Add cs3darknet_x, cs3sedarknet_l, and darknetaa53 weights from TPU sessions. Move SE btwn conv1 & conv2 in DarkBlock. Improve SE/attn handling in Csp/DarkNet. Fix leaky_relu bug on older csp models.

2 years ago · 05313940e2
parent 4283c0c478
commit 05313940e2
1 changed files with 65 additions and 31 deletions
--- a/timm/models/cspnet.py
+++ b/timm/models/cspnet.py
@ -23,7 +23,7 @@ import torch.nn.functional as F

 from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD
 from .helpers import build_model_with_cfg, named_apply, MATCH_PREV_GROUP
-from .layers import ClassifierHead, ConvNormAct, ConvNormActAa, DropPath, create_attn, create_act_layer, make_divisible
+from .layers import ClassifierHead, ConvNormAct, ConvNormActAa, DropPath, get_attn, create_act_layer, make_divisible
 from .registry import register_model


@ -57,9 +57,10 @@ default_cfgs = {
    'sedarknet21': _cfg(url=''),
    'darknet53': _cfg(
        url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/darknet53_256_c2ns-3aeff817.pth',
-        interpolation='bicubic', test_input_size=(3, 288, 288), test_crop_pct=1.0,
-    ),
-    'darknetaa53': _cfg(url=''),
+        interpolation='bicubic', test_input_size=(3, 288, 288), test_crop_pct=1.0),
+    'darknetaa53': _cfg(
+        url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/darknetaa53_c2ns-5c28ec8a.pth',
+        test_input_size=(3, 288, 288), test_crop_pct=1.0),

    'cs3darknet_s': _cfg(
        url='', interpolation='bicubic'),
@ -71,7 +72,8 @@ default_cfgs = {
        url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/cs3darknet_l_c2ns-16220c5d.pth',
        interpolation='bicubic', test_input_size=(3, 288, 288), test_crop_pct=0.95),
    'cs3darknet_x': _cfg(
-        url=''),
+        url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/cs3darknet_x_c2ns-4e4490aa.pth',
+        interpolation='bicubic', crop_pct=0.95, test_input_size=(3, 288, 288), test_crop_pct=1.0),

    'cs3darknet_focus_s': _cfg(
        url='', interpolation='bicubic'),
@ -84,6 +86,10 @@ default_cfgs = {
    'cs3darknet_focus_x': _cfg(
        url='', interpolation='bicubic'),

+    'cs3sedarknet_l': _cfg(
+        url='https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/cs3sedarknet_l_c2ns-e8d1dc13.pth',
+        interpolation='bicubic', test_input_size=(3, 288, 288), test_crop_pct=0.95),
+
    'cs3sedarknet_xdw': _cfg(
        url='', interpolation='bicubic'),
 }
@ -119,6 +125,7 @@ class CspStagesCfg:
    bottle_ratio: Union[float, Tuple[float, ...]] = 1.  # bottleneck-ratio of blocks in stage
    avg_down: Union[bool, Tuple[bool, ...]] = False
    attn_layer: Optional[Union[str, Tuple[str, ...]]] = None
+    attn_kwargs: Optional[Union[Dict, Tuple[Dict]]] = None
    stage_type: Union[str, Tuple[str]] = 'csp'  # stage type ('csp', 'cs2', 'dark')
    block_type: Union[str, Tuple[str]] = 'bottle'  # blocks type for stages ('bottle', 'dark')

@ -136,6 +143,7 @@ class CspStagesCfg:
        self.bottle_ratio = _pad_arg(self.bottle_ratio, n)
        self.avg_down = _pad_arg(self.avg_down, n)
        self.attn_layer = _pad_arg(self.attn_layer, n)
+        self.attn_kwargs = _pad_arg(self.attn_kwargs, n)
        self.stage_type = _pad_arg(self.stage_type, n)
        self.block_type = _pad_arg(self.block_type, n)

@ -149,12 +157,20 @@ class CspModelCfg:
    stem: CspStemCfg
    stages: CspStagesCfg
    zero_init_last: bool = True  # zero init last weight (usually bn) in residual path
-    act_layer: str = 'relu'
+    act_layer: str = 'leaky_relu'
    norm_layer: str = 'batchnorm'
    aa_layer: Optional[str] = None  # FIXME support string factory for this


