Config

packnet_sfm.utils.config.backwards_config(config)[source]

Add or update configuration for backwards compatibility (no need for it right now, pretrained models are up-to-date with configuration files).

Parameters

config (CfgNode) – Model configuration

Returns

config – Updated model configuration

Return type

CfgNode

packnet_sfm.utils.config.get_default_config(cfg_default)[source]

Get default configuration from file

packnet_sfm.utils.config.merge_cfg_file(config, cfg_file=None)[source]

Merge configuration file

packnet_sfm.utils.config.merge_cfgs(original, override)[source]

Updates CfgNode with information from another one

Parameters
  • original (CfgNode) – Original configuration node

  • override (CfgNode) – Another configuration node used for overriding

Returns

updated – Updated configuration node

Return type

CfgNode

packnet_sfm.utils.config.parse_test_config(ckpt_file, cfg_default, cfg_file)[source]

Parse model configuration for testing

Parameters
  • ckpt_file (str) – Checkpoint file, with pretrained model

  • cfg_default (str) – Default configuration file, with default values

  • cfg_file (str) – Configuration file with updated information

Returns

  • config (CfgNode) – Parsed model configuration

  • state_dict (dict) – Model state dict with pretrained weights

packnet_sfm.utils.config.parse_test_file(ckpt_file, cfg_file=None)[source]

Parse model configuration for testing

Parameters
  • ckpt_file (str) – Checkpoint file, with pretrained model

  • cfg_file – Configuration file, to update pretrained model configuration

Returns

  • config (CfgNode) – Parsed model configuration

  • state_dict (dict) – Model state dict with pretrained weights

packnet_sfm.utils.config.parse_train_config(cfg_default, cfg_file)[source]

Parse model configuration for training

Parameters
  • cfg_default (str) – Default .py configuration file

  • cfg_file (str) – Configuration .yaml file to override the default parameters

Returns

config – Parsed model configuration

Return type

CfgNode

packnet_sfm.utils.config.parse_train_file(file)[source]

Parse file for training

Parameters

file (str) – File, can be either a .yaml for a yacs configuration file or a .ckpt for a pre-trained checkpoint file

Returns

  • config (CfgNode) – Parsed model configuration

  • ckpt (str) – Parsed checkpoint file

packnet_sfm.utils.config.prep_dataset(config)[source]

Expand dataset configuration to match split length

Parameters

config (CfgNode) – Dataset configuration

Returns

config – Updated dataset configuration

Return type

CfgNode

packnet_sfm.utils.config.prepare_test_config(config)[source]

Prepare model configuration for testing

Parameters

config (CfgNode) – Model configuration

Returns

config – Prepared model configuration

Return type

CfgNode

packnet_sfm.utils.config.prepare_train_config(config)[source]

Prepare model configuration for training

Parameters

config (CfgNode) – Model configuration

Returns

config – Prepared model configuration

Return type

CfgNode

packnet_sfm.utils.config.set_checkpoint(config)[source]

Set checkpoint information

Parameters

config (CfgNode) – Model configuration

Returns

config – Updated model configuration

Return type

CfgNode

packnet_sfm.utils.config.set_name(config)[source]

Set run name based on available information

Parameters

config (CfgNode) – Model configuration

Returns

name – Updated run name

Return type

str