(save_dir / 'opt.yaml', 'w') as f:
yaml.dump(vars(opt), f, sort_keys=False)
# Configure
cuda = device.type != 'cpu'
init_seeds(2)
with open(opt.data) as f:
data_dict = yaml.load(f, Loader=yaml.SafeLoader) # data dict
# with torch_distributed_zero_first(rank):
# check_dataset(data_dict) # check
train_path = data_dict['train']
test_path = data_dict['val']
# print(data_dict)
# Model
pretrained, ckpt, self.model = self.build_model(weights)
# Freeze
# self.freeze(model)
# Optimizer
self.optimizer = self.build_optim(total_batch_size)
self.scheduler, lf = self.build_scheduler(epochs=epochs)
wandb = False
# Logging
if wandb and wandb.run is None:
opt.hyp = hyp # add hyperparameters
wandb_run = wandb.init(
config=opt,
resume="allow",
project='YOLODB'
if opt.project == 'runs/train' else Path(opt.project).stem,
name=save_dir.stem,
entity=opt.entity,
id=ckpt.get('wandb_id') if ckpt is not None else None)
loggers = {'wandb': wandb} # loggers dict
# EMA
# ema = ModelEMA(self.model) if rank in [-1, 0] else None
ema = ModelEMA(self.model)
# Resume
start_epoch, best_fitness = 0, 0.0
if pretrained:
# Optimizer
if ckpt and ckpt['optimizer'] is not None:
self.optimizer.load_state_dict(ckpt['optimizer'])
best_fitness = ckpt['best_fitness']
# EMA
if ema and ckpt.get('ema'):
ema.ema.load_state_dict(ckpt['ema'][0].float().state_dict())
ema.updates = ckpt['ema'][1]
# Results
if ckpt.get('training_results') is not None:
results_file.write_text(
ckpt['training_results']) # write results.txt
# Epochs
start_epoch = ckpt['epoch'] + 1
if opt.resume:
assert start_epoch > 0, '%s training to %g epochs is finished, nothing to resume.' % (
weights, epochs)
if epochs < start_epoch:
logger.info(
'%s has been trained for %g epochs. Fine-tuning for %g additional epochs.'
% (weights, ckpt['epoch'], epochs))
epochs += ckpt['epoch'] # finetune additional epochs
del ckpt
imgsz, imgsz_test = opt.img_size
train_process = data_dict['process']['train']
val_process = data_dict['process']['val']
# Trainloader
dataloader, dataset = build_dataloader(train_path,
imgsz,
batch_size,
opt,
hyp=hyp,
augment=True,
cache=opt.cache_images,
rect=opt.rect,
workers=opt.workers,
image_weights=opt.image_weights,
quad=opt.quad,
prefix=colorstr('train: '),
process_list=train_process,
mode='train')
num_of_batches = len(dataloader) # number of batches
testloader = build_dataloader(
test_path,
imgsz_test,
batch_size,
opt, # testloader
hyp=hyp,
cache=opt.cache_images and not opt.notest,
rect=True,
rank=-1,
workers=opt.workers,
pad=0.5,
prefix=colorstr('val: '),
process_list=val_process,
mode='valid')[0]
# if not opt.resume:
# labels = np.concatenate(dataset.labels, 0)
# # c = torch.tensor(labels[:, 0]) # classes
# print(labels)
# if plots:
# plot_labels(labels, save_dir, loggers)
# if tb_writer:
# tb_writer.add_histogram('classes', c, 0)
# Start training
t0 = time.time()
nw = max(
round(hyp['warmup_epochs'] * num_of_batches),
1000) # number of warmup iterations, max(3 epochs, 1k iterations)
# nw = min(nw, (epochs - start_epoch) / 2 * nb) # limit warmup to < 1/2 of training
nc = 1
maps = np.zeros(nc) # mAP per class
results = (0, 0, 0, 0, 0, 0, 0
) # P, R, [email protected], [email protected], val_loss(box, obj, cls)
self.scheduler.last_epoch = start_epoch - 1 # do not move
scaler = amp.GradScaler(enabled=False)
logger.info(f'Image sizes {imgsz} train, {imgsz_test} test\n'
f'Using {dataloader.num_workers} dataloader workers\n'
f'Logging results to {save_dir}\n'
f'Starting training for {epochs} epochs...')
for epoch in range(
start_epoch, epochs
): # epoch ------------------------------------------------------------------
self.model.train()
mloss = torch.zeros(4, device=device) # mean losses
pbar = enumerate(dataloader)
logger.info(
('\n' + '%10s' * 8) % ('Epoch', 'gpu_mem', 'box', 'obj', 'cls',
'total', 'targets', 'img_size'))
pbar = tqdm(pbar, total=num_of_batches) # progress bar
self.optimizer.zero_grad()
for i, batch in pbar: # batch -------------------------------------------------------------
ni = i + num_of_batches * epoch # number integrated batches (since train start)
# if i> 1:
# break
# Warmup
if ni <= nw:
xi = [0, nw] # x interp
# model.gr = np.interp(ni, xi, [0.0, 1.0]) # iou loss ratio (obj_loss = 1.0 or iou)
accumulate = max(
1,
np.interp(ni, xi, [1, nbs / total_batch_size]).round())
for j, x in enumerate(self.optimizer.param_groups):
# bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
x['lr'] = np.interp(ni, xi, [
hyp['warmup_bias_lr'] if j == 2 else 0.0,
x['initial_lr'] * lf(epoch)
])
if 'momentum' in x:
x['momentum'] = np.interp(
ni, xi,
[hyp['warmup_momentum'], hyp['momentum']])
# Multi-scale
# if opt.multi_scale:
# sz = random.randrange(imgsz * 0.5, imgsz * 1.5 + gs) // gs * gs # size
# sf = sz / max(imgs.shape[2:]) # scale factor
# if sf != 1:
# ns = [math.ceil(x * sf / gs) * gs for x in imgs.shape[2:]] # new shape (stretched to gs-multiple)
# imgs = F.interpolate(imgs, size=ns, mode='bilinear', align_corners=False)
# Forward
with amp.autocast(enabled=False):
