272 lines
14 KiB
Python
272 lines
14 KiB
Python
# coding=utf-8
|
|
# Copyright 2022 SenseTime and The HuggingFace Inc. team. All rights reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
""" DETA model configuration"""
|
|
|
|
|
|
from ...configuration_utils import PretrainedConfig
|
|
from ...utils import logging
|
|
from ..auto import CONFIG_MAPPING
|
|
|
|
|
|
logger = logging.get_logger(__name__)
|
|
|
|
|
|
from ..deprecated._archive_maps import DETA_PRETRAINED_CONFIG_ARCHIVE_MAP # noqa: F401, E402
|
|
|
|
|
|
class DetaConfig(PretrainedConfig):
|
|
r"""
|
|
This is the configuration class to store the configuration of a [`DetaModel`]. It is used to instantiate a DETA
|
|
model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
|
|
defaults will yield a similar configuration to that of the DETA
|
|
[SenseTime/deformable-detr](https://huggingface.co/SenseTime/deformable-detr) architecture.
|
|
|
|
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
|
|
documentation from [`PretrainedConfig`] for more information.
|
|
|
|
Args:
|
|
backbone_config (`PretrainedConfig` or `dict`, *optional*, defaults to `ResNetConfig()`):
|
|
The configuration of the backbone model.
|
|
backbone (`str`, *optional*):
|
|
Name of backbone to use when `backbone_config` is `None`. If `use_pretrained_backbone` is `True`, this
|
|
will load the corresponding pretrained weights from the timm or transformers library. If `use_pretrained_backbone`
|
|
is `False`, this loads the backbone's config and uses that to initialize the backbone with random weights.
|
|
use_pretrained_backbone (`bool`, *optional*, `False`):
|
|
Whether to use pretrained weights for the backbone.
|
|
use_timm_backbone (`bool`, *optional*, `False`):
|
|
Whether to load `backbone` from the timm library. If `False`, the backbone is loaded from the transformers
|
|
library.
|
|
backbone_kwargs (`dict`, *optional*):
|
|
Keyword arguments to be passed to AutoBackbone when loading from a checkpoint
|
|
e.g. `{'out_indices': (0, 1, 2, 3)}`. Cannot be specified if `backbone_config` is set.
|
|
num_queries (`int`, *optional*, defaults to 900):
|
|
Number of object queries, i.e. detection slots. This is the maximal number of objects [`DetaModel`] can
|
|
detect in a single image. In case `two_stage` is set to `True`, we use `two_stage_num_proposals` instead.
|
|
d_model (`int`, *optional*, defaults to 256):
|
|
Dimension of the layers.
|
|
encoder_layers (`int`, *optional*, defaults to 6):
|
|
Number of encoder layers.
|
|
decoder_layers (`int`, *optional*, defaults to 6):
|
|
Number of decoder layers.
|
|
encoder_attention_heads (`int`, *optional*, defaults to 8):
|
|
Number of attention heads for each attention layer in the Transformer encoder.
|
|
decoder_attention_heads (`int`, *optional*, defaults to 8):
|
|
Number of attention heads for each attention layer in the Transformer decoder.
|
|
decoder_ffn_dim (`int`, *optional*, defaults to 2048):
|
|
Dimension of the "intermediate" (often named feed-forward) layer in decoder.
|
|
encoder_ffn_dim (`int`, *optional*, defaults to 2048):
|
|
Dimension of the "intermediate" (often named feed-forward) layer in decoder.
|
|
activation_function (`str` or `function`, *optional*, defaults to `"relu"`):
|
|
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
|
|
`"relu"`, `"silu"` and `"gelu_new"` are supported.
|
|
dropout (`float`, *optional*, defaults to 0.1):
|
|
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
|
|
attention_dropout (`float`, *optional*, defaults to 0.0):
|
|
The dropout ratio for the attention probabilities.
|
|
activation_dropout (`float`, *optional*, defaults to 0.0):
|
|
The dropout ratio for activations inside the fully connected layer.
|
|
init_std (`float`, *optional*, defaults to 0.02):
|
|
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
|
|
init_xavier_std (`float`, *optional*, defaults to 1):
|
|
The scaling factor used for the Xavier initialization gain in the HM Attention map module.
|
|
encoder_layerdrop (`float`, *optional*, defaults to 0.0):
|
|
The LayerDrop probability for the encoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
|
|
for more details.
|
|
auxiliary_loss (`bool`, *optional*, defaults to `False`):
|
|
Whether auxiliary decoding losses (loss at each decoder layer) are to be used.
|
|
position_embedding_type (`str`, *optional*, defaults to `"sine"`):
|
|
Type of position embeddings to be used on top of the image features. One of `"sine"` or `"learned"`.
