File size: 22,744 Bytes
29ebc1d | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 | # ruff: noqa
# fmt: off
"""Prompt Depth Anything, vendored verbatim from DepthAnything/PromptDA @ 9161547ca7151e14fcc8f490ad3cc02fa9437483 (promptda/ and the local torchhub
copy of facebookresearch/dinov2 it loads): the files below concatenated, package-internal imports replaced by
relative imports of the other vendored files (generated by scripts/model_ports/common/vendor_flat.py; do not edit by hand). Code license: Apache-2.0."""
# ==============================================================================================================
# PromptDA/promptda/utils/logger.py
# ==============================================================================================================
import os
class Log:
log_on = True # fast switch
used_tags = dict() # To keep track of used tags
_is_main_cached = None # Cache to store the main process check result
@staticmethod
def is_main_process():
if Log._is_main_cached is not None:
return Log._is_main_cached
try:
from pytorch_lightning.utilities import rank_zero_only
if rank_zero_only.rank == 0:
Log._is_main_cached = True
else:
Log._is_main_cached = False
except:
Log._is_main_cached = True
return Log._is_main_cached
@staticmethod
def _should_log(tag):
"""
Determine if the log information should be recorded.
Conditions: log function is enabled, current process is the main process, and the tag has not been used.
"""
if not Log.log_on:
return False
if not Log.is_main_process():
return False
if tag is None:
return True
if '__' in tag:
num = int(tag.split('__')[-1])
tag = tag.split('__')[0] # can output num same information
else:
num = 3 # default 3
if tag not in Log.used_tags:
Log.used_tags[tag] = num
Log.used_tags[tag] -= 1
if Log.used_tags[tag] >= 0:
return True
else:
return False
@staticmethod
def info(*args, tag=None):
"""
Output INFO level log information.
"""
if Log._should_log(tag):
print("\033[1;32m[INFO]\033[0;0m", *args)
@staticmethod
def warn(*args, tag=None):
"""
Output WARN level log information.
"""
if Log._should_log(tag):
print("\033[1;35m[WARN]\033[0;0m", *args)
@staticmethod
def error(*args, tag=None):
print("\033[1;31m[ERROR]\033[0;0m", *args)
@staticmethod
def debug(*args, tag=None):
"""
Output DEBUG level log information.
"""
if Log._should_log(tag) and 'HT_DEBUG' in os.environ and os.environ['HT_DEBUG'] == '1':
print("\033[1;33m[DEBUG]\033[0;0m", *args)
# ==============================================================================================================
# PromptDA/promptda/model/blocks.py
# ==============================================================================================================
import torch.nn as nn
import torch
import torch.nn.functional as F
import os
import numpy as np
def _make_fusion_block(features, use_bn, size=None):
return FeatureFusionDepthBlock(
features,
nn.ReLU(False),
deconv=False,
bn=use_bn,
expand=False,
align_corners=True,
size=size,
)
def _make_scratch(in_shape, out_shape, groups=1, expand=False):
scratch = nn.Module()
out_shape1 = out_shape
out_shape2 = out_shape
out_shape3 = out_shape
if len(in_shape) >= 4:
out_shape4 = out_shape
if expand:
out_shape1 = out_shape
out_shape2 = out_shape*2
out_shape3 = out_shape*4
if len(in_shape) >= 4:
out_shape4 = out_shape*8
scratch.layer1_rn = nn.Conv2d(
in_shape[0], out_shape1, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
)
scratch.layer2_rn = nn.Conv2d(
in_shape[1], out_shape2, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
)
scratch.layer3_rn = nn.Conv2d(
in_shape[2], out_shape3, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
)
if len(in_shape) >= 4:
scratch.layer4_rn = nn.Conv2d(
in_shape[3], out_shape4, kernel_size=3, stride=1, padding=1, bias=False, groups=groups
)
return scratch
class ResidualConvUnit(nn.Module):
"""Residual convolution module.