loss, pred, metrics = self.model.compute_loss(
batch, training=True)
# print(loss)
# Backward
scaler.scale(loss).backward()
# Optimize
if ni % accumulate == 0:
scaler.step(self.optimizer) # optimizer.step
scaler.update()
self.optimizer.zero_grad()
if ema:
ema.update(self.model)
if isinstance(loss, dict):
line = []
loss = torch.tensor(0.).cuda()
for key, l_val in loss.items():
loss += l_val.mean()
line.append('loss_{0}:{1:.4f}'.format(
key, l_val.mean()))
else:
loss = loss.mean()
# Print
# for name, metric in metrics.items():
# print('%s: %6f' % (name, metric.mean()))
mloss = (mloss * i + loss) / (i + 1) # update mean losses
mem = '%.3gG' % (torch.cuda.memory_reserved() / 1E9
if torch.cuda.is_available() else 0) # (GB)
s = ('%10s' * 2 +
'%10.4g' * 6) % ('%g/%g' % (epoch, epochs - 1), mem,
*mloss, 1, batch['image'].shape[-1])
pbar.set_description(s)
# Plot
# if plots and ni < 3:
# f = save_dir / f'train_batch{ni}.jpg' # filename
# Thread(target=plot_images, args=(imgs, targets, paths, f), daemon=True).start()
# # if tb_writer:
# # tb_writer.add_image(f, result, dataformats='HWC', global_step=epoch)
# # tb_writer.add_graph(model, imgs) # add model to tensorboard
# elif plots and ni == 10 and wandb:
# wandb.log({"Mosaics": [wandb.Image(str(x), caption=x.name) for x in save_dir.glob('train*.jpg')
# if x.exists()]}, commit=False)
# end batch ------------------------------------------------------------------------------------------------
# end epoch ----------------------------------------------------------------------------------------------------
# Scheduler
lr = [x['lr']
for x in self.optimizer.param_groups] # for tensorboard
self.scheduler.step()
#ema.update_attr(self.model, include=['yaml', 'nc', 'hyp', 'gr', 'names', 'stride', 'class_weights'])
final_epoch = epoch + 1 == epochs
if not opt.notest or final_epoch: # Calculate mAP
valid_result = self.validor.validate({'test': testloader},
self.model, epoch,
num_of_batches * epoch)
print(valid_result)
# Write
# with open(results_file, 'a') as f:
# f.write(s + '%10.4g' * 7 % results + '\n') # append metrics, val_loss
# if len(opt.name) and opt.bucket:
# os.system('gsutil cp %s gs://%s/results/results%s.txt' % (results_file, opt.bucket, opt.name))
# Log
# tags = ['train/box_loss', 'train/obj_loss', 'train/cls_loss', # train loss
# 'metrics/precision', 'metrics/recall', 'metrics/mAP_0.5', 'metrics/mAP_0.5:0.95',
# 'val/box_loss', 'val/obj_loss', 'val/cls_loss', # val loss
# 'x/lr0', 'x/lr1', 'x/lr2'] # params
# for x, tag in zip(list(mloss[:-1]) + list(results) + lr, tags):
# # if tb_writer:
# # tb_writer.add_scalar(tag, x, epoch) # tensorboard
# if wandb:
# wandb.log({tag: x}, step=epoch, commit=tag == tags[-1]) # W&B
# Update best mAP
#fi = fitness(np.array(results).reshape(1, -1)) # weighted combination of [P, R, [email protected], [email protected]]
# if fi > best_fitness:
# best_fitness = fi
# # Save model
# if (not opt.nosave) or (final_epoch): # if save
# ckpt = {'epoch': epoch,
# 'best_fitness': best_fitness,
# # 'training_results': results_file.read_text(),
# 'model': ema.ema if final_epoch else deepcopy(
# self.model.module if is_parallel(self.model) else self.model).half(),
# 'ema': (deepcopy(ema.ema).half(), ema.updates),
# 'optimizer': self.optimizer.state_dict(),
# 'wandb_id': wandb_run.id if wandb else None}
# # Save last, best and delete
# torch.save(ckpt, last)
# if best_fitness == fi:
# torch.save(ckpt, best)
# del ckpt
# end epoch ----------------------------------------------------------------------------------------------------
# end training
#if rank in [-1, 0]:
# Strip optimizers
final = best if best.exists() else last # final model
for f in last, best:
if f.exists():
strip_optimizer(f)
# Plots
if plots:
plot_results(save_dir=save_dir) # save as results.png
if wandb:
files = [
'results.png', 'confusion_matrix.png',
*[f'{x}_curve.png' for x in ('F1', 'PR', 'P', 'R')]
]
wandb.log({
"Results": [
wandb.Image(str(save_dir / f), caption=f)
for f in files if (save_dir / f).exists()
]
})
if opt.log_artifacts:
wandb.log_artifact(artifact_or_path=str(final),
type='model',
name=save_dir.stem)
# Test best.pt
logger.info('%g epochs completed in %.3f hours.\n' %
(epoch - start_epoch + 1, (time.time() - t0) / 3600))
# if opt.data.endswith('coco.yaml') and nc == 80: # if COCO
# for m in (last, best) if best.exists() else (last): # speed, mAP tests
# results, _, _ = test.test(opt.data,
# batch_size=batch_size * 2,
# imgsz=imgsz_test,
# conf_thres=0.001,
# iou_thres=0.7,
# model=attempt_load(m, device).half(),
# single_cls=opt.single_cls,
# dataloader=testloader,
# save_dir=save_dir,
# save_json=True,
# plots=False)
wandb.run.finish() if wandb and wandb.run else None
torch.cuda.empty_cache()
return results
