# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. """ This code is based on https://github.com/microsoft/FocalNet/blob/main/classification/focalnet.py """ import numpy as np import paddle import paddle.nn as nn import paddle.nn.functional as F from ppdet.modeling.shape_spec import ShapeSpec from ppdet.core.workspace import register, serializable from .transformer_utils import DropPath, Identity from .transformer_utils import add_parameter, to_2tuple from .transformer_utils import ones_, zeros_, trunc_normal_ from .swin_transformer import Mlp __all__ = ['FocalNet'] MODEL_cfg = { 'focalnet_T_224_1k_srf': dict( embed_dim=96, depths=[2, 2, 6, 2], focal_levels=[2, 2, 2, 2], focal_windows=[3, 3, 3, 3], drop_path_rate=0.2, use_conv_embed=False, use_postln=False, use_postln_in_modulation=False, use_layerscale=False, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_tiny_srf_pretrained.pdparams', ), 'focalnet_S_224_1k_srf': dict( embed_dim=96, depths=[2, 2, 18, 2], focal_levels=[2, 2, 2, 2], focal_windows=[3, 3, 3, 3], drop_path_rate=0.3, use_conv_embed=False, use_postln=False, use_postln_in_modulation=False, use_layerscale=False, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_small_srf_pretrained.pdparams', ), 'focalnet_B_224_1k_srf': dict( embed_dim=128, depths=[2, 2, 18, 2], focal_levels=[2, 2, 2, 2], focal_windows=[3, 3, 3, 3], drop_path_rate=0.5, use_conv_embed=False, use_postln=False, use_postln_in_modulation=False, use_layerscale=False, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_base_srf_pretrained.pdparams', ), 'focalnet_T_224_1k_lrf': dict( embed_dim=96, depths=[2, 2, 6, 2], focal_levels=[3, 3, 3, 3], focal_windows=[3, 3, 3, 3], drop_path_rate=0.2, use_conv_embed=False, use_postln=False, use_postln_in_modulation=False, use_layerscale=False, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_tiny_lrf_pretrained.pdparams', ), 'focalnet_S_224_1k_lrf': dict( embed_dim=96, depths=[2, 2, 18, 2], focal_levels=[3, 3, 3, 3], focal_windows=[3, 3, 3, 3], drop_path_rate=0.3, use_conv_embed=False, use_postln=False, use_postln_in_modulation=False, use_layerscale=False, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_small_lrf_pretrained.pdparams', ), 'focalnet_B_224_1k_lrf': dict( embed_dim=128, depths=[2, 2, 18, 2], focal_levels=[3, 3, 3, 3], focal_windows=[3, 3, 3, 3], drop_path_rate=0.5, use_conv_embed=False, use_postln=False, use_postln_in_modulation=False, use_layerscale=False, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_base_lrf_pretrained.pdparams', ), 'focalnet_L_384_22k_fl3': dict( embed_dim=192, depths=[2, 2, 18, 2], focal_levels=[3, 3, 3, 3], focal_windows=[5, 5, 5, 5], drop_path_rate=0.5, use_conv_embed=True, use_postln=True, use_postln_in_modulation=False, use_layerscale=True, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_large_lrf_384_pretrained.pdparams', ), 'focalnet_L_384_22k_fl4': dict( embed_dim=192, depths=[2, 2, 18, 2], focal_levels=[4, 4, 4, 4], focal_windows=[3, 3, 3, 3], drop_path_rate=0.5, use_conv_embed=True, use_postln=True, use_postln_in_modulation=False, use_layerscale=True, normalize_modulator=True, # pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_large_lrf_384_fl4_pretrained.pdparams', ), 'focalnet_XL_384_22k_fl3': dict( embed_dim=256, depths=[2, 2, 18, 2], focal_levels=[3, 3, 3, 3], focal_windows=[5, 5, 5, 5], drop_path_rate=0.5, use_conv_embed=True, use_postln=True, use_postln_in_modulation=False, use_layerscale=True, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_xlarge_lrf_384_pretrained.pdparams', ), 'focalnet_XL_384_22k_fl4': dict( embed_dim=256, depths=[2, 2, 18, 2], focal_levels=[4, 4, 4, 4], focal_windows=[3, 3, 3, 3], drop_path_rate=0.5, use_conv_embed=True, use_postln=True, use_postln_in_modulation=False, use_layerscale=True, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_xlarge_lrf_384_fl4_pretrained.pdparams', ), 