From 4a4d5f2a2ee06069140b0d861018a70c63ad3d46 Mon Sep 17 00:00:00 2001
From: Gustaf Rydholm <gustaf.rydholm@gmail.com>
Date: Tue, 13 Sep 2022 21:58:26 +0200
Subject: Add convnext attention

---
 text_recognizer/networks/convnext/__init__.py  |  5 ++
 text_recognizer/networks/convnext/attention.py | 79 ++++++++++++++++++++++++++
 text_recognizer/networks/convnext/convnext.py  | 15 +++--
 3 files changed, 91 insertions(+), 8 deletions(-)

(limited to 'text_recognizer')

diff --git a/text_recognizer/networks/convnext/__init__.py b/text_recognizer/networks/convnext/__init__.py
index 8d8470f..1743f09 100644
--- a/text_recognizer/networks/convnext/__init__.py
+++ b/text_recognizer/networks/convnext/__init__.py
@@ -1,2 +1,7 @@
 """Convnext module."""
 from text_recognizer.networks.convnext.convnext import ConvNext
+from text_recognizer.networks.convnext.attention import (
+    Attention,
+    FeedForward,
+    TransformerBlock,
+)
diff --git a/text_recognizer/networks/convnext/attention.py b/text_recognizer/networks/convnext/attention.py
index e69de29..7f03436 100644
--- a/text_recognizer/networks/convnext/attention.py
+++ b/text_recognizer/networks/convnext/attention.py
@@ -0,0 +1,79 @@
+"""Convolution self attention block."""
+
+from einops import reduce, rearrange
+from torch import einsum, nn, Tensor
+import torch.nn.functional as F
+
+from text_recognizer.networks.convnext.norm import LayerNorm
+from text_recognizer.networks.convnext.residual import Residual
+
+
+def l2norm(t: Tensor) -> Tensor:
+    return F.normalize(t, dim=-1)
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim: int, mult: int = 4) -> None:
+        super().__init__()
+        inner_dim = int(dim * mult)
+        self.fn = Residual(
+            nn.Sequential(
+                LayerNorm(dim),
+                nn.Conv2d(dim, inner_dim, 1, bias=False),
+                nn.GELU(),
+                LayerNorm(inner_dim),
+                nn.Conv2d(inner_dim, dim, 1, bias=False),
+            )
+        )
+
+    def forward(self, x: Tensor) -> Tensor:
+        return self.fn(x)
+
+
+class Attention(nn.Module):
+    def __init__(
+        self, dim: int, heads: int = 4, dim_head: int = 64, scale: int = 8
+    ) -> None:
+        super().__init__()
+        self.scale = scale
+        self.heads = heads
+        inner_dim = heads * dim_head
+        self.norm = LayerNorm(dim)
+
+        self.to_qkv = nn.Conv2d(dim, inner_dim * 3, 1, bias=False)
+        self.to_out = nn.Conv2d(inner_dim, dim, 1, bias=False)
+
+    def forward(self, x: Tensor) -> Tensor:
+        h, w = x.shape[-2:]
+
+        residual = x.clone()
+
+        x = self.norm(x)
+
+        q, k, v = self.to_qkv(x).chunk(3, dim=1)
+        q, k, v = map(
+            lambda t: rearrange(t, "b (h c) ... -> b h (...) c", h=self.heads),
+            (q, k, v),
+        )
+
+        q, k = map(l2norm, (q, k))
+
+        sim = einsum("b h i d, b h j d -> b h i j", q, k) * self.scale
+        attn = sim.softmax(dim=-1)
+
+        out = einsum("b h i j, b h j d -> b h i d", attn, v)
+
+        out = rearrange(out, "b h (x y) d -> b (h d) x y", x=h, y=w)
+        return self.to_out(out) + residual
+
+
+class TransformerBlock(nn.Module):
+    def __init__(self, attn: Attention, ff: FeedForward) -> None:
+        super().__init__()
+        self.attn = attn
+        self.ff = ff
+
+    def forward(self, x: Tensor) -> Tensor:
+        x = self.attn(x)
+        x = self.ff(x)
+        return x
diff --git a/text_recognizer/networks/convnext/convnext.py b/text_recognizer/networks/convnext/convnext.py
index a4556a0..68de81a 100644
--- a/text_recognizer/networks/convnext/convnext.py
+++ b/text_recognizer/networks/convnext/convnext.py
@@ -1,13 +1,9 @@
-from typing import Sequence
+from typing import Optional, Sequence
 
-from einops import reduce, rearrange
-from einops.layers.torch import Rearrange
-import torch
-from torch import einsum, nn, Tensor
-import torch.nn.functional as F
+from text_recognizer.networks.convnext.attention import TransformerBlock
+from torch import nn, Tensor
 
 from text_recognizer.networks.convnext.downsample import Downsample
-from text_recognizer.networks.convnext.residual import Residual
 from text_recognizer.networks.convnext.norm import LayerNorm
 
 
@@ -38,9 +34,11 @@ class ConvNext(nn.Module):
         dim_mults: Sequence[int] = (2, 4, 8),
         depths: Sequence[int] = (3, 3, 6),
         downsampling_factors: Sequence[Sequence[int]] = ((2, 2), (2, 2), (2, 2)),
+        attn: Optional[TransformerBlock] = None,
     ) -> None:
         super().__init__()
         dims = (dim, *map(lambda m: m * dim, dim_mults))
+        self.attn = attn
         self.out_channels = dims[-1]
         self.stem = nn.Conv2d(1, dims[0], kernel_size=(7, 7), padding="same")
         self.layers = nn.ModuleList([])
@@ -65,11 +63,12 @@ class ConvNext(nn.Module):
             nn.init.trunc_normal_(m.weight, std=0.02)
             nn.init.constant_(m.bias, 0)
 
-    def forward(self, x):
+    def forward(self, x: Tensor) -> Tensor:
         x = self.stem(x)
         for init_block, blocks, down in self.layers:
             x = init_block(x)
             for fn in blocks:
                 x = fn(x)
             x = down(x)
+        x = self.attn(x)
         return self.norm(x)
-- 
cgit v1.2.3-70-g09d2