Merge pull request #825 from TingquanGao/reg/add_twins_levit

Add LeViT and Twins

docs/zh_CN/models/LeViT.md

+# LeViT
+
+## 概述
+LeViT是一种快速推理的、用于图像分类任务的混合神经网络。其设计之初考虑了网络模型在不同的硬件平台上的性能，因此能够更好地反映普遍应用的真实场景。通过大量实验，作者找到了卷积神经网络与Transformer体系更好的结合方式，并且提出了attention-based方法，用于整合Transformer中的位置信息编码。[论文地址](https://arxiv.org/abs/2104.01136)。
+
+## 精度、FLOPS和参数量
+
+| Models           | Top1 | Top5 | Reference<br>top1 | Reference<br>top5 | FLOPS<br>(M) | Params<br>(M) |
+|:--:|:--:|:--:|:--:|:--:|:--:|:--:|
+| LeViT-128S | 0.7621 | 0.9277 | 0.766 | 0.929 | 305  | 7.8 |
+| LeViT-128  | 0.7833 | 0.9378 | 0.786 | 0.940 | 406  | 9.2 |
+| LeViT-192  | 0.7963 | 0.9460 | 0.800 | 0.947 | 658  | 11 |
+| LeViT-256  | 0.8085 | 0.9497 | 0.816 | 0.954 | 1120 | 19 |
+| LeViT-384  | 0.8234 | 0.9587 | 0.826 | 0.960 | 2353 | 39 |
+
+
+**注**：与Reference的精度差异源于数据预处理不同。

docs/zh_CN/models/Twins.md

+# Twins
+
+## 概述
+Twins网络包括Twins-PCPVT和Twins-SVT，其重点对空间注意力机制进行了精心设计，得到了简单却更为有效的方案。由于该体系结构仅涉及矩阵乘法，而目前的深度学习框架中对矩阵乘法有较高的优化程度，因此该体系结构十分高效且易于实现。并且，该体系结构在图像分类、目标检测和语义分割等多种下游视觉任务中都能够取得优异的性能。[论文地址](https://arxiv.org/abs/2104.13840)。
+
+## 精度、FLOPS和参数量
+
+| Models        | Top1 | Top5 | Reference<br>top1 | Reference<br>top5 | FLOPS<br>(G) | Params<br>(M) |
+|:--:|:--:|:--:|:--:|:--:|:--:|:--:|
+| pcpvt_small   | 0.8082 | 0.9552 | 0.812 | - | 3.7 | 24.1   |
+| pcpvt_base    | 0.8242 | 0.9619 | 0.827 | - | 6.4 | 43.8   |
+| pcpvt_large   | 0.8273 | 0.9650 | 0.831 | - | 9.5 | 60.9   |
+| alt_gvt_small | 0.8140 | 0.9546 | 0.817 | - | 2.8  | 24   |
+| alt_gvt_base  | 0.8294 | 0.9621 | 0.832 | - | 8.3  | 56   |
+| alt_gvt_large | 0.8331 | 0.9642 | 0.837 | - | 14.8 | 99.2   |
+
+**注**：与Reference的精度差异源于数据预处理不同。

ppcls/arch/backbone/__init__.py

-# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
+# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -47,6 +47,8 @@ from ppcls.arch.backbone.model_zoo.distillation_models import ResNet50_vd_distil
 from ppcls.arch.backbone.model_zoo.swin_transformer import SwinTransformer_tiny_patch4_window7_224, SwinTransformer_small_patch4_window7_224, SwinTransformer_base_patch4_window7_224, SwinTransformer_base_patch4_window12_384, SwinTransformer_large_patch4_window7_224, SwinTransformer_large_patch4_window12_384
 from ppcls.arch.backbone.model_zoo.mixnet import MixNet_S, MixNet_M, MixNet_L
 from ppcls.arch.backbone.model_zoo.rexnet import ReXNet_1_0, ReXNet_1_3, ReXNet_1_5, ReXNet_2_0, ReXNet_3_0
+from ppcls.arch.backbone.model_zoo.gvt import pcpvt_small, pcpvt_base, pcpvt_large, alt_gvt_small, alt_gvt_base, alt_gvt_large
+from ppcls.arch.backbone.model_zoo.levit import LeViT_128S, LeViT_128, LeViT_192, LeViT_256, LeViT_384
 from ppcls.arch.backbone.model_zoo.dla import DLA34, DLA46_c, DLA46x_c, DLA60, DLA60x, DLA60x_c, DLA102, DLA102x, DLA102x2, DLA169
 from ppcls.arch.backbone.model_zoo.rednet import RedNet26, RedNet38, RedNet50, RedNet101, RedNet152 
 from ppcls.arch.backbone.model_zoo.tnt import TNT_small

