-
Notifications
You must be signed in to change notification settings - Fork 19
/
Copy pathmodel.py
233 lines (208 loc) · 9.34 KB
/
model.py
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
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
import math
import torch
import pickle
import numpy as np
import torch.nn as nn
import torch.nn.functional as F
from channel import *
from model_util import *
from functools import partial
from model_util import _cfg
from timm.models.registry import register_model
from timm.models.layers import trunc_normal_ as __call_trunc_normal_
from typing import List, Callable, Union, Any, TypeVar, Tuple
from transformers import BertModel
from base_args import IMGC_NUMCLASS
def trunc_normal_(tensor, mean=0., std=1.):
__call_trunc_normal_(tensor, mean=mean, std=std, a=-std, b=std)
__all__ = [
'ViT_Van_model',
'ViT_FIM_model']
class ViT_Van_CLS(nn.Module):
def __init__(self,
img_size=224, patch_size=16, encoder_in_chans=3, encoder_num_classes=0,
encoder_embed_dim=768, encoder_depth=12,encoder_num_heads=12, decoder_num_classes=768,
decoder_embed_dim=512, decoder_depth=8, decoder_num_heads=8, mlp_ratio=4.,
qkv_bias=False, qk_scale=None, drop_rate=0., attn_drop_rate=0., drop_path_rate=0.,
norm_layer=nn.LayerNorm, init_values=0.,use_learnable_pos_emb=False,num_classes=0,
):
super().__init__()
self.img_encoder = ViTEncoder_Van(img_size=img_size, patch_size=patch_size, in_chans=encoder_in_chans,
num_classes=encoder_num_classes, embed_dim=encoder_embed_dim,depth=encoder_depth,
num_heads=encoder_num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias,drop_rate=drop_rate,
drop_path_rate=drop_path_rate,norm_layer=norm_layer, init_values=init_values,
use_learnable_pos_emb=use_learnable_pos_emb)
self.img_decoder = ViTDecoder_Van(patch_size=patch_size, num_patches=self.img_encoder.patch_embed.num_patches,
num_classes=decoder_num_classes, embed_dim=decoder_embed_dim, depth=decoder_depth,
num_heads=decoder_num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale,
drop_rate=drop_rate, attn_drop_rate=attn_drop_rate, drop_path_rate=drop_path_rate,
norm_layer=norm_layer,init_values=init_values) if decoder_depth>0 else nn.Identity()
self.encoder_to_channel = nn.Linear(encoder_embed_dim, 32)
self.channel = Channels()
self.channel_to_decoder = nn.Linear(32, decoder_embed_dim)
# self.head = nn.Linear(decoder_embed_dim, IMGC_NUMCLASS)
self.head = nn.Linear(decoder_embed_dim, IMGC_NUMCLASS)
def _init_weights(self, m):
if isinstance(m, nn.Linear):
nn.init.xavier_uniform_(m.weight)
if isinstance(m, nn.Linear) and m.bias is not None:
nn.init.constant_(m.bias, 0)
elif isinstance(m, nn.LayerNorm):
nn.init.constant_(m.bias, 0)
nn.init.constant_(m.weight, 1.0)
def get_num_layers(self):
return len(self.blocks)
@torch.jit.ignore
def no_weight_decay(self):
return {'pos_embed', 'cls_token', 'mask_token'}
def forward(self, img, bm_pos, target=None, _eval=False, test_snr=200):
if _eval:
self.eval()
else:
self.train()
out = {}
if self.training:
noise_snr, noise_var = noise_gen(self.training)
noise_var,noise_snr = noise_var.cuda(), noise_snr.cpu().item()
else:
noise_var = torch.FloatTensor([1]) * 10**(-test_snr/20)
x = self.img_encoder(img, bm_pos)
x = self.encoder_to_channel(x)
# x = power_norm_batchwise(x)
# x = self.channel.AWGN(x, noise_var.item())
x = self.channel_to_decoder(x)
x = self.img_decoder(x)
# x = self.head(x.view(x.shape[0],-1))
x = self.head(x.mean(1))
out['out_x'] = x
return out
class ViT_FIM_CLS(nn.Module):
def __init__(self,
img_size=224, patch_size=16, encoder_in_chans=3, encoder_num_classes=0,
encoder_embed_dim=768, encoder_depth=12,encoder_num_heads=12, decoder_num_classes=768,
decoder_embed_dim=512, decoder_depth=8, decoder_num_heads=8, mlp_ratio=4.,
qkv_bias=False, qk_scale=None, drop_rate=0., attn_drop_rate=0., drop_path_rate=0.,