def train(hyp, opt, device, tb_writer=None, wandb=None):
logger.info(
colorstr("hyperparameters: ") + ", ".join(f"{k}={v}"
for k, v in hyp.items()))
save_dir, epochs, batch_size, total_batch_size, weights, rank = (
Path(opt.save_dir),
opt.epochs,
opt.batch_size,
opt.total_batch_size,
opt.weights,
opt.global_rank,
)
# Directories
wdir = save_dir / "weights"
wdir.mkdir(parents=True, exist_ok=True) # make dir
last = wdir / "last.pt"
best = wdir / "best.pt"
results_file = save_dir / "results.txt"
# Save run settings
with open(save_dir / "hyp.yaml", "w") as f:
yaml.dump(hyp, f, sort_keys=False)
with open(save_dir / "opt.yaml", "w") as f:
# yaml.dump(vars(opt), f, sort_keys=False) # opt 実行パラメータ
yaml.dump(str(opt), f, sort_keys=False)
# Configure
plots = not opt.evolve # create plots
cuda = device.type != "cpu"
init_seeds(2 + rank)
with open(opt.data) as f:
data_dict = yaml.load(f, Loader=yaml.SafeLoader) # data dict
with torch_distributed_zero_first(rank):
check_dataset(data_dict) # check
train_path = data_dict["train"]
test_path = data_dict["val"]
nc = 1 if opt.single_cls else int(data_dict["nc"]) # number of classes
names = (["item"] if opt.single_cls and len(data_dict["names"]) != 1 else
data_dict["names"]) # class names
assert len(names) == nc, "%g names found for nc=%g dataset in %s" % (
len(names),
nc,
opt.data,
) # check
# Model
pretrained = weights.endswith(".pt")
if pretrained:
with torch_distributed_zero_first(rank):
attempt_download(weights) # download if not found locally
ckpt = torch.load(weights, map_location=device) # load checkpoint
if hyp.get("anchors"):
ckpt["model"].yaml["anchors"] = round(
hyp["anchors"]) # force autoanchor
model = Model(opt.cfg or ckpt["model"].yaml, ch=3,
nc=nc).to(device) # create
exclude = ["anchor"] if opt.cfg or hyp.get("anchors") else [
] # exclude keys
state_dict = ckpt["model"].float().state_dict() # to FP32
state_dict = intersect_dicts(state_dict,
model.state_dict(),
exclude=exclude) # intersect
model.load_state_dict(state_dict, strict=False) # load
logger.info(
"Transferred %g/%g items from %s" %
(len(state_dict), len(model.state_dict()), weights)) # report
else:
model = Model(opt.cfg, ch=3, nc=nc).to(device) # create
# Freeze
freeze = [] # parameter names to freeze (full or partial)
for k, v in model.named_parameters():
v.requires_grad = True # train all layers
if any(x in k for x in freeze):
print("freezing %s" % k)
v.requires_grad = False
# Optimizer
nbs = 64 # nominal batch size
accumulate = max(round(nbs / total_batch_size),
1) # accumulate loss before optimizing
hyp["weight_decay"] *= total_batch_size * accumulate / nbs # scale weight_decay
logger.info(f"Scaled weight_decay = {hyp['weight_decay']}")
pg0, pg1, pg2 = [], [], [] # optimizer parameter groups
for k, v in model.named_modules():
if hasattr(v, "bias") and isinstance(v.bias, nn.Parameter):
pg2.append(v.bias) # biases
if isinstance(v, nn.BatchNorm2d):
pg0.append(v.weight) # no decay
elif hasattr(v, "weight") and isinstance(v.weight, nn.Parameter):
pg1.append(v.weight) # apply decay
if opt.adam:
optimizer = optim.Adam(pg0,
lr=hyp["lr0"],
betas=(hyp["momentum"],
0.999)) # adjust beta1 to momentum
else:
optimizer = optim.SGD(pg0,
lr=hyp["lr0"],
momentum=hyp["momentum"],
nesterov=True)
optimizer.add_param_group({
"params": pg1,
"weight_decay": hyp["weight_decay"]
}) # add pg1 with weight_decay
optimizer.add_param_group({"params": pg2}) # add pg2 (biases)
logger.info("Optimizer groups: %g .bias, %g conv.weight, %g other" %
(len(pg2), len(pg1), len(pg0)))
del pg0, pg1, pg2
# Scheduler https://arxiv.org/pdf/1812.01187.pdf
# https://pytorch.org/docs/stable/_modules/torch/optim/lr_scheduler.html#OneCycleLR
if opt.linear_lr:
lf = (lambda x: (1 - x / (epochs - 1)) *
(1.0 - hyp["lrf"]) + hyp["lrf"]) # linear
else:
lf = one_cycle(1, hyp["lrf"], epochs) # cosine 1->hyp['lrf']
scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lf)
# plot_lr_scheduler(optimizer, scheduler, epochs)
# Logging
if rank in [-1, 0] and wandb and wandb.run is None:
opt.hyp = hyp # add hyperparameters
wandb_run = wandb.init(
config=opt,
resume="allow",
project="YOLOv5"
if opt.project == "runs/train" else Path(opt.project).stem,
name=save_dir.stem,
id=ckpt.get("wandb_id") if "ckpt" in locals() else None,
)
loggers = {"wandb": wandb} # loggers dict
# Resume
start_epoch, best_fitness = 0, 0.0
if pretrained:
# Optimizer
if ckpt["optimizer"] is not None:
optimizer.load_state_dict(ckpt["optimizer"])
best_fitness = ckpt["best_fitness"]
# Results
if ckpt.get("training_results") is not None:
with open(results_file, "w") as file:
file.write(ckpt["training_results"]) # write results.txt
# Epochs
start_epoch = ckpt["epoch"] + 1
if opt.resume:
assert (
start_epoch > 0
), "%s training to %g epochs is finished, nothing to resume." % (
weights,
epochs,
)
if epochs < start_epoch:
logger.info(
"%s has been trained for %g epochs. Fine-tuning for %g additional epochs."