ppcls/arch/backbone/model_zoo/levit.py

+# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
+#
+# 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.
+
+import itertools
+import math
+import warnings
+
+import paddle
+import paddle.nn as nn
+import paddle.nn.functional as F
+from paddle.nn.initializer import TruncatedNormal, Constant
+from paddle.regularizer import L2Decay
+
+from .vision_transformer import trunc_normal_, zeros_, ones_, Identity
+
+__all__ = ["LeViT_128S", "LeViT_128", "LeViT_192", "LeViT_256", "LeViT_384"]
+
+
+def cal_attention_biases(attention_biases, attention_bias_idxs):
+    gather_list = []
+    attention_bias_t = paddle.transpose(attention_biases, (1, 0))
+    for idx in attention_bias_idxs:
+        gather = paddle.gather(attention_bias_t, idx)
+        gather_list.append(gather)
+    shape0, shape1 = attention_bias_idxs.shape
+    return paddle.transpose(paddle.concat(gather_list), (1, 0)).reshape(
+        (0, shape0, shape1))
+
+
+class Conv2d_BN(nn.Sequential):
+    def __init__(self,
+                 a,
+                 b,
+                 ks=1,
+                 stride=1,
+                 pad=0,
+                 dilation=1,
+                 groups=1,
+                 bn_weight_init=1,
+                 resolution=-10000):
+        super().__init__()
+        self.add_sublayer(
+            'c',
+            nn.Conv2D(
+                a, b, ks, stride, pad, dilation, groups, bias_attr=False))
+        bn = nn.BatchNorm2D(b)
+        ones_(bn.weight)
+        zeros_(bn.bias)
+        self.add_sublayer('bn', bn)
+
+
+class Linear_BN(nn.Sequential):
+    def __init__(self, a, b, bn_weight_init=1):
+        super().__init__()
+        self.add_sublayer('c', nn.Linear(a, b, bias_attr=False))
+        bn = nn.BatchNorm1D(b)
+        ones_(bn.weight)
+        zeros_(bn.bias)
+        self.add_sublayer('bn', bn)
+
+    def forward(self, x):
+        l, bn = self._sub_layers.values()
+        x = l(x)
+        return paddle.reshape(bn(x.flatten(0, 1)), x.shape)
+
+
+class BN_Linear(nn.Sequential):
+    def __init__(self, a, b, bias=True, std=0.02):
+        super().__init__()
+        self.add_sublayer('bn', nn.BatchNorm1D(a))
+        l = nn.Linear(a, b, bias_attr=bias)
+        trunc_normal_(l.weight)
+        if bias:
+            zeros_(l.bias)
+        self.add_sublayer('l', l)
+
+
+def b16(n, activation, resolution=224):
+    return nn.Sequential(
+        Conv2d_BN(
+            3, n // 8, 3, 2, 1, resolution=resolution),
+        activation(),
+        Conv2d_BN(
+            n // 8, n // 4, 3, 2, 1, resolution=resolution // 2),
+        activation(),
+        Conv2d_BN(
+            n // 4, n // 2, 3, 2, 1, resolution=resolution // 4),
+        activation(),
+        Conv2d_BN(
+            n // 2, n, 3, 2, 1, resolution=resolution // 8))
+
+
+class Residual(nn.Layer):
+    def __init__(self, m, drop):
+        super().__init__()
+        self.m = m
+        self.drop = drop
+
+    def forward(self, x):
+        if self.training and self.drop > 0:
+            return x + self.m(x) * paddle.rand(
+                x.size(0), 1, 1,
+                device=x.device).ge_(self.drop).div(1 - self.drop).detach()
+        else:
+            return x + self.m(x)
+
+
+class Attention(nn.Layer):
+    def __init__(self,
+                 dim,
+                 key_dim,
+                 num_heads=8,
+                 attn_ratio=4,
+                 activation=None,
+                 resolution=14):
+        super().__init__()
+        self.num_heads = num_heads
+        self.scale = key_dim**-0.5
+        self.key_dim = key_dim
+        self.nh_kd = nh_kd = key_dim * num_heads
+        self.d = int(attn_ratio * key_dim)
+        self.dh = int(attn_ratio * key_dim) * num_heads
+        self.attn_ratio = attn_ratio
+        self.h = self.dh + nh_kd * 2
+        self.qkv = Linear_BN(dim, self.h)
+        self.proj = nn.Sequential(
+            activation(), Linear_BN(
+                self.dh, dim, bn_weight_init=0))
+        points = list(itertools.product(range(resolution), range(resolution)))
+        N = len(points)
+        attention_offsets = {}
+        idxs = []
+        for p1 in points:
+            for p2 in points:
+                offset = (abs(p1[0] - p2[0]), abs(p1[1] - p2[1]))
+                if offset not in attention_offsets:
+                    attention_offsets[offset] = len(attention_offsets)
+                idxs.append(attention_offsets[offset])
+        self.attention_biases = self.create_parameter(
+            shape=(num_heads, len(attention_offsets)),
+            default_initializer=zeros_,
+            attr=paddle.ParamAttr(regularizer=L2Decay(0.0)))
+        tensor_idxs = paddle.to_tensor(idxs, dtype='int64')
+        self.register_buffer('attention_bias_idxs',
+                             paddle.reshape(tensor_idxs, [N, N]))
+
+    @paddle.no_grad()
+    def train(self, mode=True):
+        if mode:
+            super().train()
+        else:
+            super().eval()
+        if mode and hasattr(self, 'ab'):
+            del self.ab
+        else:
+            self.ab = cal_attention_biases(self.attention_biases,
+                                           self.attention_bias_idxs)
+
+    def forward(self, x):
+        self.training = True
+        B, N, C = x.shape
+        qkv = self.qkv(x)
+        qkv = paddle.reshape(qkv,
+                             [B, N, self.num_heads, self.h // self.num_heads])
+        q, k, v = paddle.split(
+            qkv, [self.key_dim, self.key_dim, self.d], axis=3)
+        q = paddle.transpose(q, perm=[0, 2, 1, 3])
+        k = paddle.transpose(k, perm=[0, 2, 1, 3])
+        v = paddle.transpose(v, perm=[0, 2, 1, 3])
+        k_transpose = paddle.transpose(k, perm=[0, 1, 3, 2])
+
+        if self.training:
+            attention_biases = cal_attention_biases(self.attention_biases,
+                                                    self.attention_bias_idxs)
+        else:
+            attention_biases = self.ab
+        attn = ((q @k_transpose) * self.scale + attention_biases)
+        attn = F.softmax(attn)
+        x = paddle.transpose(attn @v, perm=[0, 2, 1, 3])
+        x = paddle.reshape(x, [B, N, self.dh])
+        x = self.proj(x)
+        return x
+
+
+class Subsample(nn.Layer):
+    def __init__(self, stride, resolution):
+        super().__init__()
+        self.stride = stride
+        self.resolution = resolution
+
+    def forward(self, x):
+        B, N, C = x.shape
+        x = paddle.reshape(x, [B, self.resolution, self.resolution,
+                               C])[:, ::self.stride, ::self.stride]
+        x = paddle.reshape(x, [B, -1, C])
+        return x
+
+
+class AttentionSubsample(nn.Layer):
+    def __init__(self,
+                 in_dim,
+                 out_dim,
+                 key_dim,
+                 num_heads=8,
+                 attn_ratio=2,
+                 activation=None,
+                 stride=2,
+                 resolution=14,
+                 resolution_=7):
+        super().__init__()
+        self.num_heads = num_heads
+        self.scale = key_dim**-0.5
+        self.key_dim = key_dim
+        self.nh_kd = nh_kd = key_dim * num_heads
+        self.d = int(attn_ratio * key_dim)
+        self.dh = int(attn_ratio * key_dim) * self.num_heads
+        self.attn_ratio = attn_ratio
+        self.resolution_ = resolution_
+        self.resolution_2 = resolution_**2
+        self.training = True
+        h = self.dh + nh_kd
+        self.kv = Linear_BN(in_dim, h)
+
+        self.q = nn.Sequential(
+            Subsample(stride, resolution), Linear_BN(in_dim, nh_kd))
+        self.proj = nn.Sequential(activation(), Linear_BN(self.dh, out_dim))
+
+        self.stride = stride
+        self.resolution = resolution
+        points = list(itertools.product(range(resolution), range(resolution)))
+        points_ = list(
+            itertools.product(range(resolution_), range(resolution_)))
+
+        N = len(points)
+        N_ = len(points_)
+        attention_offsets = {}
+        idxs = []
+        i = 0
+        j = 0
+        for p1 in points_:
+            i += 1
+            for p2 in points:
+                j += 1
+                size = 1
+                offset = (abs(p1[0] * stride - p2[0] + (size - 1) / 2),
+                          abs(p1[1] * stride - p2[1] + (size - 1) / 2))
+                if offset not in attention_offsets:
+                    attention_offsets[offset] = len(attention_offsets)