|
|
class_cost (`float`, *optional*, defaults to 1):
|
|
Relative weight of the classification error in the Hungarian matching cost.
|
|
bbox_cost (`float`, *optional*, defaults to 5):
|
|
Relative weight of the L1 error of the bounding box coordinates in the Hungarian matching cost.
|
|
giou_cost (`float`, *optional*, defaults to 2):
|
|
Relative weight of the generalized IoU loss of the bounding box in the Hungarian matching cost.
|
|
mask_loss_coefficient (`float`, *optional*, defaults to 1):
|
|
Relative weight of the Focal loss in the panoptic segmentation loss.
|
|
dice_loss_coefficient (`float`, *optional*, defaults to 1):
|
|
Relative weight of the DICE/F-1 loss in the panoptic segmentation loss.
|
|
bbox_loss_coefficient (`float`, *optional*, defaults to 5):
|
|
Relative weight of the L1 bounding box loss in the object detection loss.
|
|
giou_loss_coefficient (`float`, *optional*, defaults to 2):
|
|
Relative weight of the generalized IoU loss in the object detection loss.
|
|
eos_coefficient (`float`, *optional*, defaults to 0.1):
|
|
Relative classification weight of the 'no-object' class in the object detection loss.
|
|
num_feature_levels (`int`, *optional*, defaults to 5):
|
|
The number of input feature levels.
|
|
encoder_n_points (`int`, *optional*, defaults to 4):
|
|
The number of sampled keys in each feature level for each attention head in the encoder.
|
|
decoder_n_points (`int`, *optional*, defaults to 4):
|
|
The number of sampled keys in each feature level for each attention head in the decoder.
|
|
two_stage (`bool`, *optional*, defaults to `True`):
|
|
Whether to apply a two-stage deformable DETR, where the region proposals are also generated by a variant of
|
|
DETA, which are further fed into the decoder for iterative bounding box refinement.
|
|
two_stage_num_proposals (`int`, *optional*, defaults to 300):
|
|
The number of region proposals to be generated, in case `two_stage` is set to `True`.
|
|
with_box_refine (`bool`, *optional*, defaults to `True`):
|
|
Whether to apply iterative bounding box refinement, where each decoder layer refines the bounding boxes
|
|
based on the predictions from the previous layer.
|
|
focal_alpha (`float`, *optional*, defaults to 0.25):
|
|
Alpha parameter in the focal loss.
|
|
assign_first_stage (`bool`, *optional*, defaults to `True`):
|
|
Whether to assign each prediction i to the highest overlapping ground truth object if the overlap is larger than a threshold 0.7.
|
|
assign_second_stage (`bool`, *optional*, defaults to `True`):
|
|
Whether to assign second assignment procedure in the second stage closely follows the first stage assignment procedure.
|
|
disable_custom_kernels (`bool`, *optional*, defaults to `True`):
|
|
Disable the use of custom CUDA and CPU kernels. This option is necessary for the ONNX export, as custom
|
|
kernels are not supported by PyTorch ONNX export.
|
|
|
|
Examples:
|
|
|
|
```python
|
|
>>> from transformers import DetaConfig, DetaModel
|
|
|
|
>>> # Initializing a DETA SenseTime/deformable-detr style configuration
|
|
>>> configuration = DetaConfig()
|
|
|
|
>>> # Initializing a model (with random weights) from the SenseTime/deformable-detr style configuration
|
|
>>> model = DetaModel(configuration)
|
|
|
|
>>> # Accessing the model configuration
|
|
>>> configuration = model.config
|
|
```"""
|
|
|
|
model_type = "deta"
|
|
attribute_map = {
|
|
"hidden_size": "d_model",
|
|
"num_attention_heads": "encoder_attention_heads",
|
|
}
|
|
|
|
def __init__(
|
|
self,
|
|
backbone_config=None,
|
|
backbone=None,
|
|
use_pretrained_backbone=False,
|
|
use_timm_backbone=False,
|
|
backbone_kwargs=None,
|
|
num_queries=900,
|
|
max_position_embeddings=2048,
|
|
encoder_layers=6,
|
|
encoder_ffn_dim=2048,
|
|
encoder_attention_heads=8,
|
|
decoder_layers=6,
|
|
decoder_ffn_dim=1024,
|
|
decoder_attention_heads=8,
|
|
encoder_layerdrop=0.0,
|
|
is_encoder_decoder=True,
|
|
activation_function="relu",
|
|
d_model=256,
|
|
dropout=0.1,
|
|
attention_dropout=0.0,
|
|
activation_dropout=0.0,
|
|
init_std=0.02,
|
|
init_xavier_std=1.0,
|
|
return_intermediate=True,
|
|
auxiliary_loss=False,
|
|
position_embedding_type="sine",
|
|
num_feature_levels=5,
|
|
encoder_n_points=4,
|
|
decoder_n_points=4,
|
|
two_stage=True,
|
|
two_stage_num_proposals=300,
|
|
with_box_refine=True,
|
|
assign_first_stage=True,
|
|
assign_second_stage=True,
|
|
class_cost=1,
|
|
bbox_cost=5,
|
|
giou_cost=2,
|
|
mask_loss_coefficient=1,
|
|
dice_loss_coefficient=1,
|
|
bbox_loss_coefficient=5,
|
|
giou_loss_coefficient=2,
|
|
eos_coefficient=0.1,
|
|
focal_alpha=0.25,
|
|
disable_custom_kernels=True,
|
|
**kwargs,
|
|
):
|
|
if use_pretrained_backbone:
|
|
raise ValueError("Pretrained backbones are not supported yet.")