% (weights, ckpt["epoch"], epochs))
epochs += ckpt["epoch"] # finetune additional epochs
del ckpt, state_dict
# Image sizes
gs = int(model.stride.max()) # grid size (max stride)
nl = model.model[
-1].nl # number of detection layers (used for scaling hyp['obj'])
imgsz, imgsz_test = [check_img_size(x, gs) for x in opt.img_size
] # verify imgsz are gs-multiples
# DP mode
if cuda and rank == -1 and torch.cuda.device_count() > 1:
model = torch.nn.DataParallel(model)
# SyncBatchNorm
if opt.sync_bn and cuda and rank != -1:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model).to(device)
logger.info("Using SyncBatchNorm()")
# EMA
ema = ModelEMA(model) if rank in [-1, 0] else None
# DDP mode
if cuda and rank != -1:
model = DDP(model,
device_ids=[opt.local_rank],
output_device=opt.local_rank)
# Trainloader
dataloader, dataset = create_dataloader(
train_path,
imgsz,
batch_size,
gs,
opt,
hyp=hyp,
augment=True,
cache=opt.cache_images,
rect=opt.rect,
rank=rank,
world_size=opt.world_size,
workers=opt.workers,
image_weights=opt.image_weights,
quad=opt.quad,
prefix=colorstr("train: "),
)
mlc = np.concatenate(dataset.labels, 0)[:, 0].max() # max label class
nb = len(dataloader) # number of batches
assert (
mlc < nc
), "Label class %g exceeds nc=%g in %s. Possible class labels are 0-%g" % (
mlc,
nc,
opt.data,
nc - 1,
)
# Process 0
if rank in [-1, 0]:
ema.updates = start_epoch * nb // accumulate # set EMA updates
testloader = create_dataloader(
test_path,
imgsz_test,
batch_size * 2,
gs,
opt, # testloader
hyp=hyp,
cache=opt.cache_images and not opt.notest,
rect=True,
rank=-1,
world_size=opt.world_size,
workers=opt.workers,
pad=0.5,
prefix=colorstr("val: "),
)[0]
if not opt.resume:
labels = np.concatenate(dataset.labels, 0)
c = torch.tensor(labels[:, 0]) # classes
# cf = torch.bincount(c.long(), minlength=nc) + 1. # frequency
# model._initialize_biases(cf.to(device))
if plots:
plot_labels(labels, save_dir, loggers)
if tb_writer:
tb_writer.add_histogram("classes", c, 0)
# Anchors
if not opt.noautoanchor:
check_anchors(dataset,
model=model,
thr=hyp["anchor_t"],
imgsz=imgsz)
# Model parameters
hyp["box"] *= 3.0 / nl # scale to layers
hyp["cls"] *= nc / 80.0 * 3.0 / nl # scale to classes and layers
hyp["obj"] *= (imgsz / 640)**2 * 3.0 / nl # scale to image size and layers
model.nc = nc # attach number of classes to model
model.hyp = hyp # attach hyperparameters to model
model.gr = 1.0 # iou loss ratio (obj_loss = 1.0 or iou)
model.class_weights = (
labels_to_class_weights(dataset.labels, nc).to(device) * nc
) # attach class weights
model.names = names
# Start training
t0 = time.time()
nw = max(round(hyp["warmup_epochs"] * nb),
1000) # number of warmup iterations, max(3 epochs, 1k iterations)
# nw = min(nw, (epochs - start_epoch) / 2 * nb) # limit warmup to < 1/2 of training
maps = np.zeros(nc) # mAP per class
results = (0, 0, 0, 0, 0, 0, 0
) # P, R, [email protected], [email protected], val_loss(box, obj, cls)
scheduler.last_epoch = start_epoch - 1 # do not move
scaler = amp.GradScaler(enabled=cuda)
compute_loss = ComputeLoss(model) # init loss class
logger.info(f"Image sizes {imgsz} train, {imgsz_test} test\n"
f"Using {dataloader.num_workers} dataloader workers\n"
f"Logging results to {save_dir}\n"
f"Starting training for {epochs} epochs...")
for epoch in range(
start_epoch, epochs
): # epoch ------------------------------------------------------------------
model.train()
# Update image weights (optional)
if opt.image_weights:
# Generate indices
if rank in [-1, 0]:
cw = (model.class_weights.cpu().numpy() * (1 - maps)**2 / nc
) # class weights
iw = labels_to_image_weights(dataset.labels,
nc=nc,
class_weights=cw) # image weights
dataset.indices = random.choices(
range(dataset.n), weights=iw,
k=dataset.n) # rand weighted idx
# Broadcast if DDP
if rank != -1:
indices = (torch.tensor(dataset.indices)
if rank == 0 else torch.zeros(dataset.n)).int()
dist.broadcast(indices, 0)
if rank != 0:
dataset.indices = indices.cpu().numpy()
# Update mosaic border
# b = int(random.uniform(0.25 * imgsz, 0.75 * imgsz + gs) // gs * gs)
# dataset.mosaic_border = [b - imgsz, -b] # height, width borders
mloss = torch.zeros(4, device=device) # mean losses
if rank != -1:
dataloader.sampler.set_epoch(epoch)
pbar = enumerate(dataloader)
logger.info(
("\n" + "%10s" * 8) % ("Epoch", "gpu_mem", "box", "obj", "cls",
"total", "targets", "img_size"))
if rank in [-1, 0]:
pbar = tqdm(pbar, total=nb) # progress bar
optimizer.zero_grad()
for i, (
imgs,
targets,
paths,
_,
) in (
pbar
): # batch -------------------------------------------------------------
ni = i + nb * epoch # number integrated batches (since train start)
imgs = (imgs.to(device, non_blocking=True).float() / 255.0
) # uint8 to float32, 0-255 to 0.0-1.0
# Warmup
if ni <= nw:
xi = [0, nw] # x interp
# model.gr = np.interp(ni, xi, [0.0, 1.0]) # iou loss ratio (obj_loss = 1.0 or iou)
accumulate = max(
1,
np.interp(ni, xi, [1, nbs / total_batch_size]).round())
for j, x in enumerate(optimizer.param_groups):
# bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
x["lr"] = np.interp(
ni,
xi,
[
hyp["warmup_bias_lr"] if j == 2 else 0.0,
x["initial_lr"] * lf(epoch),
],
)
if "momentum" in x:
x["momentum"] = np.interp(
ni, xi, [hyp["warmup_momentum"], hyp["momentum"]])
# Multi-scale
if opt.multi_scale:
sz = random.randrange(imgsz * 0.5,
imgsz * 1.5 + gs) // gs * gs # size
sf = sz / max(imgs.shape[2:]) # scale factor
if sf != 1:
ns = [math.ceil(x * sf / gs) * gs for x in imgs.shape[2:]
] # new shape (stretched to gs-multiple)
