src/models/sam/configuration_sam.py

# coding=utf-8
# Copyright 2023 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.
""" SAM model configuration"""

import copy

from transformers.configuration_utils import PretrainedConfig
from transformers.utils import logging


logger = logging.get_logger(__name__)

SAM_PRETRAINED_CONFIG_ARCHIVE_MAP = {
    "facebook/sam-vit-huge": "https://huggingface.co/facebook/sam-vit-huge/resolve/main/config.json",
    "facebook/sam-vit-large": "https://huggingface.co/facebook/sam-vit-large/resolve/main/config.json",
    "facebook/sam-vit-base": "https://huggingface.co/facebook/sam-vit-base/resolve/main/config.json",
}


class SamPromptEncoderConfig(PretrainedConfig):
    r"""
    This is the configuration class to store the configuration of a [`SamPromptEncoder`]. The [`SamPromptEncoder`]
    module is used to encode the input 2D points and bounding boxes. Instantiating a configuration defaults will yield
    a similar configuration to that of the SAM-vit-h
    [facebook/sam-vit-huge](https://huggingface.co/facebook/sam-vit-huge) architecture.

    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
    documentation from [`PretrainedConfig`] for more information.

    Args:
        hidden_size (`int`, *optional*, defaults to 256):
            Dimensionality of the hidden states.
        image_size (`int`, *optional*, defaults to 1024):
            The expected output resolution of the image.
        patch_size (`int`, *optional*, defaults to 16):
            The size (resolution) of each patch.
        mask_input_channels (`int`, *optional*, defaults to 16):
            The number of channels to be fed to the `MaskDecoder` module.
        num_point_embeddings (`int`, *optional*, defaults to 4):
            The number of point embeddings to be used.
        hidden_act (`str`, *optional*, defaults to `"gelu"`):
            The non-linear activation function in the encoder and pooler.
    """

    def __init__(
        self,
        hidden_size=256,
        image_size=1024,
        patch_size=16,
        mask_input_channels=16,
        num_point_embeddings=4,
        hidden_act="gelu",
        layer_norm_eps=1e-6,
        **kwargs,
    ):
        super().__init__(**kwargs)
        self.hidden_size = hidden_size
        self.image_size = image_size
        self.patch_size = patch_size
        self.image_embedding_size = image_size // patch_size
        self.mask_input_channels = mask_input_channels
        self.num_point_embeddings = num_point_embeddings
        self.hidden_act = hidden_act
        self.layer_norm_eps = layer_norm_eps


class SamMaskDecoderConfig(PretrainedConfig):
    r"""
    This is the configuration class to store the configuration of a [`SamMaskDecoder`]. It is used to instantiate a SAM
    mask decoder to the specified arguments, defining the model architecture. Instantiating a configuration defaults
    will yield a similar configuration to that of the SAM-vit-h
    [facebook/sam-vit-huge](https://huggingface.co/facebook/sam-vit-huge) architecture.

    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
    documentation from [`PretrainedConfig`] for more information.

    Args:
        hidden_size (`int`, *optional*, defaults to 256):
            Dimensionality of the hidden states.
        hidden_act (`str`, *optional*, defaults to `"relu"`):
            The non-linear activation function used inside the `SamMaskDecoder` module.
        mlp_dim (`int`, *optional*, defaults to 2048):
            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
        num_hidden_layers (`int`, *optional*, defaults to 2):
            Number of hidden layers in the Transformer encoder.
        num_attention_heads (`int`, *optional*, defaults to 8):
            Number of attention heads for each attention layer in the Transformer encoder.
        attention_downsample_rate (`int`, *optional*, defaults to 2):
            The downsampling rate of the attention layer.
        num_multimask_outputs (`int`, *optional*, defaults to 3):
            The number of outputs from the `SamMaskDecoder` module. In the Segment Anything paper, this is set to 3.
        iou_head_depth (`int`, *optional*, defaults to 3):
            The number of layers in the IoU head module.
        iou_head_hidden_dim (`int`, *optional*, defaults to 256):
            The dimensionality of the hidden states in the IoU head module.
        layer_norm_eps (`float`, *optional*, defaults to 1e-6):
            The epsilon used by the layer normalization layers.

