fastreid.modeling.backbones.regnet.regnet Namespace Reference

Classes
class	AnyHead
class	AnyNet
class	AnyStage
class	BasicTransform
class	BottleneckTransform
class	RegNet
class	ResBasicBlock
class	ResBottleneckBlock
class	ResStemCifar
class	ResStemIN
class	SE
class	SimpleStemIN
class	VanillaBlock

Functions
	init_weights (m)
	get_stem_fun (stem_type)
	get_block_fun (block_type)
	drop_connect (x, drop_ratio)
	quantize_float (f, q)
	adjust_ws_gs_comp (ws, bms, gs)
	get_stages_from_blocks (ws, rs)
	generate_regnet (w_a, w_0, w_m, d, q=8)
	init_pretrained_weights (key)
	build_regnet_backbone (cfg)

Variables
	logger = logging.getLogger(__name__)
dict	model_urls

Function Documentation

adjust_ws_gs_comp()

fastreid.modeling.backbones.regnet.regnet.adjust_ws_gs_comp	(		ws,
			bms,
			gs )

Adjusts the compatibility of widths and groups.

Definition at line 424 of file regnet.py.

def adjust_ws_gs_comp(ws, bms, gs):
    """Adjusts the compatibility of widths and groups."""
    ws_bot = [int(w * b) for w, b in zip(ws, bms)]
    gs = [min(g, w_bot) for g, w_bot in zip(gs, ws_bot)]
    ws_bot = [quantize_float(w_bot, g) for w_bot, g in zip(ws_bot, gs)]
    ws = [int(w_bot / b) for w_bot, b in zip(ws_bot, bms)]
    return ws, gs
 
 

build_regnet_backbone()

fastreid.modeling.backbones.regnet.regnet.build_regnet_backbone

(

cfg

)

Definition at line 542 of file regnet.py.

def build_regnet_backbone(cfg):
    # 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
    depth         = cfg.MODEL.BACKBONE.DEPTH
    # fmt: on
 
    cfg_files = {
        '800x': 'fastreid/modeling/backbones/regnet/regnetx/RegNetX-800MF_dds_8gpu.yaml',
        '800y': 'fastreid/modeling/backbones/regnet/regnety/RegNetY-800MF_dds_8gpu.yaml',
        '1600x': 'fastreid/modeling/backbones/regnet/regnetx/RegNetX-1.6GF_dds_8gpu.yaml',
        '1600y': 'fastreid/modeling/backbones/regnet/regnety/RegNetY-1.6GF_dds_8gpu.yaml',
        '3200x': 'fastreid/modeling/backbones/regnet/regnetx/RegNetX-3.2GF_dds_8gpu.yaml',
        '3200y': 'fastreid/modeling/backbones/regnet/regnety/RegNetY-3.2GF_dds_8gpu.yaml',
        '4000x': 'fastreid/modeling/backbones/regnet/regnety/RegNetX-4.0GF_dds_8gpu.yaml',
        '4000y': 'fastreid/modeling/backbones/regnet/regnety/RegNetY-4.0GF_dds_8gpu.yaml',
        '6400x': 'fastreid/modeling/backbones/regnet/regnetx/RegNetX-6.4GF_dds_8gpu.yaml',
        '6400y': 'fastreid/modeling/backbones/regnet/regnety/RegNetY-6.4GF_dds_8gpu.yaml',
    }[depth]
 
    regnet_cfg.merge_from_file(cfg_files)
    model = RegNet(last_stride, bn_norm)
 
    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
            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

drop_connect()

fastreid.modeling.backbones.regnet.regnet.drop_connect	(		x,
			drop_ratio )

Drop connect (adapted from DARTS).

Definition at line 72 of file regnet.py.

def drop_connect(x, drop_ratio):
    """Drop connect (adapted from DARTS)."""
    keep_ratio = 1.0 - drop_ratio
    mask = torch.empty([x.shape[0], 1, 1, 1], dtype=x.dtype, device=x.device)
    mask.bernoulli_(keep_ratio)
    x.div_(keep_ratio)
    x.mul_(mask)
    return x
 

generate_regnet()

fastreid.modeling.backbones.regnet.regnet.generate_regnet	(		w_a,
			w_0,
			w_m,
			d,
			q = 8 )

Generates per block ws from RegNet parameters.

Definition at line 442 of file regnet.py.

def generate_regnet(w_a, w_0, w_m, d, q=8):
    """Generates per block ws from RegNet parameters."""
    assert w_a >= 0 and w_0 > 0 and w_m > 1 and w_0 % q == 0
    ws_cont = np.arange(d) * w_a + w_0
    ks = np.round(np.log(ws_cont / w_0) / np.log(w_m))
    ws = w_0 * np.power(w_m, ks)
    ws = np.round(np.divide(ws, q)) * q
    num_stages, max_stage = len(np.unique(ws)), ks.max() + 1
    ws, ws_cont = ws.astype(int).tolist(), ws_cont.tolist()
    return ws, num_stages, max_stage, ws_cont
 
 

get_block_fun()

fastreid.modeling.backbones.regnet.regnet.get_block_fun

(

block_type

)

Retrieves the block function by name.