"""
def __init__(self, features, activation, bn):
"""Init.
Args:
features (int): number of features
"""
super().__init__()
self.bn = bn
self.groups = 1
self.conv1 = nn.Conv2d(
features, features, kernel_size=3, stride=1, padding=1, bias=True, groups=self.groups
)
self.conv2 = nn.Conv2d(
features, features, kernel_size=3, stride=1, padding=1, bias=True, groups=self.groups
)
if self.bn == True:
self.bn1 = nn.BatchNorm2d(features)
self.bn2 = nn.BatchNorm2d(features)
self.activation = activation
self.skip_add = nn.quantized.FloatFunctional()
def forward(self, x):
"""Forward pass.
Args:
x (tensor): input
Returns:
tensor: output
"""
out = self.activation(x)
out = self.conv1(out)
if self.bn == True:
out = self.bn1(out)
out = self.activation(out)
out = self.conv2(out)
if self.bn == True:
out = self.bn2(out)
if self.groups > 1:
out = self.conv_merge(out)
return self.skip_add.add(out, x)
class FeatureFusionBlock(nn.Module):
"""Feature fusion block.
"""
def __init__(self, features, activation, deconv=False, bn=False, expand=False, align_corners=True, size=None):
"""Init.
Args:
features (int): number of features
"""
super(FeatureFusionBlock, self).__init__()
self.deconv = deconv
self.align_corners = align_corners
self.groups = 1
self.expand = expand
out_features = features
if self.expand == True:
out_features = features//2
self.out_conv = nn.Conv2d(
features, out_features, kernel_size=1, stride=1, padding=0, bias=True, groups=1)
self.resConfUnit1 = ResidualConvUnit(features, activation, bn)
self.resConfUnit2 = ResidualConvUnit(features, activation, bn)
self.skip_add = nn.quantized.FloatFunctional()
self.size = size
def forward(self, *xs, size=None):
"""Forward pass.
Returns:
tensor: output
"""
output = xs[0]
if len(xs) == 2:
res = self.resConfUnit1(xs[1])
output = self.skip_add.add(output, res)
output = self.resConfUnit2(output)
if (size is None) and (self.size is None):
modifier = {"scale_factor": 2}
elif size is None:
modifier = {"size": self.size}
else:
modifier = {"size": size}
output = nn.functional.interpolate(
output, **modifier, mode="bilinear", align_corners=self.align_corners
)
output = self.out_conv(output)
return output
class FeatureFusionControlBlock(FeatureFusionBlock):
"""Feature fusion block.
"""
def __init__(self, features, activation, deconv=False, bn=False, expand=False, align_corners=True, size=None):
"""Init.
Args:
features (int): number of features
"""
super.__init__(features, activation, deconv,
bn, expand, align_corners, size)
self.copy_block = FeatureFusionBlock(
features, activation, deconv, bn, expand, align_corners, size)
def forward(self, *xs, size=None):
"""Forward pass.
Returns:
tensor: output
"""
output = xs[0]
if len(xs) == 2:
res = self.resConfUnit1(xs[1])
output = self.skip_add.add(output, res)
output = self.resConfUnit2(output)
if (size is None) and (self.size is None):
modifier = {"scale_factor": 2}
elif size is None:
modifier = {"size": self.size}
else:
modifier = {"size": size}
output = nn.functional.interpolate(
output, **modifier, mode="bilinear", align_corners=self.align_corners
)
output = self.out_conv(output)
return output
def zero_module(module):
"""
Zero out the parameters of a module and return it.
"""
for p in module.parameters():
p.detach().zero_()
return module
class FeatureFusionDepthBlock(nn.Module):
"""Feature fusion block.
"""
def __init__(self, features, activation, deconv=False, bn=False, expand=False, align_corners=True, size=None):
"""Init.