imgs = F.interpolate(imgs,
size=ns,
mode="bilinear",
align_corners=False)
# Forward
with amp.autocast(enabled=cuda):
pred = model(imgs) # forward
loss, loss_items = compute_loss(
pred, targets.to(device)) # loss scaled by batch_size
if rank != -1:
loss *= (opt.world_size
) # gradient averaged between devices in DDP mode
if opt.quad:
loss *= 4.0
# Backward
scaler.scale(loss).backward()
# Optimize
if ni % accumulate == 0:
scaler.step(optimizer) # optimizer.step
scaler.update()
optimizer.zero_grad()
if ema:
ema.update(model)
# Print
if rank in [-1, 0]:
mloss = (mloss * i + loss_items) / (i + 1
) # update mean losses
mem = "%.3gG" % (torch.cuda.memory_reserved() / 1e9
if torch.cuda.is_available() else 0) # (GB)
s = ("%10s" * 2 + "%10.4g" * 6) % (
"%g/%g" % (epoch, epochs - 1),
mem,
*mloss,
targets.shape[0],
imgs.shape[-1],
)
pbar.set_description(s)
# Plot
if plots and ni < 3:
f = save_dir / f"train_batch{ni}.jpg" # filename
Thread(target=plot_images,
args=(imgs, targets, paths, f),
daemon=True).start()
# if tb_writer:
# tb_writer.add_image(f, result, dataformats='HWC', global_step=epoch)
# tb_writer.add_graph(model, imgs) # add model to tensorboard
elif plots and ni == 10 and wandb:
wandb.log(
{
"Mosaics": [
wandb.Image(str(x), caption=x.name)
for x in save_dir.glob("train*.jpg")
if x.exists()
]
},
commit=False,
)
# end batch ------------------------------------------------------------------------------------------------
# end epoch ----------------------------------------------------------------------------------------------------
# Scheduler
lr = [x["lr"] for x in optimizer.param_groups] # for tensorboard
scheduler.step()
# DDP process 0 or single-GPU
if rank in [-1, 0]:
# mAP
if ema:
ema.update_attr(
model,
include=[
"yaml",
"nc",
"hyp",
"gr",
"names",
"stride",
"class_weights",
],
)
final_epoch = epoch + 1 == epochs
if not opt.notest or final_epoch: # Calculate mAP
results, maps, times = test.test(
opt.data,
batch_size=batch_size * 2,
imgsz=imgsz_test,
model=ema.ema,
single_cls=opt.single_cls,
dataloader=testloader,
save_dir=save_dir,
verbose=nc < 50 and final_epoch,
plots=plots and final_epoch,
log_imgs=opt.log_imgs if wandb else 0,
compute_loss=compute_loss,
)
# Write
with open(results_file, "a") as f:
f.write(
s + "%10.4g" * 7 % results +
"\n") # P, R, [email protected], [email protected], val_loss(box, obj, cls)
if len(opt.name) and opt.bucket:
os.system("gsutil cp %s gs://%s/results/results%s.txt" %
(results_file, opt.bucket, opt.name))
# Log
tags = [
"train/box_loss",
"train/obj_loss",
"train/cls_loss", # train loss
"metrics/precision",
"metrics/recall",
"metrics/mAP_0.5",
"metrics/mAP_0.5:0.95",
"val/box_loss",
"val/obj_loss",
"val/cls_loss", # val loss
"x/lr0",
"x/lr1",
"x/lr2",
] # params
for x, tag in zip(list(mloss[:-1]) + list(results) + lr, tags):
if tb_writer:
tb_writer.add_scalar(tag, x, epoch) # tensorboard
if wandb:
wandb.log({tag: x}, step=epoch,
commit=tag == tags[-1]) # W&B
# Update best mAP
fi = fitness(np.array(results).reshape(
1, -1)) # weighted combination of [P, R, [email protected], [email protected]]
if fi > best_fitness:
best_fitness = fi
# Save model
save = (not opt.nosave) or (final_epoch and not opt.evolve)
if save:
with open(results_file, "r") as f: # create checkpoint
ckpt = {
"epoch":
epoch,
"best_fitness":
best_fitness,
"training_results":
f.read(),
"model":
ema.ema,
"optimizer":
None if final_epoch else optimizer.state_dict(),
"wandb_id":
wandb_run.id if wandb else None,
}
# Save last, best and delete
torch.save(ckpt, last)
if best_fitness == fi:
torch.save(ckpt, best)
del ckpt
# end epoch ----------------------------------------------------------------------------------------------------
# end training
if rank in [-1, 0]:
# Strip optimizers
final = best if best.exists() else last # final model
for f in [last, best]:
if f.exists():
strip_optimizer(f) # strip optimizers
if opt.bucket:
os.system(f"gsutil cp {final} gs://{opt.bucket}/weights") # upload
# Plots
if plots:
plot_results(save_dir=save_dir) # save as results.png
if wandb:
files = [
"results.png",
"confusion_matrix.png",
*[f"{x}_curve.png" for x in ("F1", "PR", "P", "R")],
]
wandb.log({
"Results": [
wandb.Image(str(save_dir / f), caption=f)
for f in files if (save_dir / f).exists()
]
})
if opt.log_artifacts:
wandb.log_artifact(artifact_or_path=str(final),
type="model",
name=save_dir.stem)
# Test best.pt
logger.info("%g epochs completed in %.3f hours.\n" %
(epoch - start_epoch + 1, (time.time() - t0) / 3600))
if opt.data.endswith("coco.yaml") and nc == 80: # if COCO
for conf, iou, save_json in (
[0.25, 0.45, False],
[0.001, 0.65, True],
): # speed, mAP tests
results, _, _ = test.test(
opt.data,
batch_size=batch_size * 2,
imgsz=imgsz_test,
conf_thres=conf,
iou_thres=iou,
model=attempt_load(final, device).half(),
single_cls=opt.single_cls,
dataloader=testloader,
save_dir=save_dir,
save_json=save_json,
plots=False,
)
else:
dist.destroy_process_group()
wandb.run.finish() if wandb and wandb.run else None
torch.cuda.empty_cache()
# mlflow
with mlflow.start_run() as run:
# Log args into mlflow
for key, value in hyp.items():
mlflow.log_param(key, value)
for key, value in vars(opt).items():
mlflow.log_param(key, value)
# Log results into mlflow
for x, tag in zip(list(mloss[:-1]) + list(results) + lr, tags):
# xがtorch.Tensorだったらfloatに直す
if torch.is_tensor(x):
x = x.item()
# tag名に特殊記号があれば削除する
if ":" in tag:
tag = re.sub(r":", " ", tag)
mlflow.log_metric(tag, x)
# Log model
mlflow.pytorch.log_model(model, "model")
return results
def train(