|
|
|
|
if backbone_config is not None and backbone is not None:
|
|
raise ValueError("You can't specify both `backbone` and `backbone_config`.")
|
|
|
|
if backbone_config is None and backbone is None:
|
|
logger.info("`backbone_config` is `None`. Initializing the config with the default `ResNet` backbone.")
|
|
backbone_config = CONFIG_MAPPING["resnet"](out_features=["stage2", "stage3", "stage4"])
|
|
else:
|
|
if isinstance(backbone_config, dict):
|
|
backbone_model_type = backbone_config.pop("model_type")
|
|
config_class = CONFIG_MAPPING[backbone_model_type]
|
|
backbone_config = config_class.from_dict(backbone_config)
|
|
|
|
if backbone_kwargs is not None and backbone_kwargs and backbone_config is not None:
|
|
raise ValueError("You can't specify both `backbone_kwargs` and `backbone_config`.")
|
|
|
|
self.backbone_config = backbone_config
|
|
self.backbone = backbone
|
|
self.use_pretrained_backbone = use_pretrained_backbone
|
|
self.use_timm_backbone = use_timm_backbone
|
|
self.backbone_kwargs = backbone_kwargs
|
|
self.num_queries = num_queries
|
|
self.max_position_embeddings = max_position_embeddings
|
|
self.d_model = d_model
|
|
self.encoder_ffn_dim = encoder_ffn_dim
|
|
self.encoder_layers = encoder_layers
|
|
self.encoder_attention_heads = encoder_attention_heads
|
|
self.decoder_ffn_dim = decoder_ffn_dim
|
|
self.decoder_layers = decoder_layers
|
|
self.decoder_attention_heads = decoder_attention_heads
|
|
self.dropout = dropout
|
|
self.attention_dropout = attention_dropout
|
|
self.activation_dropout = activation_dropout
|
|
self.activation_function = activation_function
|
|
self.init_std = init_std
|
|
self.init_xavier_std = init_xavier_std
|
|
self.encoder_layerdrop = encoder_layerdrop
|
|
self.auxiliary_loss = auxiliary_loss
|
|
self.position_embedding_type = position_embedding_type
|
|
# deformable attributes
|
|
self.num_feature_levels = num_feature_levels
|
|
self.encoder_n_points = encoder_n_points
|
|
self.decoder_n_points = decoder_n_points
|
|
self.two_stage = two_stage
|
|
self.two_stage_num_proposals = two_stage_num_proposals
|
|
self.with_box_refine = with_box_refine
|
|
self.assign_first_stage = assign_first_stage
|
|
self.assign_second_stage = assign_second_stage
|
|
if two_stage is True and with_box_refine is False:
|
|
raise ValueError("If two_stage is True, with_box_refine must be True.")
|
|
# Hungarian matcher
|
|
self.class_cost = class_cost
|
|
self.bbox_cost = bbox_cost
|
|
self.giou_cost = giou_cost
|
|
# Loss coefficients
|
|
self.mask_loss_coefficient = mask_loss_coefficient
|
|
self.dice_loss_coefficient = dice_loss_coefficient
|
|
self.bbox_loss_coefficient = bbox_loss_coefficient
|
|
self.giou_loss_coefficient = giou_loss_coefficient
|
|
self.eos_coefficient = eos_coefficient
|
|
self.focal_alpha = focal_alpha
|
|
self.disable_custom_kernels = disable_custom_kernels
|
|
super().__init__(is_encoder_decoder=is_encoder_decoder, **kwargs)
|
|
|
|
@property
|
|
def num_attention_heads(self) -> int:
|
|
return self.encoder_attention_heads
|
|
|
|
@property
|
|
def hidden_size(self) -> int:
|
|
return self.d_model
|