+                idxs.append(attention_offsets[offset])
+        self.attention_biases = self.create_parameter(
+            shape=(num_heads, len(attention_offsets)),
+            default_initializer=zeros_,
+            attr=paddle.ParamAttr(regularizer=L2Decay(0.0)))
+
+        tensor_idxs_ = paddle.to_tensor(idxs, dtype='int64')
+        self.register_buffer('attention_bias_idxs',
+                             paddle.reshape(tensor_idxs_, [N_, N]))
+
+    @paddle.no_grad()
+    def train(self, mode=True):
+        if mode:
+            super().train()
+        else:
+            super().eval()
+        if mode and hasattr(self, 'ab'):
+            del self.ab
+        else:
+            self.ab = cal_attention_biases(self.attention_biases,
+                                           self.attention_bias_idxs)
+
+    def forward(self, x):
+        self.training = True
+        B, N, C = x.shape
+        kv = self.kv(x)
+        kv = paddle.reshape(kv, [B, N, self.num_heads, -1])
+        k, v = paddle.split(kv, [self.key_dim, self.d], axis=3)
+        k = paddle.transpose(k, perm=[0, 2, 1, 3])  # BHNC
+        v = paddle.transpose(v, perm=[0, 2, 1, 3])
+        q = paddle.reshape(
+            self.q(x), [B, self.resolution_2, self.num_heads, self.key_dim])
+        q = paddle.transpose(q, perm=[0, 2, 1, 3])
+
+        if self.training:
+            attention_biases = cal_attention_biases(self.attention_biases,
+                                                    self.attention_bias_idxs)
+        else:
+            attention_biases = self.ab
+
+        attn = (q @paddle.transpose(
+            k, perm=[0, 1, 3, 2])) * self.scale + attention_biases
+        attn = F.softmax(attn)
+
+        x = paddle.reshape(
+            paddle.transpose(
+                (attn @v), perm=[0, 2, 1, 3]), [B, -1, self.dh])
+        x = self.proj(x)
+        return x
+
+
+class LeViT(nn.Layer):
+    """ Vision Transformer with support for patch or hybrid CNN input stage
+    """
+
+    def __init__(self,
+                 img_size=224,
+                 patch_size=16,
+                 in_chans=3,
+                 class_dim=1000,
+                 embed_dim=[192],
+                 key_dim=[64],
+                 depth=[12],
+                 num_heads=[3],
+                 attn_ratio=[2],
+                 mlp_ratio=[2],
+                 hybrid_backbone=None,
+                 down_ops=[],
+                 attention_activation=nn.Hardswish,
+                 mlp_activation=nn.Hardswish,
+                 distillation=True,
+                 drop_path=0):
+        super().__init__()
+
+        self.class_dim = class_dim
+        self.num_features = embed_dim[-1]
+        self.embed_dim = embed_dim
+        self.distillation = distillation
+
+        self.patch_embed = hybrid_backbone
+
+        self.blocks = []
+        down_ops.append([''])
+        resolution = img_size // patch_size
+        for i, (ed, kd, dpth, nh, ar, mr, do) in enumerate(
+                zip(embed_dim, key_dim, depth, num_heads, attn_ratio,
+                    mlp_ratio, down_ops)):
+            for _ in range(dpth):
+                self.blocks.append(
+                    Residual(
+                        Attention(
+                            ed,
+                            kd,
+                            nh,
+                            attn_ratio=ar,
+                            activation=attention_activation,
+                            resolution=resolution, ),
+                        drop_path))
+                if mr > 0:
+                    h = int(ed * mr)
+                    self.blocks.append(
+                        Residual(
+                            nn.Sequential(
+                                Linear_BN(ed, h),
+                                mlp_activation(),
+                                Linear_BN(
+                                    h, ed, bn_weight_init=0), ),
+                            drop_path))
+            if do[0] == 'Subsample':
+                #('Subsample',key_dim, num_heads, attn_ratio, mlp_ratio, stride)
+                resolution_ = (resolution - 1) // do[5] + 1
+                self.blocks.append(
+                    AttentionSubsample(
+                        *embed_dim[i:i + 2],