'focalnet_H_224_22k_fl3': dict( embed_dim=352, depths=[2, 2, 18, 2], focal_levels=[3, 3, 3, 3], focal_windows=[3, 3, 3, 3], drop_path_rate=0.5, use_conv_embed=True, use_postln=True, use_postln_in_modulation=True, # use_layerscale=True, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_huge_lrf_224_pretrained.pdparams', ), 'focalnet_H_224_22k_fl4': dict( embed_dim=352, depths=[2, 2, 18, 2], focal_levels=[4, 4, 4, 4], focal_windows=[3, 3, 3, 3], drop_path_rate=0.5, use_conv_embed=True, use_postln=True, use_postln_in_modulation=True, # use_layerscale=True, normalize_modulator=False, pretrained='https://bj.bcebos.com/v1/paddledet/models/pretrained/focalnet_huge_lrf_224_fl4_pretrained.pdparams', ), } class FocalModulation(nn.Layer): """ Args: dim (int): Number of input channels. proj_drop (float, optional): Dropout ratio of output. Default: 0.0 focal_level (int): Number of focal levels focal_window (int): Focal window size at focal level 1 focal_factor (int): Step to increase the focal window. Default: 2 use_postln_in_modulation (bool): Whether use post-modulation layernorm normalize_modulator (bool): Whether use normalize in modulator """ def __init__(self, dim, proj_drop=0., focal_level=2, focal_window=7, focal_factor=2, use_postln_in_modulation=False, normalize_modulator=False): super().__init__() self.dim = dim # specific args for focalv3 self.focal_level = focal_level self.focal_window = focal_window self.focal_factor = focal_factor self.use_postln_in_modulation = use_postln_in_modulation self.normalize_modulator = normalize_modulator self.f = nn.Linear( dim, 2 * dim + (self.focal_level + 1), bias_attr=True) self.h = nn.Conv2D( dim, dim, kernel_size=1, stride=1, padding=0, groups=1, bias_attr=True) self.act = nn.GELU() self.proj = nn.Linear(dim, dim) self.proj_drop = nn.Dropout(proj_drop) self.focal_layers = nn.LayerList() if self.use_postln_in_modulation: self.ln = nn.LayerNorm(dim) for k in range(self.focal_level): kernel_size = self.focal_factor * k + self.focal_window self.focal_layers.append( nn.Sequential( nn.Conv2D( dim, dim, kernel_size=kernel_size, stride=1, groups=dim, padding=kernel_size // 2, bias_attr=False), nn.GELU())) def forward(self, x): """ Forward function. Args: x: input features with shape of (B, H, W, C) """ _, _, _, C = x.shape x = self.f(x) x = x.transpose([0, 3, 1, 2]) q, ctx, gates = paddle.split(x, (C, C, self.focal_level + 1), 1) ctx_all = 0 for l in range(self.focal_level): ctx = self.focal_layers[l](ctx) ctx_all = ctx_all + ctx * gates[:, l:l + 1] ctx_global = self.act(ctx.mean(2, keepdim=True).mean(3, keepdim=True)) ctx_all = ctx_all + ctx_global * gates[:, self.focal_level:] if self.normalize_modulator: ctx_all = ctx_all / (self.focal_level + 1) x_out = q * self.h(ctx_all) x_out = x_out.transpose([0, 2, 3, 1]) if self.use_postln_in_modulation: x_out = self.ln(x_out) x_out = self.proj(x_out) x_out = self.proj_drop(x_out) return x_out class FocalModulationBlock(nn.Layer): """ Focal Modulation Block. Args: dim (int): Number of input channels. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. drop (float, optional): Dropout rate. Default: 0.0 drop_path (float, optional): Stochastic depth rate. Default: 0.0 act_layer (nn.Layer, optional): Activation layer. Default: nn.GELU norm_layer (nn.Layer, optional): Normalization layer. Default: nn.LayerNorm focal_level (int): number of focal levels focal_window (int): focal kernel size at level 1 use_postln (bool): Whether use layernorm after modulation. Default: False. use_postln_in_modulation (bool): Whether use post-modulation layernorm. Default: False. normalize_modulator (bool): Whether use normalize in modulator use_layerscale (bool): Whether use layerscale proposed in CaiT. Default: False layerscale_value (float): Value for layer scale. Default: 1e-4 """ def __init__(self, dim, mlp_ratio=4., drop=0., drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, focal_level=2, focal_window=9, use_postln=False, use_postln_in_modulation=False, normalize_modulator=False, use_layerscale=False, layerscale_value=1e-4): super().