    """

    def __init__(
        self,
        hidden_size=256,
        hidden_act="relu",
        mlp_dim=2048,
        num_hidden_layers=2,
        num_attention_heads=8,
        attention_downsample_rate=2,
        num_multimask_outputs=3,
        iou_head_depth=3,
        iou_head_hidden_dim=256,
        layer_norm_eps=1e-6,
        **kwargs,
    ):
        super().__init__(**kwargs)
        self.hidden_size = hidden_size
        self.hidden_act = hidden_act
        self.mlp_dim = mlp_dim
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.attention_downsample_rate = attention_downsample_rate
        self.num_multimask_outputs = num_multimask_outputs
        self.iou_head_depth = iou_head_depth
        self.iou_head_hidden_dim = iou_head_hidden_dim
        self.layer_norm_eps = layer_norm_eps


class SamVisionConfig(PretrainedConfig):
    r"""
    This is the configuration class to store the configuration of a [`SamVisionModel`]. It is used to instantiate a SAM
    vision encoder according to the specified arguments, defining the model architecture. Instantiating a configuration
    defaults will yield a similar configuration to that of the SAM ViT-h
    [facebook/sam-vit-huge](https://huggingface.co/facebook/sam-vit-huge) architecture.

    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
    documentation from [`PretrainedConfig`] for more information.

    Args:
        hidden_size (`int`, *optional*, defaults to 768):
            Dimensionality of the encoder layers and the pooler layer.
        intermediate_size (`int`, *optional*, defaults to 6144):
            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
        projection_dim (`int`, *optional*, defaults to 512):
            Dimensionality of the projection layer in the Transformer encoder.
        output_channels (`int`, *optional*, defaults to 256):
            Dimensionality of the output channels in the Patch Encoder.
        num_hidden_layers (`int`, *optional*, defaults to 12):
            Number of hidden layers in the Transformer encoder.
        num_attention_heads (`int`, *optional*, defaults to 12):
            Number of attention heads for each attention layer in the Transformer encoder.
        num_channels (`int`, *optional*, defaults to 3):
            Number of channels in the input image.
        image_size (`int`, *optional*, defaults to 1024):
            Expected resolution. Target size of the resized input image.
        patch_size (`int`, *optional*, defaults to 16):
            Size of the patches to be extracted from the input image.
        hidden_act (`str`, *optional*, defaults to `"gelu"`):
            The non-linear activation function (function or string)
        layer_norm_eps (`float`, *optional*, defaults to 1e-6):
            The epsilon used by the layer normalization layers.
        dropout (`float`, *optional*, defaults to 0.0):
            The dropout probability.
        attention_dropout (`float`, *optional*, defaults to 0.0):
            The dropout ratio for the attention probabilities.
        initializer_range (`float`, *optional*, defaults to 1e-10):
            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
        initializer_factor (`float`, *optional*, defaults to 1.0):
            A factor for multiplying the initializer range.
        qkv_bias (`bool`, *optional*, defaults to `True`):
            Whether to add a bias to query, key, value projections.
        mlp_ratio (`float`, *optional*, defaults to 4.0):
            Ratio of mlp hidden dim to embedding dim.
        use_abs_pos (`bool`, *optional*, defaults to True):
            Whether to use absolute position embedding.
        use_rel_pos (`bool`, *optional*, defaults to True):
            Whether to use relative position embedding.
        window_size (`int`, *optional*, defaults to 14):
            Window size for relative position.
        global_attn_indexes (`List[int]`, *optional*, defaults to `[2, 5, 8, 11]`):
            The indexes of the global attention layers.
        num_pos_feats (`int`, *optional*, defaults to 128):
            The dimensionality of the position embedding.
        mlp_dim (`int`, *optional*, defaults to None):
            The dimensionality of the MLP layer in the Transformer encoder. If `None`, defaults to `mlp_ratio *
            hidden_size`.
    """

    def __init__(
        self,
        hidden_size=768,
        intermediate_size=6144,
        projection_dim=512,
        output_channels=256,
        num_hidden_layers=12,
        num_attention_heads=12,
        num_channels=3,
        image_size=1024,
        patch_size=16,
        hidden_act="gelu",
        layer_norm_eps=1e-06,
        dropout=0.0,
        attention_dropout=0.0,
        initializer_range=1e-10,
        initializer_factor=1.0,
        qkv_bias=True,
        mlp_ratio=4.0,
        use_abs_pos=True,
        use_rel_pos=True,
        window_size=14,
        global_attn_indexes=[2, 5, 8, 11],
        num_pos_feats=128,
        mlp_dim=None,
        **kwargs,
    ):
        super().__init__(**kwargs)