+                        key_dim=do[1],
+                        num_heads=do[2],
+                        attn_ratio=do[3],
+                        activation=attention_activation,
+                        stride=do[5],
+                        resolution=resolution,
+                        resolution_=resolution_))
+                resolution = resolution_
+                if do[4] > 0:  # mlp_ratio
+                    h = int(embed_dim[i + 1] * do[4])
+                    self.blocks.append(
+                        Residual(
+                            nn.Sequential(
+                                Linear_BN(embed_dim[i + 1], h),
+                                mlp_activation(),
+                                Linear_BN(
+                                    h, embed_dim[i + 1], bn_weight_init=0), ),
+                            drop_path))
+        self.blocks = nn.Sequential(*self.blocks)
+
+        # Classifier head
+        self.head = BN_Linear(embed_dim[-1],
+                              class_dim) if class_dim > 0 else Identity()
+        if distillation:
+            self.head_dist = BN_Linear(
+                embed_dim[-1], class_dim) if class_dim > 0 else Identity()
+
+    def forward(self, x):
+        x = self.patch_embed(x)
+        x = x.flatten(2)
+        x = paddle.transpose(x, perm=[0, 2, 1])
+        x = self.blocks(x)
+        x = x.mean(1)
+        if self.distillation:
+            x = self.head(x), self.head_dist(x)
+            if not self.training:
+                x = (x[0] + x[1]) / 2
+        else:
+            x = self.head(x)
+        return x
+
+
+def model_factory(C, D, X, N, drop_path, class_dim, distillation):
+    embed_dim = [int(x) for x in C.split('_')]
+    num_heads = [int(x) for x in N.split('_')]
+    depth = [int(x) for x in X.split('_')]
+    act = nn.Hardswish
+    model = LeViT(
+        patch_size=16,
+        embed_dim=embed_dim,
+        num_heads=num_heads,
+        key_dim=[D] * 3,
+        depth=depth,
+        attn_ratio=[2, 2, 2],
+        mlp_ratio=[2, 2, 2],
+        down_ops=[
+            #('Subsample',key_dim, num_heads, attn_ratio, mlp_ratio, stride)
+            ['Subsample', D, embed_dim[0] // D, 4, 2, 2],
+            ['Subsample', D, embed_dim[1] // D, 4, 2, 2],
+        ],
+        attention_activation=act,
+        mlp_activation=act,
+        hybrid_backbone=b16(embed_dim[0], activation=act),
+        class_dim=class_dim,
+        drop_path=drop_path,
+        distillation=distillation)
+
+    return model
+
+
+specification = {
+    'LeViT_128S': {
+        'C': '128_256_384',
+        'D': 16,
+        'N': '4_6_8',
+        'X': '2_3_4',
+        'drop_path': 0
+    },
+    'LeViT_128': {
+        'C': '128_256_384',
+        'D': 16,
+        'N': '4_8_12',
+        'X': '4_4_4',
+        'drop_path': 0
+    },
+    'LeViT_192': {
+        'C': '192_288_384',
+        'D': 32,
+        'N': '3_5_6',
+        'X': '4_4_4',
+        'drop_path': 0
+    },
+    'LeViT_256': {
+        'C': '256_384_512',
+        'D': 32,
+        'N': '4_6_8',
+        'X': '4_4_4',
+        'drop_path': 0
+    },
+    'LeViT_384': {
+        'C': '384_512_768',
+        'D': 32,
+        'N': '6_9_12',
+        'X': '4_4_4',
+        'drop_path': 0.1
+    },
+}
+
+
+def LeViT_128S(class_dim=1000, distillation=True, pretrained=False):
+    return model_factory(
+        **specification['LeViT_128S'],
+        class_dim=class_dim,
+        distillation=distillation)
+
+
+def LeViT_128(class_dim=1000, distillation=True):
+    return model_factory(
+        **specification['LeViT_128'],
+        class_dim=class_dim,
+        distillation=distillation)
+
+
+def LeViT_192(class_dim=1000, distillation=True):
+    return model_factory(
+        **specification['LeViT_192'],
+        class_dim=class_dim,
+        distillation=distillation)
+
+
+def LeViT_256(class_dim=1000, distillation=False):
+    return model_factory(
+        **specification['LeViT_256'],
+        class_dim=class_dim,
+        distillation=distillation)
+
+
+def LeViT_384(class_dim=1000, distillation=True):
+    return model_factory(
+        **specification['LeViT_384'],
+        class_dim=class_dim,
+        distillation=distillation)