__init__() self.dim = dim self.mlp_ratio = mlp_ratio self.focal_window = focal_window self.focal_level = focal_level self.use_postln = use_postln self.use_layerscale = use_layerscale self.norm1 = norm_layer(dim) self.modulation = FocalModulation( dim, proj_drop=drop, focal_level=self.focal_level, focal_window=self.focal_window, use_postln_in_modulation=use_postln_in_modulation, normalize_modulator=normalize_modulator) self.drop_path = DropPath(drop_path) if drop_path > 0. else Identity() self.norm2 = norm_layer(dim) mlp_hidden_dim = int(dim * mlp_ratio) self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) self.H = None self.W = None self.gamma_1 = 1.0 self.gamma_2 = 1.0 if self.use_layerscale: self.gamma_1 = add_parameter(self, layerscale_value * paddle.ones([dim])) self.gamma_2 = add_parameter(self, layerscale_value * paddle.ones([dim])) def forward(self, x): """ Args: x: Input feature, tensor size (B, H*W, C). """ B, L, C = x.shape H, W = self.H, self.W assert L == H * W, "input feature has wrong size" shortcut = x if not self.use_postln: x = self.norm1(x) x = x.reshape([-1, H, W, C]) # FM x = self.modulation(x).reshape([-1, H * W, C]) if self.use_postln: x = self.norm1(x) # FFN x = shortcut + self.drop_path(self.gamma_1 * x) if self.use_postln: x = x + self.drop_path(self.gamma_2 * self.norm2(self.mlp(x))) else: x = x + self.drop_path(self.gamma_2 * self.mlp(self.norm2(x))) return x class BasicLayer(nn.Layer): """ A basic focal modulation layer for one stage. Args: dim (int): Number of feature channels depth (int): Depths of this stage. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4. drop (float, optional): Dropout rate. Default: 0.0 drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 norm_layer (nn.Layer, optional): Normalization layer. Default: nn.LayerNorm downsample (nn.Layer | None, optional): Downsample layer at the end of the layer. Default: None focal_level (int): Number of focal levels focal_window (int): Focal window size at focal level 1 use_conv_embed (bool): Whether use overlapped convolution for patch embedding use_layerscale (bool): Whether use layerscale proposed in CaiT. Default: False layerscale_value (float): Value of layerscale use_postln (bool): Whether use layernorm after modulation. Default: False. use_postln_in_modulation (bool): Whether use post-modulation layernorm. Default: False. normalize_modulator (bool): Whether use normalize in modulator use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. """ def __init__(self, dim, depth, mlp_ratio=4., drop=0., drop_path=0., norm_layer=nn.LayerNorm, downsample=None, focal_level=2, focal_window=9, use_conv_embed=False, use_layerscale=False, layerscale_value=1e-4, use_postln=False, use_postln_in_modulation=False, normalize_modulator=False, use_checkpoint=False): super().__init__() self.depth = depth self.use_checkpoint = use_checkpoint # build blocks self.blocks = nn.LayerList([ FocalModulationBlock( dim=dim, mlp_ratio=mlp_ratio, drop=drop, drop_path=drop_path[i] if isinstance(drop_path, np.ndarray) else drop_path, act_layer=nn.GELU, norm_layer=norm_layer, focal_level=focal_level, focal_window=focal_window, use_postln=use_postln, use_postln_in_modulation=use_postln_in_modulation, normalize_modulator=normalize_modulator, use_layerscale=use_layerscale, layerscale_value=layerscale_value) for i in range(depth) ]) # patch merging layer if downsample is not None: self.downsample = downsample( patch_size=2, in_chans=dim, embed_dim=2 * dim, use_conv_embed=use_conv_embed, norm_layer=norm_layer, is_stem=False) else: self.downsample = None