Spaces:
Paused
Paused
File size: 15,453 Bytes
3650c12 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 |
# --------------------------------------------------------
# Based on BEiT, timm, DINO and DeiT code bases
# https://github.com/microsoft/unilm/tree/master/beit
# https://github.com/rwightman/pytorch-image-models/tree/master/timm
# https://github.com/facebookresearch/deit
# https://github.com/facebookresearch/dino
# --------------------------------------------------------'
from functools import partial
import torch
import torch.nn as nn
import torch.utils.checkpoint as cp
from .videomaev2_finetune import (
Block,
PatchEmbed,
_cfg,
get_sinusoid_encoding_table,
)
from .videomaev2_finetune import trunc_normal_ as __call_trunc_normal_
def trunc_normal_(tensor, mean=0., std=1.):
__call_trunc_normal_(tensor, mean=mean, std=std, a=-std, b=std)
class PretrainVisionTransformerEncoder(nn.Module):
""" Vision Transformer with support for patch or hybrid CNN input stage
"""
def __init__(self,
img_size=224,
patch_size=16,
in_chans=3,
num_classes=0,
embed_dim=768,
depth=12,
num_heads=12,
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=None,
tubelet_size=2,
use_learnable_pos_emb=False,
with_cp=False,
all_frames=16,
cos_attn=False):
super().__init__()
self.num_classes = num_classes
# num_features for consistency with other models
self.num_features = self.embed_dim = embed_dim
self.patch_embed = PatchEmbed(
img_size=img_size,
patch_size=patch_size,
in_chans=in_chans,
embed_dim=embed_dim,
num_frames=all_frames,
tubelet_size=tubelet_size)
num_patches = self.patch_embed.num_patches
self.with_cp = with_cp
if use_learnable_pos_emb:
self.pos_embed = nn.Parameter(
torch.zeros(1, num_patches + 1, embed_dim))
else:
# sine-cosine positional embeddings
self.pos_embed = get_sinusoid_encoding_table(
num_patches, embed_dim)
dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)
] # stochastic depth decay rule
self.blocks = nn.ModuleList([
Block(
dim=embed_dim,
num_heads=num_heads,
mlp_ratio=mlp_ratio,
qkv_bias=qkv_bias,
qk_scale=qk_scale,
drop=drop_rate,
attn_drop=attn_drop_rate,
drop_path=dpr[i],
norm_layer=norm_layer,
init_values=init_values,
cos_attn=cos_attn) for i in range(depth)
])
self.norm = norm_layer(embed_dim)
self.head = nn.Linear(
embed_dim, num_classes) if num_classes > 0 else nn.Identity()
if use_learnable_pos_emb:
trunc_normal_(self.pos_embed, std=.02)
self.apply(self._init_weights)
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'}
def get_classifier(self):
return self.head
def reset_classifier(self, num_classes, global_pool=''):
self.num_classes = num_classes
self.head = nn.Linear(
self.embed_dim, num_classes) if num_classes > 0 else nn.Identity()
def forward_features(self, x, mask):
x = self.patch_embed(x)
x = x + self.pos_embed.type_as(x).to(x.device).clone().detach()
B, _, C = x.shape
x_vis = x[~mask].reshape(B, -1, C) # ~mask means visible
for blk in self.blocks:
if self.with_cp:
x_vis = cp.checkpoint(blk, x_vis)
else:
x_vis = blk(x_vis)
x_vis = self.norm(x_vis)
return x_vis
def forward(self, x, mask):
x = self.forward_features(x, mask)
x = self.head(x)
return x
class PretrainVisionTransformerDecoder(nn.Module):
""" Vision Transformer with support for patch or hybrid CNN input stage
"""
def __init__(self,
patch_size=16,
num_classes=768,
embed_dim=768,
depth=12,
num_heads=12,
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=None,
num_patches=196,
tubelet_size=2,
with_cp=False,
cos_attn=False):
super().__init__()
self.num_classes = num_classes
assert num_classes == 3 * tubelet_size * patch_size**2
# num_features for consistency with other models
self.num_features = self.embed_dim = embed_dim
self.patch_size = patch_size
self.with_cp = with_cp
dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)
] # stochastic depth decay rule
self.blocks = nn.ModuleList([
Block(
dim=embed_dim,
num_heads=num_heads,
mlp_ratio=mlp_ratio,
qkv_bias=qkv_bias,
qk_scale=qk_scale,
drop=drop_rate,
attn_drop=attn_drop_rate,
drop_path=dpr[i],
norm_layer=norm_layer,
init_values=init_values,
cos_attn=cos_attn) for i in range(depth)
])
self.norm = norm_layer(embed_dim)
self.head = nn.Linear(
embed_dim, num_classes) if num_classes > 0 else nn.Identity()
self.apply(self._init_weights)
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'}
def get_classifier(self):