Definition at line 59 of file regnet.py.

def get_block_fun(block_type):
    """Retrieves the block function by name."""
    block_funs = {
        "vanilla_block": VanillaBlock,
        "res_basic_block": ResBasicBlock,
        "res_bottleneck_block": ResBottleneckBlock,
    }
    assert block_type in block_funs.keys(), "Block type '{}' not supported".format(
        block_type
    )
    return block_funs[block_type]
 
 

get_stages_from_blocks()

fastreid.modeling.backbones.regnet.regnet.get_stages_from_blocks	(		ws,
			rs )

Gets ws/ds of network at each stage from per block values.

Definition at line 433 of file regnet.py.

def get_stages_from_blocks(ws, rs):
    """Gets ws/ds of network at each stage from per block values."""
    ts_temp = zip(ws + [0], [0] + ws, rs + [0], [0] + rs)
    ts = [w != wp or r != rp for w, wp, r, rp in ts_temp]
    s_ws = [w for w, t in zip(ws, ts[:-1]) if t]
    s_ds = np.diff([d for d, t in zip(range(len(ts)), ts) if t]).tolist()
    return s_ws, s_ds
 
 

get_stem_fun()

fastreid.modeling.backbones.regnet.regnet.get_stem_fun

(

stem_type

)

Retrives the stem function by name.

Definition at line 46 of file regnet.py.

def get_stem_fun(stem_type):
    """Retrives the stem function by name."""
    stem_funs = {
        "res_stem_cifar": ResStemCifar,
        "res_stem_in": ResStemIN,
        "simple_stem_in": SimpleStemIN,
    }
    assert stem_type in stem_funs.keys(), "Stem type '{}' not supported".format(
        stem_type
    )
    return stem_funs[stem_type]
 
 

init_pretrained_weights()

fastreid.modeling.backbones.regnet.regnet.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 490 of file regnet.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'))['model_state']
 
    return state_dict
 
 
@BACKBONE_REGISTRY.register()

init_weights()

fastreid.modeling.backbones.regnet.regnet.init_weights

(

m

)

Performs ResNet-style weight initialization.

Definition at line 29 of file regnet.py.

def init_weights(m):
    """Performs ResNet-style weight initialization."""
    if isinstance(m, nn.Conv2d):
        # Note that there is no bias due to BN
        fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
        m.weight.data.normal_(mean=0.0, std=math.sqrt(2.0 / fan_out))
    elif isinstance(m, nn.BatchNorm2d):
        zero_init_gamma = (
                hasattr(m, "final_bn") and m.final_bn and regnet_cfg.BN.ZERO_INIT_FINAL_GAMMA
        )
        m.weight.data.fill_(0.0 if zero_init_gamma else 1.0)
        m.bias.data.zero_()
    elif isinstance(m, nn.Linear):
        m.weight.data.normal_(mean=0.0, std=0.01)
        m.bias.data.zero_()
 
 

quantize_float()

fastreid.modeling.backbones.regnet.regnet.quantize_float	(		f,
			q )

Converts a float to closest non-zero int divisible by q.

Definition at line 419 of file regnet.py.

def quantize_float(f, q):
    """Converts a float to closest non-zero int divisible by q."""
    return int(round(f / q) * q)
 
 

Variable Documentation

logger

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

Definition at line 14 of file regnet.py.

model_urls

dict fastreid.modeling.backbones.regnet.regnet.model_urls

Initial value:

=  {
    '800x': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160905981/RegNetX-200MF_dds_8gpu.pyth',
    '800y': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160906567/RegNetY-800MF_dds_8gpu.pyth',
    '1600x': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160990626/RegNetX-1.6GF_dds_8gpu.pyth',
    '1600y': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160906681/RegNetY-1.6GF_dds_8gpu.pyth',
    '3200x': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160906139/RegNetX-3.2GF_dds_8gpu.pyth',
    '3200y': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160906834/RegNetY-3.2GF_dds_8gpu.pyth',
    '4000x': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160906383/RegNetX-4.0GF_dds_8gpu.pyth',
    '4000y': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160906838/RegNetY-4.0GF_dds_8gpu.pyth',
    '6400x': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/161116590/RegNetX-6.4GF_dds_8gpu.pyth',
    '6400y': 'https://dl.fbaipublicfiles.com/pycls/dds_baselines/160907112/RegNetY-6.4GF_dds_8gpu.pyth',
}

Definition at line 15 of file regnet.py.