def forward(self, x, H, W): """ Args: x: Input feature, tensor size (B, H*W, C). """ for blk in self.blocks: blk.H, blk.W = H, W x = blk(x) if self.downsample is not None: x_reshaped = x.transpose([0, 2, 1]).reshape( [x.shape[0], x.shape[-1], H, W]) x_down = self.downsample(x_reshaped) x_down = x_down.flatten(2).transpose([0, 2, 1]) Wh, Ww = (H + 1) // 2, (W + 1) // 2 return x, H, W, x_down, Wh, Ww else: return x, H, W, x, H, W class PatchEmbed(nn.Layer): """ Image to Patch Embedding Args: patch_size (int): Patch token size. Default: 4. in_chans (int): Number of input image channels. Default: 3. embed_dim (int): Number of linear projection output channels. Default: 96. norm_layer (nn.Layer, optional): Normalization layer. Default: None use_conv_embed (bool): Whether use overlapped convolution for patch embedding. Default: False is_stem (bool): Is the stem block or not. """ def __init__(self, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None, use_conv_embed=False, is_stem=False): super().__init__() patch_size = to_2tuple(patch_size) self.patch_size = patch_size self.in_chans = in_chans self.embed_dim = embed_dim if use_conv_embed: # if we choose to use conv embedding, then we treat the stem and non-stem differently if is_stem: kernel_size = 7 padding = 2 stride = 4 else: kernel_size = 3 padding = 1 stride = 2 self.proj = nn.Conv2D( in_chans, embed_dim, kernel_size=kernel_size, stride=stride, padding=padding) else: self.proj = nn.Conv2D( in_chans, embed_dim, kernel_size=patch_size, stride=patch_size) if norm_layer is not None: self.norm = norm_layer(embed_dim) else: self.norm = None def forward(self, x): _, _, H, W = x.shape if W % self.patch_size[1] != 0: # for 3D tensor: [pad_left, pad_right] # for 4D tensor: [pad_left, pad_right, pad_top, pad_bottom] x = F.pad(x, [0, self.patch_size[1] - W % self.patch_size[1], 0, 0]) W += W % self.patch_size[1] if H % self.patch_size[0] != 0: x = F.pad(x, [0, 0, 0, self.patch_size[0] - H % self.patch_size[0]]) H += H % self.patch_size[0] x = self.proj(x) if self.norm is not None: _, _, Wh, Ww = x.shape x = x.flatten(2).transpose([0, 2, 1]) x = self.norm(x) x = x.transpose([0, 2, 1]).reshape([-1, self.embed_dim, Wh, Ww]) return x @register @serializable class FocalNet(nn.Layer): """ FocalNet backbone Args: arch (str): Architecture of FocalNet out_indices (Sequence[int]): Output from which stages. frozen_stages (int): Stages to be frozen (stop grad and set eval mode). -1 means not freezing any parameters. patch_size (int | tuple(int)): Patch size. Default: 4. in_chans (int): Number of input image channels. Default: 3. embed_dim (int): Number of linear projection output channels. Default: 96. depths (tuple[int]): Depths of each FocalNet Transformer stage. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4. drop_rate (float): Dropout rate. drop_path_rate (float): Stochastic depth rate. Default: 0.2. norm_layer (nn.Layer): Normalization layer. Default: nn.LayerNorm. patch_norm (bool): If True, add normalization after patch embedding. Default: True. focal_levels (Sequence[int]): Number of focal levels at four stages focal_windows (Sequence[int]): Focal window sizes at first focal level at four stages use_conv_embed (bool): Whether use overlapped convolution for patch embedding use_layerscale (bool): Whether use layerscale proposed in CaiT. Default: False layerscale_value (float): Value of layerscale use_postln (bool): Whether use layernorm after modulation. Default: False. use_postln_in_modulation (bool): Whether use post-modulation layernorm. Default: False. normalize_modulator (bool): Whether use normalize in modulator use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. """ def __init__( self, arch='focalnet_T_224_1k_srf', out_indices=(0, 1, 2, 3), frozen_stages=-1, patch_size=4, in_chans=3, embed_dim=96, depths=[2, 2, 6, 2], mlp_ratio=4., drop_rate=0., drop_path_rate=0.2, # 0.5 better for large+ models norm_layer=nn.LayerNorm, patch_norm=True, focal_levels=[2, 2, 2, 2], focal_windows=[3, 3, 3, 3], use_conv_embed=False, use_layerscale=False, layerscale_value=1e-4, use_postln=False, use_postln_in_modulation=False, normalize_modulator=False, use_checkpoint=False, pretrained=None): super(FocalNet, self).