Args:
features (int): number of features
"""
super(FeatureFusionDepthBlock, self).__init__()
self.deconv = deconv
self.align_corners = align_corners
self.groups = 1
self.expand = expand
out_features = features
if self.expand == True:
out_features = features//2
self.out_conv = nn.Conv2d(
features, out_features, kernel_size=1, stride=1, padding=0, bias=True, groups=1)
self.resConfUnit1 = ResidualConvUnit(features, activation, bn)
self.resConfUnit2 = ResidualConvUnit(features, activation, bn)
self.resConfUnit_depth = nn.Sequential(
nn.Conv2d(1, features, kernel_size=3, stride=1,
padding=1, bias=True, groups=1),
activation,
nn.Conv2d(features, features, kernel_size=3,
stride=1, padding=1, bias=True, groups=1),
activation,
zero_module(
nn.Conv2d(features, features, kernel_size=3,
stride=1, padding=1, bias=True, groups=1)
)
)
self.skip_add = nn.quantized.FloatFunctional()
self.size = size
def forward(self, *xs, prompt_depth=None, size=None):
"""Forward pass.
Returns:
tensor: output
"""
output = xs[0]
if len(xs) == 2:
res = self.resConfUnit1(xs[1])
output = self.skip_add.add(output, res)
output = self.resConfUnit2(output)
if prompt_depth is not None:
prompt_depth = F.interpolate(
prompt_depth, output.shape[2:], mode='bilinear', align_corners=False)
res = self.resConfUnit_depth(prompt_depth)
output = self.skip_add.add(output, res)
if (size is None) and (self.size is None):
modifier = {"scale_factor": 2}
elif size is None:
modifier = {"size": self.size}
else:
modifier = {"size": size}
output = nn.functional.interpolate(
output, **modifier, mode="bilinear", align_corners=self.align_corners
)
output = self.out_conv(output)
return output
# ==============================================================================================================
# PromptDA/promptda/model/dpt.py
# ==============================================================================================================
# Copyright (c) 2024, Depth Anything V2
# https://github.com/DepthAnything/Depth-Anything-V2/blob/main/depth_anything_v2/dpt.py
import torch
import torch.nn as nn
import torch.nn.functional as F
class DPTHead(nn.Module):
def __init__(self,
nclass,
in_channels,
features=256,
out_channels=[256, 512, 1024, 1024],
use_bn=False,
use_clstoken=False,
output_act='sigmoid'):
super(DPTHead, self).__init__()
self.nclass = nclass
self.use_clstoken = use_clstoken
self.projects = nn.ModuleList([
nn.Conv2d(
in_channels=in_channels,
out_channels=out_channel,
kernel_size=1,
stride=1,
padding=0,
) for out_channel in out_channels
])
self.resize_layers = nn.ModuleList([
nn.ConvTranspose2d(
in_channels=out_channels[0],
out_channels=out_channels[0],
kernel_size=4,
stride=4,
padding=0),
nn.ConvTranspose2d(
in_channels=out_channels[1],
out_channels=out_channels[1],
kernel_size=2,
stride=2,
padding=0),
nn.Identity(),
nn.Conv2d(
in_channels=out_channels[3],
out_channels=out_channels[3],
kernel_size=3,
stride=2,
padding=1)
])
if use_clstoken:
self.readout_projects = nn.ModuleList()
for _ in range(len(self.projects)):
self.readout_projects.append(
nn.Sequential(
nn.Linear(2 * in_channels, in_channels),
nn.GELU()))
self.scratch = _make_scratch(
out_channels,
features,
groups=1,
expand=False,
)
self.scratch.stem_transpose = None
self.scratch.refinenet1 = _make_fusion_block(
features, use_bn)
self.scratch.refinenet2 = _make_fusion_block(
features, use_bn)
self.scratch.refinenet3 = _make_fusion_block(
features, use_bn)
self.scratch.refinenet4 = _make_fusion_block(
features, use_bn)
head_features_1 = features
head_features_2 = 32
act_func = nn.Sigmoid() if output_act == 'sigmoid' else nn.Identity()
if nclass > 1:
self.scratch.output_conv = nn.Sequential(
nn.Conv2d(head_features_1, head_features_1,
kernel_size=3, stride=1, padding=1),
nn.ReLU(True),
nn.Conv2d(head_features_1, nclass,
kernel_size=1, stride=1, padding=0),
)
else:
self.scratch.output_conv1 = nn.Conv2d(
head_features_1, head_features_1 // 2, kernel_size=3, stride=1, padding=1)