ppcls/arch/backbone/model_zoo/vision_transformer.py

@@ -12,10 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from collections import Callable
+
 import numpy as np
 import paddle
 import paddle.nn as nn
-from paddle.nn.initializer import TruncatedNormal, Constant
+from paddle.nn.initializer import TruncatedNormal, Constant, Normal
 
 __all__ = [
     "VisionTransformer", "ViT_small_patch16_224", "ViT_base_patch16_224",
@@ -25,6 +27,7 @@ __all__ = [
 ]
 
 trunc_normal_ = TruncatedNormal(std=.02)
+normal_ = Normal
 zeros_ = Constant(value=0.)
 ones_ = Constant(value=1.)
 
@@ -141,7 +144,13 @@ class Block(nn.Layer):
                  norm_layer='nn.LayerNorm',
                  epsilon=1e-5):
         super().__init__()
-        self.norm1 = eval(norm_layer)(dim, epsilon=epsilon)
+        if isinstance(norm_layer, str):
+            self.norm1 = eval(norm_layer)(dim, epsilon=epsilon)
+        elif isinstance(norm_layer, Callable):
+            self.norm1 = norm_layer(dim)
+        else:
+            raise TypeError(
+                "The norm_layer must be str or paddle.nn.layer.Layer class")
         self.attn = Attention(
             dim,
             num_heads=num_heads,
@@ -151,7 +160,13 @@ class Block(nn.Layer):
             proj_drop=drop)
         # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
         self.drop_path = DropPath(drop_path) if drop_path > 0. else Identity()
-        self.norm2 = eval(norm_layer)(dim, epsilon=epsilon)
+        if isinstance(norm_layer, str):
+            self.norm2 = eval(norm_layer)(dim, epsilon=epsilon)
+        elif isinstance(norm_layer, Callable):
+            self.norm2 = norm_layer(dim)
+        else:
+            raise TypeError(
+                "The norm_layer must be str or paddle.nn.layer.Layer class")
         mlp_hidden_dim = int(dim * mlp_ratio)
         self.mlp = Mlp(in_features=dim,
                        hidden_features=mlp_hidden_dim,