def main(cfg):
OmegaConf.set_struct(cfg, False)
# Get device
device = torch.device("cuda" if (torch.cuda.is_available() and cfg.training.cuda) else "cpu")
log.info("DEVICE : {}".format(device))
# Enable CUDNN BACKEND
torch.backends.cudnn.enabled = cfg.training.enable_cudnn
# Checkpoint
checkpoint = ModelCheckpoint(cfg.training.checkpoint_dir, cfg.model_name, cfg.training.weight_name, strict=True)
# Setup the dataset config
# Generic config
dataset = instantiate_dataset(cfg.data)
model = checkpoint.create_model(dataset, weight_name=cfg.training.weight_name)
log.info(model)
log.info("Model size = %i", sum(param.numel() for param in model.parameters() if param.requires_grad))
log.info(dataset)
model.eval()
if cfg.enable_dropout:
model.enable_dropout_in_eval()
model = model.to(device)
# Run training / evaluation
output_path = os.path.join(cfg.training.checkpoint_dir, cfg.data.name, "features")
if not os.path.exists(output_path):
os.makedirs(output_path, exist_ok=True)
run(model, dataset, device, output_path, cfg)
def main(cfg):
OmegaConf.set_struct(cfg, False)
# Get device
device = torch.device("cuda" if (torch.cuda.is_available() and cfg.training.cuda) else "cpu")
log.info("DEVICE : {}".format(device))
# Enable CUDNN BACKEND
torch.backends.cudnn.enabled = cfg.training.enable_cudnn
# Checkpoint
checkpoint = ModelCheckpoint(cfg.training.checkpoint_dir, cfg.model_name, cfg.training.weight_name, strict=True)
# Setup the dataset config
# Generic config
dataset = instantiate_dataset(cfg.data)
if not checkpoint.is_empty:
model = checkpoint.create_model(dataset, weight_name=cfg.training.weight_name)
else:
log.info("No Checkpoint for this model")
model = instantiate_model(copy.deepcopy(cfg), dataset)
model.set_pretrained_weights()
log.info(model)
log.info("Model size = %i", sum(param.numel() for param in model.parameters() if param.requires_grad))
log.info(dataset)
model.eval()
if cfg.enable_dropout:
model.enable_dropout_in_eval()
model = model.to(device)
run(model, dataset, device, cfg)
def main(cfg):
OmegaConf.set_struct(cfg, False)
# Get device
device = torch.device("cuda" if (
torch.cuda.is_available() and cfg.cuda) else "cpu")
log.info("DEVICE : {}".format(device))
# Enable CUDNN BACKEND
torch.backends.cudnn.enabled = cfg.enable_cudnn
# Checkpoint
checkpoint = ModelCheckpoint(cfg.checkpoint_dir,
cfg.model_name,
cfg.weight_name,
strict=True)
# Setup the dataset config
# Generic config
train_dataset_cls = get_dataset_class(checkpoint.data_config)
setattr(checkpoint.data_config, "class", train_dataset_cls.FORWARD_CLASS)
setattr(checkpoint.data_config, "dataroot", cfg.input_path)
# Datset specific configs
if cfg.data:
for key, value in cfg.data.items():
checkpoint.data_config.update(key, value)
# Create dataset and mdoel
dataset = instantiate_dataset(checkpoint.data_config)
model = checkpoint.create_model(dataset, weight_name=cfg.weight_name)
log.info(model)
log.info(
"Model size = %i",
sum(param.numel() for param in model.parameters()
if param.requires_grad))
# Set dataloaders
dataset.create_dataloaders(
model,
cfg.batch_size,
cfg.shuffle,
cfg.num_workers,
False,
)
log.info(dataset)
model.eval()
if cfg.enable_dropout:
model.enable_dropout_in_eval()
model = model.to(device)
# Run training / evaluation
if not os.path.exists(cfg.output_path):
os.makedirs(cfg.output_path)
run(model, dataset, device, cfg.output_path)
def main(cfg):
OmegaConf.set_struct(cfg, False)
# Get device
device = torch.device("cuda" if (
torch.cuda.is_available() and cfg.cuda) else "cpu")
log.info("DEVICE : {}".format(device))
# Enable CUDNN BACKEND
torch.backends.cudnn.enabled = cfg.enable_cudnn
# Checkpoint
checkpoint = ModelCheckpoint(cfg.checkpoint_dir,
cfg.model_name,
cfg.weight_name,
strict=True)
# Create model and datasets
dataset = instantiate_dataset(checkpoint.data_config)