hyp, # path/to/hyp.yaml or hyp dictionary
opt,
device,
):
save_dir, epochs, batch_size, weights, single_cls, evolve, data, cfg, resume, noval, nosave, workers, = \
Path(opt.save_dir), opt.epochs, opt.batch_size, opt.weights, opt.single_cls, opt.evolve, opt.data, opt.cfg, \
opt.resume, opt.noval, opt.nosave, opt.workers
# Directories
w = save_dir / 'weights' # weights dir
w.mkdir(parents=True, exist_ok=True) # make dir
last, best, results_file = w / 'last.pt', w / 'best.pt', save_dir / 'results.txt'
# Hyperparameters
if isinstance(hyp, str):
with open(hyp) as f:
hyp = yaml.safe_load(f) # load hyps dict
LOGGER.info(
colorstr('hyperparameters: ') + ', '.join(f'{k}={v}'
for k, v in hyp.items()))
# Save run settings
with open(save_dir / 'hyp.yaml', 'w') as f:
yaml.safe_dump(hyp, f, sort_keys=False)
with open(save_dir / 'opt.yaml', 'w') as f:
yaml.safe_dump(vars(opt), f, sort_keys=False)
# Configure
plots = not evolve # create plots
cuda = device.type != 'cpu'
init_seeds(1 + RANK)
with open(data) as f:
data_dict = yaml.safe_load(f) # data dict
# Loggers
loggers = {'wandb': None, 'tb': None} # loggers dict
if RANK in [-1, 0]:
# TensorBoard
if plots:
prefix = colorstr('tensorboard: ')
LOGGER.info(
f"{prefix}Start with 'tensorboard --logdir {opt.project}', view at http://localhost:6006/"
)
loggers['tb'] = SummaryWriter(str(save_dir))
# W&B
opt.hyp = hyp # add hyperparameters
run_id = torch.load(weights).get('wandb_id') if weights.endswith(
'.pt') and os.path.isfile(weights) else None
run_id = run_id if opt.resume else None # start fresh run if transfer learning
wandb_logger = WandbLogger(opt, save_dir.stem, run_id, data_dict)
loggers['wandb'] = wandb_logger.wandb
if loggers['wandb']:
data_dict = wandb_logger.data_dict
weights, epochs, hyp = opt.weights, opt.epochs, opt.hyp # may update values if resuming
nc = 1 if single_cls else int(data_dict['nc']) # number of classes
names = ['item'] if single_cls and len(
data_dict['names']) != 1 else data_dict['names'] # class names
assert len(
names
) == nc, f'{len(names)} names found for nc={nc} dataset in {data}' # check
is_coco = data.endswith('coco.yaml') and nc == 80 # COCO dataset
# Model
pretrained = weights.endswith('.pt')
if pretrained:
with torch_distributed_zero_first(RANK):
weights = attempt_download(
weights) # download if not found locally
ckpt = torch.load(weights, map_location=device) # load checkpoint
model = Model(cfg or ckpt['model'].yaml,
ch=3,
nc=nc,
anchors=hyp.get('anchors')).to(device) # create
exclude = [
'anchor'
] if (cfg or hyp.get('anchors')) and not resume else [] # exclude keys
csd = ckpt['model'].float().state_dict(
) # checkpoint state_dict as FP32
csd = intersect_dicts(csd, model.state_dict(),
exclude=exclude) # intersect
model.load_state_dict(csd, strict=False) # load
LOGGER.info(
f'Transferred {len(csd)}/{len(model.state_dict())} items from {weights}'
) # report
else:
model = Model(cfg, ch=3, nc=nc,
anchors=hyp.get('anchors')).to(device) # create
with torch_distributed_zero_first(RANK):
check_dataset(data_dict) # check
train_path, val_path = data_dict['train'], data_dict['val']
# Freeze
freeze = [] # parameter names to freeze (full or partial)
for k, v in model.named_parameters():
v.requires_grad = True # train all layers
if any(x in k for x in freeze):
print(f'freezing {k}')
v.requires_grad = False
# Optimizer
nbs = 64 # nominal batch size
accumulate = max(round(nbs / batch_size),
1) # accumulate loss before optimizing
hyp['weight_decay'] *= batch_size * accumulate / nbs # scale weight_decay
LOGGER.info(f"Scaled weight_decay = {hyp['weight_decay']}")
g0, g1, g2 = [], [], [] # optimizer parameter groups
for v in model.modules():
if hasattr(v, 'bias') and isinstance(v.bias, nn.Parameter): # bias
g2.append(v.bias)
if isinstance(v, nn.BatchNorm2d): # weight with decay
g0.append(v.weight)
elif hasattr(v, 'weight') and isinstance(
v.weight, nn.Parameter): # weight without decay
g1.append(v.weight)
if opt.adam:
optimizer = Adam(g0, lr=hyp['lr0'],
betas=(hyp['momentum'],
0.999)) # adjust beta1 to momentum
else:
optimizer = SGD(g0,
lr=hyp['lr0'],
momentum=hyp['momentum'],
nesterov=True)
optimizer.add_param_group({
'params': g1,
'weight_decay': hyp['weight_decay']
}) # add g1 with weight_decay
optimizer.add_param_group({'params': g2}) # add g2 (biases)
LOGGER.info(
f"{colorstr('optimizer:')} {type(optimizer).__name__} with parameter groups "
f"{len(g0)} weight, {len(g1)} weight (no decay), {len(g2)} bias")
del g0, g1, g2
# Scheduler
if opt.linear_lr:
lf = lambda x: (1 - x / (epochs - 1)) * (1.0 - hyp['lrf']) + hyp[
'lrf'] # linear
else:
lf = one_cycle(1, hyp['lrf'], epochs) # cosine 1->hyp['lrf']
scheduler = lr_scheduler.LambdaLR(
optimizer,
lr_lambda=lf) # plot_lr_scheduler(optimizer, scheduler, epochs)
# EMA
ema = ModelEMA(model) if RANK in [-1, 0] else None
# Resume
start_epoch, best_fitness = 0, 0.0
if pretrained:
# Optimizer
if ckpt['optimizer'] is not None:
optimizer.load_state_dict(ckpt['optimizer'])
best_fitness = ckpt['best_fitness']
# EMA
if ema and ckpt.get('ema'):
ema.ema.load_state_dict(ckpt['ema'].float().state_dict())
ema.updates = ckpt['updates']
# Results
if ckpt.get('training_results') is not None:
results_file.write_text(
ckpt['training_results']) # write results.txt
# Epochs
start_epoch = ckpt['epoch'] + 1
if resume:
assert start_epoch > 0, f'{weights} training to {epochs} epochs is finished, nothing to resume.'
if epochs < start_epoch:
LOGGER.info(
f"{weights} has been trained for {ckpt['epoch']} epochs. Fine-tuning for {epochs} more epochs."