self.scratch.output_conv2 = nn.Sequential(
nn.Conv2d(head_features_1 // 2, head_features_2,
kernel_size=3, stride=1, padding=1),
nn.ReLU(True),
nn.Conv2d(head_features_2, 1, kernel_size=1,
stride=1, padding=0),
act_func,
)
def forward(self, out_features, patch_h, patch_w, prompt_depth=None):
out = []
for i, x in enumerate(out_features):
if self.use_clstoken:
x, cls_token = x[0], x[1]
readout = cls_token.unsqueeze(1).expand_as(x)
x = self.readout_projects[i](torch.cat((x, readout), -1))
else:
x = x[0]
x = x.permute(0, 2, 1).reshape(
(x.shape[0], x.shape[-1], patch_h, patch_w))
x = self.projects[i](x)
x = self.resize_layers[i](x)
out.append(x)
layer_1, layer_2, layer_3, layer_4 = out
layer_1_rn = self.scratch.layer1_rn(layer_1)
layer_2_rn = self.scratch.layer2_rn(layer_2)
layer_3_rn = self.scratch.layer3_rn(layer_3)
layer_4_rn = self.scratch.layer4_rn(layer_4)
path_4 = self.scratch.refinenet4(
layer_4_rn, size=layer_3_rn.shape[2:], prompt_depth=prompt_depth)
path_3 = self.scratch.refinenet3(
path_4, layer_3_rn, size=layer_2_rn.shape[2:], prompt_depth=prompt_depth)
path_2 = self.scratch.refinenet2(
path_3, layer_2_rn, size=layer_1_rn.shape[2:], prompt_depth=prompt_depth)
path_1 = self.scratch.refinenet1(
path_2, layer_1_rn, prompt_depth=prompt_depth)
out = self.scratch.output_conv1(path_1)
out_feat = F.interpolate(
out, (int(patch_h * 14), int(patch_w * 14)),
mode="bilinear", align_corners=True)
out = self.scratch.output_conv2(out_feat)
return out
# ==============================================================================================================
# PromptDA/promptda/model/config.py
# ==============================================================================================================
model_configs = {
'vits': {'encoder': 'vits', 'features': 64, 'out_channels': [48, 96, 192, 384], 'layer_idxs': [2, 5, 8, 11]},
'vitb': {'encoder': 'vitb', 'features': 128, 'out_channels': [96, 192, 384, 768], 'layer_idxs': [2, 5, 8, 11]},
'vitl': {'encoder': 'vitl', 'features': 256, 'out_channels': [256, 512, 1024, 1024], 'layer_idxs': [4, 11, 17, 23]},
'vitg': {'encoder': 'vitg', 'features': 384, 'out_channels': [1536, 1536, 1536, 1536], 'layer_idxs': [9, 19, 29, 39]}
}
# ==============================================================================================================
# PromptDA/promptda/promptda.py
# ==============================================================================================================
import torch
import torch.nn as nn
import os
from pathlib import Path
from huggingface_hub import hf_hub_download
class PromptDA(nn.Module):
patch_size = 14 # patch size of the pretrained dinov2 model
use_bn = False
use_clstoken = False
output_act = 'sigmoid'
def __init__(self,
encoder='vitl',
ckpt_path='checkpoints/promptda_vitl.ckpt'):
super().__init__()
model_config = model_configs[encoder]
self.encoder = encoder
self.model_config = model_config
module_path = Path(__file__) # From anywhere else: module_path = Path(inspect.getfile(PromptDA))
package_base_dir = str(Path(*module_path.parts[:-2])) # extract path to PromptDA module, then resolve to repo base dir for dinov2 load
self.pretrained = torch.hub.load(
f'{package_base_dir}/torchhub/facebookresearch_dinov2_main',
'dinov2_{:}14'.format(encoder),
source='local',
pretrained=False)
dim = self.pretrained.blocks[0].attn.qkv.in_features
self.depth_head = DPTHead(nclass=1,
in_channels=dim,
features=model_config['features'],
out_channels=model_config['out_channels'],
use_bn=self.use_bn,
use_clstoken=self.use_clstoken,
output_act=self.output_act)
# mean and std of the pretrained dinov2 model
self.register_buffer('_mean', torch.tensor(
[0.485, 0.456, 0.406]).view(1, 3, 1, 1))
self.register_buffer('_std', torch.tensor(
[0.229, 0.224, 0.225]).view(1, 3, 1, 1))
self.load_checkpoint(ckpt_path)
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path = None, model_kwargs = None, **hf_kwargs):
"""
Load a model from a checkpoint file.