model = checkpoint.create_model(dataset, weight_name=cfg.weight_name)
log.info(model)
log.info(
"Model size = %i",
sum(param.numel() for param in model.parameters()
if param.requires_grad))
# Set dataloaders
dataset.create_dataloaders(
model,
cfg.batch_size,
cfg.shuffle,
cfg.num_workers,
cfg.precompute_multi_scale,
)
log.info(dataset)
model.eval()
if cfg.enable_dropout:
model.enable_dropout_in_eval()
model = model.to(device)
tracker: BaseTracker = dataset.get_tracker(model, dataset, False, False)
# Run training / evaluation
run(
cfg,
model,
dataset,
device,
tracker,
checkpoint,
voting_runs=cfg.voting_runs,
tracker_options=cfg.tracker_options,
)
def __init__(self, checkpoint_dir, model_name, weight_name, feat_name, num_classes=None, mock_dataset=True):
# Checkpoint
from torch_points3d.datasets.base_dataset import BaseDataset
from torch_points3d.datasets.dataset_factory import instantiate_dataset
from torch_points3d.utils.mock import MockDataset
import torch_points3d.metrics.model_checkpoint as model_checkpoint
checkpoint = model_checkpoint.ModelCheckpoint(checkpoint_dir, model_name, weight_name, strict=True)
if mock_dataset:
dataset = MockDataset(num_classes)
dataset.num_classes = num_classes
else:
dataset = instantiate_dataset(checkpoint.data_config)
BaseDataset.set_transform(self, checkpoint.data_config)
self.model = checkpoint.create_model(dataset, weight_name=weight_name)
self.model.eval()
def from_pretrained(model_tag,
download=True,
out_file=None,
weight_name="latest",
mock_dataset=True):
# Convert inputs to registry format
if PretainedRegistry.MODELS.get(model_tag) is not None:
url = PretainedRegistry.MODELS.get(model_tag)
else:
raise Exception(
"model_tag {} doesn't exist within available models. Here is the list of pre-trained models {}"
.format(model_tag, PretainedRegistry.available_models()))
checkpoint_name = model_tag + ".pt"
out_file = os.path.join(CHECKPOINT_DIR, checkpoint_name)
if download:
download_file(url, out_file)
weight_name = weight_name if weight_name is not None else "latest"
checkpoint: ModelCheckpoint = ModelCheckpoint(
CHECKPOINT_DIR,
model_tag,
weight_name if weight_name is not None else "latest",
resume=False,
)
if mock_dataset:
dataset = checkpoint.dataset_properties.copy()
if PretainedRegistry.MOCK_USED_PROPERTIES.get(
model_tag) is not None:
for k, v in PretainedRegistry.MOCK_USED_PROPERTIES.get(
model_tag).items():
dataset[k] = v
else:
dataset = instantiate_dataset(checkpoint.data_config)
model: BaseModel = checkpoint.create_model(dataset,
weight_name=weight_name)
Wandb.set_urls_to_model(model, url)
BaseDataset.set_transform(model, checkpoint.data_config)
return model
def main(cfg):
OmegaConf.set_struct(
cfg,
False) # This allows getattr and hasattr methods to function correctly
if cfg.pretty_print:
print(cfg.pretty())
set_debugging_vars_to_global(cfg.debugging)
# Get device
device = torch.device("cuda" if (
torch.cuda.is_available() and cfg.training.cuda) else "cpu")
log.info("DEVICE : {}".format(device))
# Enable CUDNN BACKEND
torch.backends.cudnn.enabled = cfg.training.enable_cudnn
dataset = instantiate_dataset(cfg.data)
model = instantiate_model(cfg, dataset)
log.info(model)
log.info(
"Model size = %i",
sum(param.numel() for param in model.parameters()
if param.requires_grad))
# Set dataloaders
dataset.create_dataloaders(
model,
cfg.training.batch_size,
cfg.training.shuffle,
cfg.training.num_workers,
cfg.training.precompute_multi_scale,
)
log.info(dataset)
# Run training / evaluation
model = model.to(device)
measurement_name = "{}_{}".format(cfg.model_name,
dataset.__class__.__name__)
run(cfg, model, dataset, device, measurement_name)
def from_file(path, weight_name="latest", mock_property=None):
"""
Load a pretrained model trained with torch-points3d from file.