)
epochs += ckpt['epoch'] # finetune additional epochs
del ckpt, csd
# Image sizes
gs = max(int(model.stride.max()), 32) # grid size (max stride)
nl = model.model[
-1].nl # number of detection layers (used for scaling hyp['obj'])
imgsz = check_img_size(opt.imgsz, gs,
floor=gs * 2) # verify imgsz is gs-multiple
# DP mode
if cuda and RANK == -1 and torch.cuda.device_count() > 1:
logging.warning(
'DP not recommended, instead use torch.distributed.run for best DDP Multi-GPU results.\n'
'See Multi-GPU Tutorial at https://github.com/ultralytics/yolov5/issues/475 to get started.'
)
model = torch.nn.DataParallel(model)
# SyncBatchNorm
if opt.sync_bn and cuda and RANK != -1:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model).to(device)
LOGGER.info('Using SyncBatchNorm()')
# Trainloader
train_loader, dataset = create_dataloader(train_path,
imgsz,
batch_size // WORLD_SIZE,
gs,
single_cls,
hyp=hyp,
augment=True,
cache=opt.cache_images,
rect=opt.rect,
rank=RANK,
workers=workers,
image_weights=opt.image_weights,
quad=opt.quad,
prefix=colorstr('train: '))
mlc = np.concatenate(dataset.labels, 0)[:, 0].max() # max label class
nb = len(train_loader) # number of batches
assert mlc < nc, f'Label class {mlc} exceeds nc={nc} in {data}. Possible class labels are 0-{nc - 1}'
# Process 0
if RANK in [-1, 0]:
val_loader = create_dataloader(val_path,
imgsz,
batch_size // WORLD_SIZE * 2,
gs,
single_cls,
hyp=hyp,
cache=opt.cache_images and not noval,
rect=True,
rank=-1,
workers=workers,
pad=0.5,
prefix=colorstr('val: '))[0]
if not resume:
labels = np.concatenate(dataset.labels, 0)
# c = torch.tensor(labels[:, 0]) # classes
# cf = torch.bincount(c.long(), minlength=nc) + 1. # frequency
# model._initialize_biases(cf.to(device))
if plots:
plot_labels(labels, names, save_dir, loggers)
# Anchors
if not opt.noautoanchor:
check_anchors(dataset,
model=model,
thr=hyp['anchor_t'],
imgsz=imgsz)
model.half().float() # pre-reduce anchor precision
# DDP mode
if cuda and RANK != -1:
model = DDP(model, device_ids=[LOCAL_RANK], output_device=LOCAL_RANK)
# Model parameters
hyp['box'] *= 3. / nl # scale to layers
hyp['cls'] *= nc / 80. * 3. / nl # scale to classes and layers
hyp['obj'] *= (imgsz / 640)**2 * 3. / nl # scale to image size and layers
hyp['label_smoothing'] = opt.label_smoothing
model.nc = nc # attach number of classes to model
model.hyp = hyp # attach hyperparameters to model
model.class_weights = labels_to_class_weights(
dataset.labels, nc).to(device) * nc # attach class weights
model.names = names
# Start training
t0 = time.time()
nw = max(round(hyp['warmup_epochs'] * nb),
1000) # number of warmup iterations, max(3 epochs, 1k iterations)
# nw = min(nw, (epochs - start_epoch) / 2 * nb) # limit warmup to < 1/2 of training
last_opt_step = -1
maps = np.zeros(nc) # mAP per class
results = (0, 0, 0, 0, 0, 0, 0
) # P, R, [email protected], [email protected], val_loss(box, obj, cls)
scheduler.last_epoch = start_epoch - 1 # do not move
scaler = amp.GradScaler(enabled=cuda)
compute_loss = ComputeLoss(model) # init loss class
LOGGER.info(f'Image sizes {imgsz} train, {imgsz} val\n'
f'Using {train_loader.num_workers} dataloader workers\n'
f'Logging results to {save_dir}\n'
f'Starting training for {epochs} epochs...')
for epoch in range(
start_epoch, epochs
): # epoch ------------------------------------------------------------------
model.train()
# Update image weights (optional)
if opt.image_weights:
# Generate indices
if RANK in [-1, 0]:
cw = model.class_weights.cpu().numpy() * (
1 - maps)**2 / nc # class weights
iw = labels_to_image_weights(dataset.labels,
nc=nc,
class_weights=cw) # image weights
dataset.indices = random.choices(
range(dataset.n), weights=iw,
k=dataset.n) # rand weighted idx
# Broadcast if DDP
if RANK != -1:
indices = (torch.tensor(dataset.indices)
if RANK == 0 else torch.zeros(dataset.n)).int()
dist.broadcast(indices, 0)
if RANK != 0:
dataset.indices = indices.cpu().numpy()
# Update mosaic border
# b = int(random.uniform(0.25 * imgsz, 0.75 * imgsz + gs) // gs * gs)
# dataset.mosaic_border = [b - imgsz, -b] # height, width borders
mloss = torch.zeros(4, device=device) # mean losses
if RANK != -1:
train_loader.sampler.set_epoch(epoch)
pbar = enumerate(train_loader)
LOGGER.info(
('\n' + '%10s' * 8) % ('Epoch', 'gpu_mem', 'box', 'obj', 'cls',
'total', 'labels', 'img_size'))
if RANK in [-1, 0]:
pbar = tqdm(pbar, total=nb) # progress bar
optimizer.zero_grad()
for i, (
imgs, targets, paths, _
) in pbar: # batch -------------------------------------------------------------
ni = i + nb * epoch # number integrated batches (since train start)
imgs = imgs.to(device, non_blocking=True).float(
) / 255.0 # uint8 to float32, 0-255 to 0.0-1.0
# Warmup
if ni <= nw:
xi = [0, nw] # x interp
# compute_loss.gr = np.interp(ni, xi, [0.0, 1.0]) # iou loss ratio (obj_loss = 1.0 or iou)
accumulate = max(
1,
np.interp(ni, xi, [1, nbs / batch_size]).round())
for j, x in enumerate(optimizer.param_groups):
# bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
x['lr'] = np.interp(ni, xi, [
hyp['warmup_bias_lr'] if j == 2 else 0.0,
x['initial_lr'] * lf(epoch)
])
if 'momentum' in x:
x['momentum'] = np.interp(
ni, xi, [hyp['warmup_momentum'], hyp['momentum']])
# Multi-scale
if opt.multi_scale:
sz = random.randrange(imgsz * 0.5,
imgsz * 1.5 + gs) // gs * gs # size
sf = sz / max(imgs.shape[2:]) # scale factor
if sf != 1:
ns = [math.ceil(x * sf / gs) * gs for x in imgs.shape[2:]
] # new shape (stretched to gs-multiple)
imgs = nn.functional.interpolate(imgs,
size=ns,
mode='bilinear',
align_corners=False)
# Forward
with amp.autocast(enabled=cuda):
pred = model(imgs) # forward
loss, loss_items = compute_loss(
pred, targets.to(device)) # loss scaled by batch_size
if RANK != -1:
loss *= WORLD_SIZE # gradient averaged between devices in DDP mode
if opt.quad:
loss *= 4.