-def _cs3darknet_cfg(width_multiplier=1.0, depth_multiplier=1.0, avg_down=False, act_layer='silu', focus=False):
+def _cs3darknet_cfg(
+        width_multiplier=1.0,
+        depth_multiplier=1.0,
+        avg_down=False,
+        act_layer='silu',
+        focus=False,
+        attn_layer=None,
+        attn_kwargs=None,
+):
    if focus:
        stem_cfg = CspStemCfg(
            out_chs=make_divisible(64 * width_multiplier),
@ -172,6 +188,8 @@ def _cs3darknet_cfg(width_multiplier=1.0, depth_multiplier=1.0, avg_down=False,
            bottle_ratio=1.,
            block_ratio=0.5,
            avg_down=avg_down,
+            attn_layer=attn_layer,
+            attn_kwargs=attn_kwargs,
            stage_type='cs3',
            block_type='dark',
        ),
@ -201,7 +219,7 @@ model_cfgs = dict(
            bottle_ratio=0.5,
            block_ratio=1.,
            cross_linear=True,
-        )
+        ),
    ),
    cspresnet50w=CspModelCfg(
        stem=CspStemCfg(out_chs=(32, 32, 64), kernel_size=3, stride=4, pool='max'),
@ -213,7 +231,7 @@ model_cfgs = dict(
            bottle_ratio=0.25,
            block_ratio=0.5,
            cross_linear=True,
-        )
+        ),
    ),
    cspresnext50=CspModelCfg(
        stem=CspStemCfg(out_chs=64, kernel_size=7, stride=4, pool='max'),
@ -226,7 +244,7 @@ model_cfgs = dict(
            bottle_ratio=1.,
            block_ratio=0.5,
            cross_linear=True,
-        )
+        ),
    ),
    cspdarknet53=CspModelCfg(
        stem=CspStemCfg(out_chs=32, kernel_size=3, stride=1, pool=''),
@ -240,7 +258,6 @@ model_cfgs = dict(
            down_growth=True,
            block_type='dark',
        ),
-        act_layer='leaky_relu',
    ),
    darknet17=CspModelCfg(
        stem=CspStemCfg(out_chs=32, kernel_size=3, stride=1, pool=''),
@ -253,7 +270,6 @@ model_cfgs = dict(
            stage_type='dark',
            block_type='dark',
        ),
-        act_layer='leaky_relu',
    ),
    darknet21=CspModelCfg(
        stem=CspStemCfg(out_chs=32, kernel_size=3, stride=1, pool=''),
@ -267,7 +283,6 @@ model_cfgs = dict(
            block_type='dark',

        ),
-        act_layer='leaky_relu',
    ),
    sedarknet21=CspModelCfg(
        stem=CspStemCfg(out_chs=32, kernel_size=3, stride=1, pool=''),
@ -282,7 +297,6 @@ model_cfgs = dict(
            block_type='dark',

        ),
-        act_layer='leaky_relu',
    ),
    darknet53=CspModelCfg(
        stem=CspStemCfg(out_chs=32, kernel_size=3, stride=1, pool=''),
@ -295,7 +309,6 @@ model_cfgs = dict(
            stage_type='dark',
            block_type='dark',
        ),
-        act_layer='leaky_relu',
    ),
    darknetaa53=CspModelCfg(
        stem=CspStemCfg(out_chs=32, kernel_size=3, stride=1, pool=''),
@ -309,7 +322,6 @@ model_cfgs = dict(
            stage_type='dark',
            block_type='dark',
        ),
-        act_layer='leaky_relu',
    ),

    cs3darknet_s=_cs3darknet_cfg(width_multiplier=0.5, depth_multiplier=0.5),
@ -322,6 +334,8 @@ model_cfgs = dict(
    cs3darknet_focus_l=_cs3darknet_cfg(focus=True),
    cs3darknet_focus_x=_cs3darknet_cfg(width_multiplier=1.25, depth_multiplier=1.33, focus=True),

+    cs3sedarknet_l=_cs3darknet_cfg(attn_layer='se', attn_kwargs=dict(rd_ratio=.25)),
+
    cs3sedarknet_xdw=CspModelCfg(
        stem=CspStemCfg(out_chs=(32, 64), kernel_size=3, stride=2, pool=''),
        stages=CspStagesCfg(
@ -333,6 +347,7 @@ model_cfgs = dict(
            block_ratio=0.5,
            attn_layer='se',
        ),
+        act_layer='silu',
    ),
 )

@ -359,14 +374,16 @@ class BottleneckBlock(nn.Module):
        super(BottleneckBlock, self).__init__()
        mid_chs = int(round(out_chs * bottle_ratio))
        ckwargs = dict(act_layer=act_layer, norm_layer=norm_layer)
+        attn_last = attn_layer is not None and attn_last
+        attn_first = attn_layer is not None and not attn_last

        self.conv1 = ConvNormAct(in_chs, mid_chs, kernel_size=1, **ckwargs)
        self.conv2 = ConvNormActAa(
            mid_chs, mid_chs, kernel_size=3, dilation=dilation, groups=groups,
            aa_layer=aa_layer, drop_layer=drop_block, **ckwargs)
-        self.attn2 = create_attn(attn_layer, channels=mid_chs) if not attn_last else None
+        self.attn2 = attn_layer(mid_chs, act_layer=act_layer) if attn_first else nn.Identity()
        self.conv3 = ConvNormAct(mid_chs, out_chs, kernel_size=1, apply_act=False, **ckwargs)
-        self.attn3 = create_attn(attn_layer, channels=out_chs) if attn_last else None
+        self.attn3 = attn_layer(out_chs, act_layer=act_layer) if attn_last else nn.Identity()
        self.drop_path = DropPath(drop_path) if drop_path else nn.Identity()
        self.act3 = create_act_layer(act_layer)