__init__() assert arch in MODEL_cfg.keys(), "Unsupported arch: {}".format(arch) embed_dim = MODEL_cfg[arch]['embed_dim'] depths = MODEL_cfg[arch]['depths'] drop_path_rate = MODEL_cfg[arch]['drop_path_rate'] focal_levels = MODEL_cfg[arch]['focal_levels'] focal_windows = MODEL_cfg[arch]['focal_windows'] use_conv_embed = MODEL_cfg[arch]['use_conv_embed'] use_layerscale = MODEL_cfg[arch]['use_layerscale'] use_postln = MODEL_cfg[arch]['use_postln'] use_postln_in_modulation = MODEL_cfg[arch]['use_postln_in_modulation'] normalize_modulator = MODEL_cfg[arch]['normalize_modulator'] if pretrained is None: pretrained = MODEL_cfg[arch]['pretrained'] self.out_indices = out_indices self.frozen_stages = frozen_stages self.num_layers = len(depths) self.patch_norm = patch_norm # split image into non-overlapping patches self.patch_embed = PatchEmbed( patch_size=patch_size, in_chans=in_chans, embed_dim=embed_dim, norm_layer=norm_layer if self.patch_norm else None, use_conv_embed=use_conv_embed, is_stem=True) self.pos_drop = nn.Dropout(p=drop_rate) # stochastic depth decay rule dpr = np.linspace(0, drop_path_rate, sum(depths)) # build layers self.layers = nn.LayerList() for i_layer in range(self.num_layers): layer = BasicLayer( dim=int(embed_dim * 2**i_layer), depth=depths[i_layer], mlp_ratio=mlp_ratio, drop=drop_rate, drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], norm_layer=norm_layer, downsample=PatchEmbed if (i_layer < self.num_layers - 1) else None, focal_level=focal_levels[i_layer], focal_window=focal_windows[i_layer], use_conv_embed=use_conv_embed, use_layerscale=use_layerscale, layerscale_value=layerscale_value, use_postln=use_postln, use_postln_in_modulation=use_postln_in_modulation, normalize_modulator=normalize_modulator, use_checkpoint=use_checkpoint) self.layers.append(layer) num_features = [int(embed_dim * 2**i) for i in range(self.num_layers)] self.num_features = num_features # add a norm layer for each output for i_layer in out_indices: layer = norm_layer(num_features[i_layer]) layer_name = f'norm{i_layer}' self.add_sublayer(layer_name, layer) self.apply(self._init_weights) self._freeze_stages() if pretrained: if 'http' in pretrained: #URL path = paddle.utils.download.get_weights_path_from_url( pretrained) else: #model in local path path = pretrained self.set_state_dict(paddle.load(path)) def _freeze_stages(self): if self.frozen_stages >= 0: self.patch_embed.eval() for param in self.patch_embed.parameters(): param.stop_gradient = True if self.frozen_stages >= 2: self.pos_drop.eval() for i in range(0, self.frozen_stages - 1): m = self.layers[i] m.eval() for param in m.parameters(): param.stop_gradient = True def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight) if isinstance(m, nn.Linear) and m.bias is not None: zeros_(m.bias) elif isinstance(m, nn.LayerNorm): zeros_(m.bias) ones_(m.weight) def forward(self, x): x = self.patch_embed(x['image']) B, _, Wh, Ww = x.shape x = x.flatten(2).transpose([0, 2, 1]) x = self.pos_drop(x) outs = [] for i in range(self.num_layers): layer = self.layers[i] x_out, H, W, x, Wh, Ww = layer(x, Wh, Ww) if i in self.out_indices: norm_layer = getattr(self, f'norm{i}') x_out = norm_layer(x_out) out = x_out.reshape([-1, H, W, self.num_features[i]]).transpose( (0, 3, 1, 2)) outs.append(out) return outs @property def out_shape(self): out_strides = [4, 8, 16, 32] return [ ShapeSpec( channels=self.num_features[i], stride=out_strides[i]) for i in self.out_indices ]