# Backward
scaler.scale(loss).backward()
# Optimize
if ni - last_opt_step >= accumulate:
scaler.step(optimizer) # optimizer.step
scaler.update()
optimizer.zero_grad()
if ema:
ema.update(model)
last_opt_step = ni
# Print
if RANK in [-1, 0]:
mloss = (mloss * i + loss_items) / (i + 1
) # update mean losses
mem = f'{torch.cuda.memory_reserved() / 1E9 if torch.cuda.is_available() else 0:.3g}G' # (GB)
s = ('%10s' * 2 +
'%10.4g' * 6) % (f'{epoch}/{epochs - 1}', mem, *mloss,
targets.shape[0], imgs.shape[-1])
pbar.set_description(s)
# Plot
if plots and ni < 3:
f = save_dir / f'train_batch{ni}.jpg' # filename
Thread(target=plot_images,
args=(imgs, targets, paths, f),
daemon=True).start()
if loggers['tb'] and ni == 0: # TensorBoard
with warnings.catch_warnings():
warnings.simplefilter(
'ignore') # suppress jit trace warning
loggers['tb'].add_graph(
torch.jit.trace(de_parallel(model),
imgs[0:1],
strict=False), [])
elif plots and ni == 10 and loggers['wandb']:
wandb_logger.log({
'Mosaics': [
loggers['wandb'].Image(str(x), caption=x.name)
for x in save_dir.glob('train*.jpg') if x.exists()
]
})
# end batch ------------------------------------------------------------------------------------------------
# Scheduler
lr = [x['lr'] for x in optimizer.param_groups] # for loggers
scheduler.step()
# DDP process 0 or single-GPU
if RANK in [-1, 0]:
# mAP
ema.update_attr(model,
include=[
'yaml', 'nc', 'hyp', 'names', 'stride',
'class_weights'
])
final_epoch = epoch + 1 == epochs
if not noval or final_epoch: # Calculate mAP
wandb_logger.current_epoch = epoch + 1
results, maps, _ = val.run(data_dict,
batch_size=batch_size //
WORLD_SIZE * 2,
imgsz=imgsz,
model=ema.ema,
single_cls=single_cls,
dataloader=val_loader,
save_dir=save_dir,
save_json=is_coco and final_epoch,
verbose=nc < 50 and final_epoch,
plots=plots and final_epoch,
wandb_logger=wandb_logger,
compute_loss=compute_loss)
# Write
with open(results_file, 'a') as f:
f.write(s + '%10.4g' * 7 % results +
'\n') # append metrics, val_loss
# Log
tags = [
'train/box_loss',
'train/obj_loss',
'train/cls_loss', # train loss
'metrics/precision',
'metrics/recall',
'metrics/mAP_0.5',
'metrics/mAP_0.5:0.95',
'val/box_loss',
'val/obj_loss',
'val/cls_loss', # val loss
'x/lr0',
'x/lr1',
'x/lr2'
] # params
for x, tag in zip(list(mloss[:-1]) + list(results) + lr, tags):
if loggers['tb']:
loggers['tb'].add_scalar(tag, x, epoch) # TensorBoard
if loggers['wandb']:
wandb_logger.log({tag: x}) # W&B
# Update best mAP
fi = fitness(np.array(results).reshape(
1, -1)) # weighted combination of [P, R, [email protected], [email protected]]
if fi > best_fitness:
best_fitness = fi
wandb_logger.end_epoch(best_result=best_fitness == fi)
# Save model
if (not nosave) or (final_epoch and not evolve): # if save
ckpt = {
'epoch':
epoch,
'best_fitness':
best_fitness,
'training_results':
results_file.read_text(),
'model':
deepcopy(de_parallel(model)).half(),
'ema':
deepcopy(ema.ema).half(),
'updates':
ema.updates,
'optimizer':
optimizer.state_dict(),
'wandb_id':
wandb_logger.wandb_run.id if loggers['wandb'] else None
}
# Save last, best and delete
torch.save(ckpt, last)
if best_fitness == fi:
torch.save(ckpt, best)
if loggers['wandb']:
if ((epoch + 1) % opt.save_period == 0
and not final_epoch) and opt.save_period != -1:
wandb_logger.log_model(last.parent,
opt,
epoch,
fi,
best_model=best_fitness == fi)
del ckpt
# end epoch ----------------------------------------------------------------------------------------------------
# end training -----------------------------------------------------------------------------------------------------
if RANK in [-1, 0]:
LOGGER.info(
f'{epoch - start_epoch + 1} epochs completed in {(time.time() - t0) / 3600:.3f} hours.\n'
)
if plots:
plot_results(save_dir=save_dir) # save as results.png
if loggers['wandb']:
files = [
'results.png', 'confusion_matrix.png',
*[f'{x}_curve.png' for x in ('F1', 'PR', 'P', 'R')]
]
wandb_logger.log({
"Results": [
loggers['wandb'].Image(str(save_dir / f), caption=f)
for f in files if (save_dir / f).exists()
]
})
if not evolve:
if is_coco: # COCO dataset
for m in [last, best
] if best.exists() else [last]: # speed, mAP tests
results, _, _ = val.run(
data_dict,
batch_size=batch_size // WORLD_SIZE * 2,
imgsz=imgsz,
model=attempt_load(m, device).half(),
iou_thres=
0.7, # NMS IoU threshold for best pycocotools results
single_cls=single_cls,
dataloader=val_loader,
save_dir=save_dir,
save_json=True,
plots=False)
# Strip optimizers
for f in last, best:
if f.exists():
strip_optimizer(f) # strip optimizers
if loggers['wandb']: # Log the stripped model
loggers['wandb'].log_artifact(
str(best if best.exists() else last),
type='model',
name='run_' + wandb_logger.wandb_run.id + '_model',
aliases=['latest', 'best', 'stripped'])
wandb_logger.finish_run()
torch.cuda.empty_cache()
return results