@ -377,11 +394,9 @@ class BottleneckBlock(nn.Module):
        shortcut = x
        x = self.conv1(x)
        x = self.conv2(x)
-        if self.attn2 is not None:
-            x = self.attn2(x)
+        x = self.attn2(x)
        x = self.conv3(x)
-        if self.attn3 is not None:
-            x = self.attn3(x)
+        x = self.attn3(x)
        x = self.drop_path(x) + shortcut
        # FIXME partial shortcut needed if first block handled as per original, not used for my current impl
        #x[:, :shortcut.size(1)] += shortcut
@ -410,11 +425,12 @@ class DarkBlock(nn.Module):
        super(DarkBlock, self).__init__()
        mid_chs = int(round(out_chs * bottle_ratio))
        ckwargs = dict(act_layer=act_layer, norm_layer=norm_layer)
+
        self.conv1 = ConvNormAct(in_chs, mid_chs, kernel_size=1, **ckwargs)
+        self.attn = attn_layer(mid_chs, act_layer=act_layer) if attn_layer is not None else nn.Identity()
        self.conv2 = ConvNormActAa(
            mid_chs, out_chs, kernel_size=3, dilation=dilation, groups=groups,
            aa_layer=aa_layer, drop_layer=drop_block, **ckwargs)
-        self.attn = create_attn(attn_layer, channels=out_chs, act_layer=act_layer)
        self.drop_path = DropPath(drop_path) if drop_path else nn.Identity()

    def zero_init_last(self):
@ -423,9 +439,8 @@ class DarkBlock(nn.Module):
    def forward(self, x):
        shortcut = x
        x = self.conv1(x)
+        x = self.attn(x)
        x = self.conv2(x)
-        if self.attn is not None:
-            x = self.attn(x)
        x = self.drop_path(x) + shortcut
        return x

@ -688,7 +703,8 @@ def create_csp_stem(
    return stem, feature_info


-def _get_stage_fn(stage_type: str, stage_args):
+def _get_stage_fn(stage_args):
+    stage_type = stage_args.pop('stage_type')
    assert stage_type in ('dark', 'csp', 'cs3')
    if stage_type == 'dark':
        stage_args.pop('expand_ratio', None)
@ -702,14 +718,25 @@ def _get_stage_fn(stage_type: str, stage_args):
    return stage_fn, stage_args


-def _get_block_fn(stage_type: str, stage_args):
-    assert stage_type in ('dark', 'bottle')
-    if stage_type == 'dark':
+def _get_block_fn(stage_args):
+    block_type = stage_args.pop('block_type')
+    assert block_type in ('dark', 'bottle')
+    if block_type == 'dark':
        return DarkBlock, stage_args
    else:
        return BottleneckBlock, stage_args


+def _get_attn_fn(stage_args):
+    attn_layer = stage_args.pop('attn_layer')
+    attn_kwargs = stage_args.pop('attn_kwargs', None) or {}
+    if attn_layer is not None:
+        attn_layer = get_attn(attn_layer)
+        if attn_kwargs:
+            attn_layer = partial(attn_layer, **attn_kwargs)
+    return attn_layer, stage_args
+
+
 def create_csp_stages(
        cfg: CspModelCfg,
        drop_path_rate: float,
@ -734,8 +761,9 @@ def create_csp_stages(
    feature_info = []
    stages = []
    for stage_idx, stage_args in enumerate(stage_args):
-        stage_fn, stage_args = _get_stage_fn(stage_args.pop('stage_type'), stage_args)
-        block_fn, stage_args = _get_block_fn(stage_args.pop('block_type'), stage_args)
+        stage_fn, stage_args = _get_stage_fn(stage_args)
+        block_fn, stage_args = _get_block_fn(stage_args)
+        attn_fn, stage_args = _get_attn_fn(stage_args)
        stride = stage_args.pop('stride')
        if stride != 1 and prev_feat:
            feature_info.append(prev_feat)
@ -752,6 +780,7 @@ def create_csp_stages(
            first_dilation=first_dilation,
            dilation=dilation,
            block_fn=block_fn,
+            attn_layer=attn_fn,  # will be passed through stage as block_kwargs
            **block_kwargs,
        )]
        prev_chs = stage_args['out_chs']
@ -968,6 +997,11 @@ def cs3darknet_focus_x(pretrained=False, **kwargs):
    return _create_cspnet('cs3darknet_focus_x', pretrained=pretrained, **kwargs)


+@register_model
+def cs3sedarknet_l(pretrained=False, **kwargs):
+    return _create_cspnet('cs3sedarknet_l', pretrained=pretrained, **kwargs)
+
+
@register_model
 def cs3sedarknet_xdw(pretrained=False, **kwargs):
    return _create_cspnet('cs3sedarknet_xdw', pretrained=pretrained, **kwargs)