return a pretrained model
Parameters
----------
path: str
path of a pretrained model
weight_name: str, optional
name of the weight
mock_property: dict, optional
mock dataset
"""
weight_name = weight_name if weight_name is not None else "latest"
path_dir, name = os.path.split(path)
name = name.split(".")[0] # ModelCheckpoint will add the extension
checkpoint: ModelCheckpoint = ModelCheckpoint(
path_dir,
name,
weight_name if weight_name is not None else "latest",
resume=False,
)
dataset = checkpoint.data_config
if mock_property is not None:
for k, v in mock_property.items():
dataset[k] = v
else:
dataset = instantiate_dataset(checkpoint.data_config)
model: BaseModel = checkpoint.create_model(dataset,
weight_name=weight_name)
BaseDataset.set_transform(model, checkpoint.data_config)
return model
def _initialize_trainer(self):
# Enable CUDNN BACKEND
torch.backends.cudnn.enabled = self.enable_cudnn
if not self.has_training:
self._cfg.training = self._cfg
resume = bool(self._cfg.checkpoint_dir)
else:
resume = bool(self._cfg.training.checkpoint_dir)
# Get device
if self._cfg.training.cuda > -1 and torch.cuda.is_available():
device = "cuda"
torch.cuda.set_device(self._cfg.training.cuda)
else:
device = "cpu"
self._device = torch.device(device)
log.info("DEVICE : {}".format(self._device))
# Profiling
if self.profiling:
# Set the num_workers as torch.utils.bottleneck doesn't work well with it
self._cfg.training.num_workers = 0
# Start Wandb if public
if self.wandb_log:
Wandb.launch(self._cfg, self._cfg.wandb.public and self.wandb_log)
# Checkpoint
self._checkpoint: ModelCheckpoint = ModelCheckpoint(
self._cfg.training.checkpoint_dir,
self._cfg.model_name,
self._cfg.training.weight_name,
run_config=self._cfg,
resume=resume,
)
# Create model and datasets
if not self._checkpoint.is_empty:
self._dataset: BaseDataset = instantiate_dataset(
self._checkpoint.data_config)
self._model: BaseModel = self._checkpoint.create_model(
self._dataset, weight_name=self._cfg.training.weight_name)
else:
self._dataset: BaseDataset = instantiate_dataset(self._cfg.data)
self._model: BaseModel = instantiate_model(
copy.deepcopy(self._cfg), self._dataset)
self._model.instantiate_optimizers(self._cfg, "cuda" in device)
self._model.set_pretrained_weights()
if not self._checkpoint.validate(self._dataset.used_properties):
log.warning(
"The model will not be able to be used from pretrained weights without the corresponding dataset. Current properties are {}"
.format(self._dataset.used_properties))
self._checkpoint.dataset_properties = self._dataset.used_properties
log.info(self._model)
self._model.log_optimizers()
log.info(
"Model size = %i",
sum(param.numel() for param in self._model.parameters()
if param.requires_grad))
# Set dataloaders
self._dataset.create_dataloaders(
self._model,
self._cfg.training.batch_size,
self._cfg.training.shuffle,
self._cfg.training.num_workers,
self.precompute_multi_scale,
)
log.info(self._dataset)
# Verify attributes in dataset
self._model.verify_data(self._dataset.train_dataset[0])
# Choose selection stage
selection_stage = getattr(self._cfg, "selection_stage", "")
self._checkpoint.selection_stage = self._dataset.resolve_saving_stage(
selection_stage)
self._tracker: BaseTracker = self._dataset.get_tracker(
self.wandb_log, self.tensorboard_log)
if self.wandb_log:
Wandb.launch(self._cfg, not self._cfg.wandb.public
and self.wandb_log)
# Run training / evaluation
self._model = self._model.to(self._device)
if self.has_visualization:
self._visualizer = Visualizer(self._cfg.visualization,
self._dataset.num_batches,
self._dataset.batch_size,
os.getcwd())
def main(cfg):
OmegaConf.set_struct(
cfg,
False) # This allows getattr and hasattr methods to function correctly
if cfg.pretty_print:
print(cfg.pretty())
# Get device
device = torch.device("cuda" if (
torch.cuda.is_available() and cfg.training.cuda) else "cpu")
log.info("DEVICE : {}".format(device))
# Enable CUDNN BACKEND
torch.backends.cudnn.enabled = cfg.training.enable_cudnn
# Profiling
profiling = getattr(cfg.debugging, "profiling", False)
if profiling:
# Set the num_workers as torch.utils.bottleneck doesn't work well with it
cfg.training.num_workers = 0
# Start Wandb if public
launch_wandb(cfg, cfg.wandb.public and cfg.wandb.log)
# Checkpoint
checkpoint = ModelCheckpoint(
cfg.training.checkpoint_dir,
cfg.model_name,
cfg.training.weight_name,
run_config=cfg,
resume=bool(cfg.training.checkpoint_dir),
)
# Create model and datasets
if not checkpoint.is_empty:
dataset = instantiate_dataset(checkpoint.data_config)
model = checkpoint.create_model(dataset,
weight_name=cfg.training.weight_name)
else:
dataset = instantiate_dataset(cfg.data)
model = instantiate_model(cfg, dataset)
model.instantiate_optimizers(cfg)
log.info(model)
model.log_optimizers()
log.info(
"Model size = %i",
sum(param.numel() for param in model.parameters()
if param.requires_grad))
# Set dataloaders
dataset.create_dataloaders(
model,
cfg.training.batch_size,
cfg.training.shuffle,
cfg.training.num_workers,
cfg.training.precompute_multi_scale,
)
log.info(dataset)
# Choose selection stage
selection_stage = getattr(cfg, "selection_stage", "")
checkpoint.selection_stage = dataset.resolve_saving_stage(selection_stage)
tracker: BaseTracker = dataset.get_tracker(model, dataset, cfg.wandb.log,
cfg.tensorboard.log)
launch_wandb(cfg, not cfg.wandb.public and cfg.wandb.log)
# Run training / evaluation
model = model.to(device)
visualizer = Visualizer(cfg.visualization, dataset.num_batches,
dataset.batch_size, os.getcwd())
run(cfg, model, dataset, device, tracker, checkpoint, visualizer)
# https://github.com/facebookresearch/hydra/issues/440
hydra._internal.hydra.GlobalHydra.get_state().clear()
return 0