norm_layer=nn.LayerNorm, init_values=0.,use_learnable_pos_emb=False,num_classes=0,
):
super().__init__()
self.img_encoder = ViTEncoder_FIM(img_size=img_size, patch_size=patch_size, in_chans=encoder_in_chans,
num_classes=encoder_num_classes, embed_dim=encoder_embed_dim,depth=encoder_depth,
num_heads=encoder_num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias,drop_rate=drop_rate,
drop_path_rate=drop_path_rate,norm_layer=norm_layer, init_values=init_values,
use_learnable_pos_emb=use_learnable_pos_emb)
self.img_decoder = ViTDecoder_Van(patch_size=patch_size, num_patches=self.img_encoder.patch_embed.num_patches,
num_classes=decoder_num_classes, embed_dim=decoder_embed_dim, depth=decoder_depth,
num_heads=decoder_num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale,
drop_rate=drop_rate, attn_drop_rate=attn_drop_rate, drop_path_rate=drop_path_rate,
norm_layer=norm_layer,init_values=init_values) if decoder_depth>0 else nn.Identity()
self.encoder_to_channel = nn.Linear(encoder_embed_dim, 32)
self.channel = Channels()
self.channel_to_decoder = nn.Linear(32, decoder_embed_dim)
self.head = nn.Linear(decoder_embed_dim, IMGC_NUMCLASS)
self.bit_per_digit = 12
self.vq_layer = VectorQuantizer(num_embeddings=2**self.bit_per_digit,
embedding_dim=32)
def _init_weights(self, m):
if isinstance(m, nn.Linear):
nn.init.xavier_uniform_(m.weight)
if isinstance(m, nn.Linear) and m.bias is not None:
nn.init.constant_(m.bias, 0)
elif isinstance(m, nn.LayerNorm):
nn.init.constant_(m.bias, 0)
nn.init.constant_(m.weight, 1.0)
def get_num_layers(self):
return len(self.blocks)
@torch.jit.ignore
def no_weight_decay(self):
return {'pos_embed', 'cls_token', 'mask_token'}
def forward(self, img, bm_pos, target=None, _eval=False, test_snr=200):
if _eval:
self.eval()
else:
self.train()
out = {}
if self.training:
noise_snr, noise_var = noise_gen(self.training)
noise_var,noise_snr = noise_var.cuda(), noise_snr.cpu().item()
else:
noise_var = torch.FloatTensor([1]) * 10**(-test_snr/20)
noise_snr = test_snr
x, cls_out = self.img_encoder(img, bm_pos, target)
x = self.encoder_to_channel(x)
# x, vq_loss = self.vq_layer(x, noise_snr, self.bit_per_digit)
# x = power_norm_batchwise(x)
# x = self.channel.AWGN(x, noise_var.item())
x = self.channel_to_decoder(x)
x = self.img_decoder(x)
# x = self.head(x.view(x.shape[0],-1))
x = self.head(x.mean(1))
out['out_x'] = x
out['out_c'] = cls_out
# out['vq_loss'] = vq_loss
return out
@register_model
def ViT_Van_model(pretrained=False, **kwargs):
model = ViT_Van_CLS(
img_size=224,
patch_size=16,
encoder_embed_dim=384,
encoder_depth=4,
encoder_num_heads=6,
decoder_embed_dim=12,
decoder_depth=1,
decoder_num_heads=4,
mlp_ratio=4,
qkv_bias=True,
norm_layer=partial(nn.LayerNorm, eps=1e-6),
**kwargs)
model.default_cfg = _cfg()
if pretrained:
checkpoint = torch.load(
kwargs["init_ckpt"], map_location="cpu"
)
model.load_state_dict(checkpoint["model"])
return model
@register_model
def ViT_FIM_model_S(pretrained=False, **kwargs):
model = ViT_FIM_CLS(
img_size=224,
patch_size=16,
encoder_embed_dim=384,
encoder_depth=4,
encoder_num_heads=6,
decoder_embed_dim=12,
decoder_depth=1,
decoder_num_heads=4,
mlp_ratio=4,
qkv_bias=True,
norm_layer=partial(nn.LayerNorm, eps=1e-6),
**kwargs)
model.default_cfg = _cfg()
if pretrained:
checkpoint = torch.load(
kwargs["init_ckpt"], map_location="cpu"
)
model.load_state_dict(checkpoint["model"])
return model
@register_model
def ViT_FIM_model_L(pretrained=False, **kwargs):
model = ViT_FIM_CLS(
img_size=224,
patch_size=16,
encoder_embed_dim=784,
encoder_depth=8,
encoder_num_heads=6,
decoder_embed_dim=12,
decoder_depth=4,
decoder_num_heads=8,
mlp_ratio=4,
qkv_bias=True,
norm_layer=partial(nn.LayerNorm, eps=1e-6),
**kwargs)
model.default_cfg = _cfg()
if pretrained:
checkpoint = torch.load(
kwargs["init_ckpt"], map_location="cpu"
)
model.load_state_dict(checkpoint["model"])
return model