return self.head
def reset_classifier(self, num_classes, global_pool=''):
self.num_classes = num_classes
self.head = nn.Linear(
self.embed_dim, num_classes) if num_classes > 0 else nn.Identity()
def forward(self, x, return_token_num):
for blk in self.blocks:
if self.with_cp:
x = cp.checkpoint(blk, x)
else:
x = blk(x)
if return_token_num > 0:
# only return the mask tokens predict pixels
x = self.head(self.norm(x[:, -return_token_num:]))
else:
# [B, N, 3*16^2]
x = self.head(self.norm(x))
return x
class PretrainVisionTransformer(nn.Module):
""" Vision Transformer with support for patch or hybrid CNN input stage
"""
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=1536, # 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,
tubelet_size=2,
num_classes=0, # avoid the error from create_fn in timm
in_chans=0, # avoid the error from create_fn in timm
with_cp=False,
all_frames=16,
cos_attn=False,
):
super().__init__()
self.encoder = PretrainVisionTransformerEncoder(
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,
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,
tubelet_size=tubelet_size,
use_learnable_pos_emb=use_learnable_pos_emb,
with_cp=with_cp,
all_frames=all_frames,
cos_attn=cos_attn)
self.decoder = PretrainVisionTransformerDecoder(
patch_size=patch_size,
num_patches=self.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,
tubelet_size=tubelet_size,
with_cp=with_cp,
cos_attn=cos_attn)
self.encoder_to_decoder = nn.Linear(
encoder_embed_dim, decoder_embed_dim, bias=False)
self.mask_token = nn.Parameter(torch.zeros(1, 1, decoder_embed_dim))
self.pos_embed = get_sinusoid_encoding_table(
self.encoder.patch_embed.num_patches, decoder_embed_dim)
trunc_normal_(self.mask_token, std=.02)
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, x, mask, decode_mask=None):
decode_vis = mask if decode_mask is None else ~decode_mask
x_vis = self.encoder(x, mask) # [B, N_vis, C_e]
x_vis = self.encoder_to_decoder(x_vis) # [B, N_vis, C_d]
B, N_vis, C = x_vis.shape
# we don't unshuffle the correct visible token order,
# but shuffle the pos embedding accorddingly.
expand_pos_embed = self.pos_embed.expand(B, -1, -1).type_as(x).to(
x.device).clone().detach()
pos_emd_vis = expand_pos_embed[~mask].reshape(B, -1, C)
pos_emd_mask = expand_pos_embed[decode_vis].reshape(B, -1, C)
# [B, N, C_d]
x_full = torch.cat(
[x_vis + pos_emd_vis, self.mask_token + pos_emd_mask], dim=1)
# NOTE: if N_mask==0, the shape of x is [B, N_mask, 3 * 16 * 16]
x = self.decoder(x_full, pos_emd_mask.shape[1])
return x
def pretrain_videomae_small_patch16_224(pretrained=False, **kwargs):
model = PretrainVisionTransformer(
img_size=224,
patch_size=16,
encoder_embed_dim=384,
encoder_depth=12,
encoder_num_heads=6,
encoder_num_classes=0,
decoder_num_classes=1536, # 16 * 16 * 3 * 2
decoder_embed_dim=192,
decoder_num_heads=3,
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
def pretrain_videomae_base_patch16_224(pretrained=False, **kwargs):
model = PretrainVisionTransformer(
img_size=224,
patch_size=16,
encoder_embed_dim=768,
encoder_depth=12,
encoder_num_heads=12,
encoder_num_classes=0,
decoder_num_classes=1536, # 16 * 16 * 3 * 2
decoder_embed_dim=384,
decoder_num_heads=6,
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
def pretrain_videomae_large_patch16_224(pretrained=False, **kwargs):
model = PretrainVisionTransformer(
img_size=224,
patch_size=16,
encoder_embed_dim=1024,
encoder_depth=24,
encoder_num_heads=16,
encoder_num_classes=0,
decoder_num_classes=1536, # 16 * 16 * 3 * 2
decoder_embed_dim=512,
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
def pretrain_videomae_huge_patch16_224(pretrained=False, **kwargs):
model = PretrainVisionTransformer(
img_size=224,
patch_size=16,
encoder_embed_dim=1280,
encoder_depth=32,
encoder_num_heads=16,
encoder_num_classes=0,
decoder_num_classes=1536, # 16 * 16 * 3 * 2
decoder_embed_dim=512,
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
def pretrain_videomae_giant_patch14_224(pretrained=False, **kwargs):
model = PretrainVisionTransformer(
img_size=224,
patch_size=14,
encoder_embed_dim=1408,
encoder_depth=40,
encoder_num_heads=16,
encoder_num_classes=0,
decoder_num_classes=1176, # 14 * 14 * 3 * 2,
decoder_embed_dim=512,
decoder_num_heads=8,
mlp_ratio=48 / 11,
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
|