### Parameters:
- `pretrained_model_name_or_path`: path to the checkpoint file or repo id.
- `model_kwargs`: additional keyword arguments to override the parameters in the checkpoint.
- `hf_kwargs`: additional keyword arguments to pass to the `hf_hub_download` function. Ignored if `pretrained_model_name_or_path` is a local path.
### Returns:
- A new instance of `MoGe` with the parameters loaded from the checkpoint.
"""
ckpt_path = None
if Path(pretrained_model_name_or_path).exists():
ckpt_path = pretrained_model_name_or_path
else:
cached_checkpoint_path = hf_hub_download(
repo_id=pretrained_model_name_or_path,
repo_type="model",
filename="model.ckpt",
**hf_kwargs
)
ckpt_path = cached_checkpoint_path
# model_config = checkpoint['model_config']
# if model_kwargs is not None:
# model_config.update(model_kwargs)
if model_kwargs is None:
model_kwargs = {}
model_kwargs.update({'ckpt_path': ckpt_path})
model = cls(**model_kwargs)
return model
def load_checkpoint(self, ckpt_path):
if os.path.exists(ckpt_path):
Log.info(f'Loading checkpoint from {ckpt_path}')
checkpoint = torch.load(ckpt_path, map_location='cpu')
self.load_state_dict(
{k[9:]: v for k, v in checkpoint['state_dict'].items()})
else:
Log.warn(f'Checkpoint {ckpt_path} not found')
def forward(self, x, prompt_depth=None):
assert prompt_depth is not None, 'prompt_depth is required'
prompt_depth, min_val, max_val = self.normalize(prompt_depth)
h, w = x.shape[-2:]
features = self.pretrained.get_intermediate_layers(
(x - self._mean) / self._std, self.model_config['layer_idxs'],
return_class_token=True)
patch_h, patch_w = h // self.patch_size, w // self.patch_size
depth = self.depth_head(features, patch_h, patch_w, prompt_depth)
depth = self.denormalize(depth, min_val, max_val)
return depth
@torch.no_grad()
def predict(self,
image: torch.Tensor,
prompt_depth: torch.Tensor):
return self.forward(image, prompt_depth)
def normalize(self,
prompt_depth: torch.Tensor):
B, C, H, W = prompt_depth.shape
min_val = torch.quantile(
prompt_depth.reshape(B, -1), 0., dim=1, keepdim=True)[:, :, None, None]
max_val = torch.quantile(
prompt_depth.reshape(B, -1), 1., dim=1, keepdim=True)[:, :, None, None]
prompt_depth = (prompt_depth - min_val) / (max_val - min_val)
return prompt_depth, min_val, max_val
def denormalize(self,
depth: torch.Tensor,
min_val: torch.Tensor,
max_val: torch.Tensor):
return depth * (max_val - min_val) + min_val
|