File size: 7,265 Bytes
2366e36
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
# Copyright (c) OpenMMLab. All rights reserved.
import torch
from mmcv.runner import BaseModule
from torch import nn
from torch.nn import functional as F

from mmocr.models.builder import HEADS, build_loss


@HEADS.register_module()
class SDMGRHead(BaseModule):

    def __init__(self,
                 num_chars=92,
                 visual_dim=64,
                 fusion_dim=1024,
                 node_input=32,
                 node_embed=256,
                 edge_input=5,
                 edge_embed=256,
                 num_gnn=2,
                 num_classes=26,
                 loss=dict(type='SDMGRLoss'),
                 bidirectional=False,
                 train_cfg=None,
                 test_cfg=None,
                 init_cfg=dict(
                     type='Normal',
                     override=dict(name='edge_embed'),
                     mean=0,
                     std=0.01)):
        super().__init__(init_cfg=init_cfg)

        self.fusion = Block([visual_dim, node_embed], node_embed, fusion_dim)
        self.node_embed = nn.Embedding(num_chars, node_input, 0)
        hidden = node_embed // 2 if bidirectional else node_embed
        self.rnn = nn.LSTM(
            input_size=node_input,
            hidden_size=hidden,
            num_layers=1,
            batch_first=True,
            bidirectional=bidirectional)
        self.edge_embed = nn.Linear(edge_input, edge_embed)
        self.gnn_layers = nn.ModuleList(
            [GNNLayer(node_embed, edge_embed) for _ in range(num_gnn)])
        self.node_cls = nn.Linear(node_embed, num_classes)
        self.edge_cls = nn.Linear(edge_embed, 2)
        self.loss = build_loss(loss)

    def forward(self, relations, texts, x=None):
        node_nums, char_nums = [], []
        for text in texts:
            node_nums.append(text.size(0))
            char_nums.append((text > 0).sum(-1))

        max_num = max([char_num.max() for char_num in char_nums])
        all_nodes = torch.cat([
            torch.cat(
                [text,
                 text.new_zeros(text.size(0), max_num - text.size(1))], -1)
            for text in texts
        ])
        embed_nodes = self.node_embed(all_nodes.clamp(min=0).long())
        rnn_nodes, _ = self.rnn(embed_nodes)

        nodes = rnn_nodes.new_zeros(*rnn_nodes.shape[::2])
        all_nums = torch.cat(char_nums)
        valid = all_nums > 0
        nodes[valid] = rnn_nodes[valid].gather(
            1, (all_nums[valid] - 1).unsqueeze(-1).unsqueeze(-1).expand(
                -1, -1, rnn_nodes.size(-1))).squeeze(1)

        if x is not None:
            nodes = self.fusion([x, nodes])

        all_edges = torch.cat(
            [rel.view(-1, rel.size(-1)) for rel in relations])
        embed_edges = self.edge_embed(all_edges.float())
        embed_edges = F.normalize(embed_edges)

        for gnn_layer in self.gnn_layers:
            nodes, cat_nodes = gnn_layer(nodes, embed_edges, node_nums)

        node_cls, edge_cls = self.node_cls(nodes), self.edge_cls(cat_nodes)
        return node_cls, edge_cls


class GNNLayer(nn.Module):

    def __init__(self, node_dim=256, edge_dim=256):
        super().__init__()
        self.in_fc = nn.Linear(node_dim * 2 + edge_dim, node_dim)
        self.coef_fc = nn.Linear(node_dim, 1)
        self.out_fc = nn.Linear(node_dim, node_dim)
        self.relu = nn.ReLU()

    def forward(self, nodes, edges, nums):
        start, cat_nodes = 0, []
        for num in nums:
            sample_nodes = nodes[start:start + num]
            cat_nodes.append(
                torch.cat([
                    sample_nodes.unsqueeze(1).expand(-1, num, -1),
                    sample_nodes.unsqueeze(0).expand(num, -1, -1)
                ], -1).view(num**2, -1))
            start += num
        cat_nodes = torch.cat([torch.cat(cat_nodes), edges], -1)
        cat_nodes = self.relu(self.in_fc(cat_nodes))
        coefs = self.coef_fc(cat_nodes)

        start, residuals = 0, []
        for num in nums:
            residual = F.softmax(
                -torch.eye(num).to(coefs.device).unsqueeze(-1) * 1e9 +
                coefs[start:start + num**2].view(num, num, -1), 1)
            residuals.append(
                (residual *
                 cat_nodes[start:start + num**2].view(num, num, -1)).sum(1))
            start += num**2

        nodes += self.relu(self.out_fc(torch.cat(residuals)))
        return nodes, cat_nodes


class Block(nn.Module):

    def __init__(self,
                 input_dims,
                 output_dim,
                 mm_dim=1600,
                 chunks=20,
                 rank=15,
                 shared=False,
                 dropout_input=0.,
                 dropout_pre_lin=0.,
                 dropout_output=0.,
                 pos_norm='before_cat'):
        super().__init__()
        self.rank = rank
        self.dropout_input = dropout_input
        self.dropout_pre_lin = dropout_pre_lin
        self.dropout_output = dropout_output
        assert (pos_norm in ['before_cat', 'after_cat'])
        self.pos_norm = pos_norm
        # Modules
        self.linear0 = nn.Linear(input_dims[0], mm_dim)
        self.linear1 = (
            self.linear0 if shared else nn.Linear(input_dims[1], mm_dim))
        self.merge_linears0 = nn.ModuleList()
        self.merge_linears1 = nn.ModuleList()
        self.chunks = self.chunk_sizes(mm_dim, chunks)
        for size in self.chunks:
            ml0 = nn.Linear(size, size * rank)
            self.merge_linears0.append(ml0)
            ml1 = ml0 if shared else nn.Linear(size, size * rank)
            self.merge_linears1.append(ml1)
        self.linear_out = nn.Linear(mm_dim, output_dim)

    def forward(self, x):
        x0 = self.linear0(x[0])
        x1 = self.linear1(x[1])
        bs = x1.size(0)
        if self.dropout_input > 0:
            x0 = F.dropout(x0, p=self.dropout_input, training=self.training)
            x1 = F.dropout(x1, p=self.dropout_input, training=self.training)
        x0_chunks = torch.split(x0, self.chunks, -1)
        x1_chunks = torch.split(x1, self.chunks, -1)
        zs = []
        for x0_c, x1_c, m0, m1 in zip(x0_chunks, x1_chunks,
                                      self.merge_linears0,
                                      self.merge_linears1):
            m = m0(x0_c) * m1(x1_c)  # bs x split_size*rank
            m = m.view(bs, self.rank, -1)
            z = torch.sum(m, 1)
            if self.pos_norm == 'before_cat':
                z = torch.sqrt(F.relu(z)) - torch.sqrt(F.relu(-z))
                z = F.normalize(z)
            zs.append(z)
        z = torch.cat(zs, 1)
        if self.pos_norm == 'after_cat':
            z = torch.sqrt(F.relu(z)) - torch.sqrt(F.relu(-z))
            z = F.normalize(z)

        if self.dropout_pre_lin > 0:
            z = F.dropout(z, p=self.dropout_pre_lin, training=self.training)
        z = self.linear_out(z)
        if self.dropout_output > 0:
            z = F.dropout(z, p=self.dropout_output, training=self.training)
        return z

    @staticmethod
    def chunk_sizes(dim, chunks):
        split_size = (dim + chunks - 1) // chunks
        sizes_list = [split_size] * chunks
        sizes_list[-1] = sizes_list[-1] - (sum(sizes_list) - dim)
        return sizes_list