fastreid.modeling.backbones.resnet Namespace Reference

Classes
class	BasicBlock
class	Bottleneck
class	ResNet

Functions
	init_pretrained_weights (key)
	build_resnet_backbone (cfg)

Variables
	logger = logging.getLogger(__name__)
dict	model_urls

Detailed Description

@author:  liaoxingyu
@contact: sherlockliao01@gmail.com

Function Documentation

build_resnet_backbone()

fastreid.modeling.backbones.resnet.build_resnet_backbone

(

cfg

)

Create a ResNet instance from config.
Returns:
    ResNet: a :class:`ResNet` instance.

Definition at line 290 of file resnet.py.

def build_resnet_backbone(cfg):
    """
    Create a ResNet instance from config.
    Returns:
        ResNet: a :class:`ResNet` instance.
    """
 
    # fmt: off
    pretrain      = cfg.MODEL.BACKBONE.PRETRAIN
    pretrain_path = cfg.MODEL.BACKBONE.PRETRAIN_PATH
    last_stride   = cfg.MODEL.BACKBONE.LAST_STRIDE
    bn_norm       = cfg.MODEL.BACKBONE.NORM
    with_ibn      = cfg.MODEL.BACKBONE.WITH_IBN
    with_se       = cfg.MODEL.BACKBONE.WITH_SE
    with_nl       = cfg.MODEL.BACKBONE.WITH_NL
    depth         = cfg.MODEL.BACKBONE.DEPTH
    # fmt: on
 
    num_blocks_per_stage = {
        '18x': [2, 2, 2, 2],
        '34x': [3, 4, 6, 3],
        '50x': [3, 4, 6, 3],
        '101x': [3, 4, 23, 3],
        '152x': [3, 8, 36, 3],
    }[depth]
 
    nl_layers_per_stage = {
        '18x': [0, 0, 0, 0],
        '34x': [0, 0, 0, 0],
        '50x': [0, 2, 3, 0],
        '101x': [0, 2, 9, 0],
        '152x': [0, 4, 12, 0]
    }[depth]
 
    block = {
        '18x': BasicBlock,
        '34x': BasicBlock,
        '50x': Bottleneck,
        '101x': Bottleneck,
        '152x': Bottleneck,
    }[depth]
 
    model = ResNet(last_stride, bn_norm, with_ibn, with_se, with_nl, block,
                   num_blocks_per_stage, nl_layers_per_stage)
    if pretrain:
        # Load pretrain path if specifically
        if pretrain_path:
            try:
                state_dict = torch.load(pretrain_path, map_location=torch.device('cpu'))
                logger.info(f"Loading pretrained model from {pretrain_path}")
            except FileNotFoundError as e:
                logger.info(f'{pretrain_path} is not found! Please check this path.')
                raise e
            except KeyError as e:
                logger.info("State dict keys error! Please check the state dict.")
                raise e
        else:
            key = depth
            if with_ibn: key = 'ibn_' + key
            if with_se:  key = 'se_' + key
 
            state_dict = init_pretrained_weights(key)
 
        incompatible = model.load_state_dict(state_dict, strict=False)
        if incompatible.missing_keys:
            logger.info(
                get_missing_parameters_message(incompatible.missing_keys)
            )
        if incompatible.unexpected_keys:
            logger.info(
                get_unexpected_parameters_message(incompatible.unexpected_keys)
            )
 
    return model

init_pretrained_weights()

fastreid.modeling.backbones.resnet.init_pretrained_weights

(

key

)

Initializes model with pretrained weights.

Layers that don't match with pretrained layers in name or size are kept unchanged.

Definition at line 238 of file resnet.py.

def init_pretrained_weights(key):
    """Initializes model with pretrained weights.
 
    Layers that don't match with pretrained layers in name or size are kept unchanged.
    """
    import os
    import errno
    import gdown
 
    def _get_torch_home():
        ENV_TORCH_HOME = 'TORCH_HOME'
        ENV_XDG_CACHE_HOME = 'XDG_CACHE_HOME'
        DEFAULT_CACHE_DIR = '~/.cache'
        torch_home = os.path.expanduser(
            os.getenv(
                ENV_TORCH_HOME,
                os.path.join(
                    os.getenv(ENV_XDG_CACHE_HOME, DEFAULT_CACHE_DIR), 'torch'
                )
            )
        )
        return torch_home
 
    torch_home = _get_torch_home()
    model_dir = os.path.join(torch_home, 'checkpoints')
    try:
        os.makedirs(model_dir)
    except OSError as e:
        if e.errno == errno.EEXIST:
            # Directory already exists, ignore.
            pass
        else:
            # Unexpected OSError, re-raise.
            raise
 
    filename = model_urls[key].split('/')[-1]
 
    cached_file = os.path.join(model_dir, filename)
 
    if not os.path.exists(cached_file):
        if comm.is_main_process():
            gdown.download(model_urls[key], cached_file, quiet=False)
 
    comm.synchronize()
 
    logger.info(f"Loading pretrained model from {cached_file}")
    state_dict = torch.load(cached_file, map_location=torch.device('cpu'))
 
    return state_dict
 
 
@BACKBONE_REGISTRY.register()

Variable Documentation

logger

fastreid.modeling.backbones.resnet.logger = logging.getLogger(__name__)

Definition at line 24 of file resnet.py.

model_urls

dict fastreid.modeling.backbones.resnet.model_urls

Initial value:

=  {
    '18x': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',
    '34x': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',
    '50x': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
    '101x': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',
    '152x': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',
    'ibn_18x': 'https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet18_ibn_a-2f571257.pth',
    'ibn_34x': 'https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet34_ibn_a-94bc1577.pth',
    'ibn_50x': 'https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet50_ibn_a-d9d0bb7b.pth',
    'ibn_101x': 'https://github.com/XingangPan/IBN-Net/releases/download/v1.0/resnet101_ibn_a-59ea0ac6.pth',
    'se_ibn_101x': 'https://github.com/XingangPan/IBN-Net/releases/download/v1.0/se_resnet101_ibn_a-fabed4e2.pth',
}

Definition at line 25 of file resnet.py.