        self.hidden_size = hidden_size
        self.intermediate_size = intermediate_size
        self.projection_dim = projection_dim
        self.output_channels = output_channels
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.num_channels = num_channels
        self.image_size = image_size
        self.patch_size = patch_size
        self.hidden_act = hidden_act
        self.layer_norm_eps = layer_norm_eps
        self.dropout = dropout
        self.attention_dropout = attention_dropout
        self.initializer_range = initializer_range
        self.initializer_factor = initializer_factor
        self.qkv_bias = qkv_bias
        self.mlp_ratio = mlp_ratio
        self.use_abs_pos = use_abs_pos
        self.use_rel_pos = use_rel_pos
        self.window_size = window_size
        self.global_attn_indexes = global_attn_indexes
        self.num_pos_feats = num_pos_feats
        self.mlp_dim = int(hidden_size * mlp_ratio) if mlp_dim is None else mlp_dim


class SamConfig(PretrainedConfig):
    r"""
    [`SamConfig`] is the configuration class to store the configuration of a [`SamModel`]. It is used to instantiate a
    SAM model according to the specified arguments, defining the vision model, prompt-encoder model and mask decoder
    configs. Instantiating a configuration with the defaults will yield a similar configuration to that of the
    SAM-ViT-H [facebook/sam-vit-huge](https://huggingface.co/facebook/sam-vit-huge) architecture.

    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
    documentation from [`PretrainedConfig`] for more information.

    Args:
        vision_config (Union[`dict`, `SamVisionConfig`], *optional*):
            Dictionary of configuration options used to initialize [`SamVisionConfig`].
        prompt_encoder_config (Union[`dict`, `SamPromptEncoderConfig`], *optional*):
            Dictionary of configuration options used to initialize [`SamPromptEncoderConfig`].
        mask_decoder_config (Union[`dict`, `SamMaskDecoderConfig`], *optional*):
            Dictionary of configuration options used to initialize [`SamMaskDecoderConfig`].

        kwargs (*optional*):
            Dictionary of keyword arguments.

    Example:

    ```python
    >>> from transformers import (
    ...     SamVisionConfig,
    ...     SamPromptEncoderConfig,
    ...     SamMaskDecoderConfig,
    ...     SamModel,
    ... )

    >>> # Initializing a SamConfig with `"facebook/sam-vit-huge"` style configuration
    >>> configuration = SamConfig()

    >>> # Initializing a SamModel (with random weights) from the `"facebook/sam-vit-huge"` style configuration
    >>> model = SamModel(configuration)

    >>> # Accessing the model configuration
    >>> configuration = model.config

    >>> # We can also initialize a SamConfig from a SamVisionConfig, SamPromptEncoderConfig, and SamMaskDecoderConfig

    >>> # Initializing SAM vision, SAM Q-Former and language model configurations
    >>> vision_config = SamVisionConfig()
    >>> prompt_encoder_config = SamPromptEncoderConfig()
    >>> mask_decoder_config = SamMaskDecoderConfig()

    >>> config = SamConfig(vision_config, prompt_encoder_config, mask_decoder_config)
    ```"""

    model_type = "sam"
    is_composition = True

    def __init__(
        self,
        vision_config=None,
        prompt_encoder_config=None,
        mask_decoder_config=None,
        initializer_range=0.02,
        **kwargs,
    ):
        super().__init__(**kwargs)
        vision_config = vision_config if vision_config is not None else {}
        prompt_encoder_config = prompt_encoder_config if prompt_encoder_config is not None else {}
        mask_decoder_config = mask_decoder_config if mask_decoder_config is not None else {}

        if isinstance(vision_config, SamVisionConfig):
            vision_config = vision_config.to_dict()
        if isinstance(prompt_encoder_config, SamPromptEncoderConfig):
            prompt_encoder_config = prompt_encoder_config.to_dict()
        if isinstance(mask_decoder_config, SamMaskDecoderConfig):
            mask_decoder_config = mask_decoder_config.to_dict()

        self.vision_config = SamVisionConfig(**vision_config)
        self.prompt_encoder_config = SamPromptEncoderConfig(**prompt_encoder_config)
        self.mask_decoder_config = SamMaskDecoderConfig(**mask_decoder_config)
        self.initializer_range = initializer_range

    def to_dict(self):
        """
        Serializes this instance to a Python dictionary. Override the default [`~PretrainedConfig.to_dict`].

        Returns:
            `Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance,
        """
        output = copy.deepcopy(self.__dict__)
        output["vision_config"] = self.vision_config.to_dict()
        output["prompt_encoder_config"] = self.prompt_encoder_config.to_dict()
        output["mask_decoder_config"] = self.mask_decoder_config.to_dict()
        output["model_type"] = self.__class__.model_type
        return output