LLM入门5 | SAM代码从入门到出门 | MetaAI

# 读取照片
image = cv2.imread('images/dog.jpg')
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# 展示照片
plt.figure(figsize=(20,20))
plt.imshow(image)
plt.axis('off')
plt.show()

import sys
sys.path.append("..")
from segment_anything import sam_model_registry, SamAutomaticMaskGenerator, SamPredictor

sam = sam_model_registry[model_type](checkpoint=sam_checkpoint)
sam.to(device=device)

mask_generator = SamAutomaticMaskGenerator(sam)

masks = mask_generator.generate(image)
print(len(masks))
print(masks[0].keys())

plt.figure(figsize=(20,20))
plt.imshow(image)
# 这个函数就是给每个mask随机颜色
show_anns(masks)
plt.axis('off')
plt.show() 

sam_model_registry = {
    "default": build_sam,
    "vit_h": build_sam,
    "vit_l": build_sam_vit_l,
    "vit_b": build_sam_vit_b,
}


def build_sam_vit_h(checkpoint=None):
    return _build_sam(
        encoder_embed_dim=1280,
        encoder_depth=32,
        encoder_num_heads=16,
        encoder_global_attn_indexes=[7, 15, 23, 31],
        checkpoint=checkpoint,
    )


build_sam = build_sam_vit_h


def build_sam_vit_l(checkpoint=None):
    return _build_sam(
        encoder_embed_dim=1024,
        encoder_depth=24,
        encoder_num_heads=16,
        encoder_global_attn_indexes=[5, 11, 17, 23],
        checkpoint=checkpoint,
    )


def build_sam_vit_b(checkpoint=None):
    return _build_sam(
        encoder_embed_dim=768,
        encoder_depth=12,
        encoder_num_heads=12,
        encoder_global_attn_indexes=[2, 5, 8, 11],
        checkpoint=checkpoint,
    )

def _build_sam(
    encoder_embed_dim,
    encoder_depth,
    encoder_num_heads,
    encoder_global_attn_indexes,
    checkpoint=None,
):
# prompt的embedding是256
    prompt_embed_dim = 256
    # 图片尺寸是1024x1024
    image_size = 1024
    # 把整张图片划分成16个patch
    vit_patch_size = 16
    # 所以每一个patch是64x64的大小
    image_embedding_size = image_size // vit_patch_size
    # 这里和论文说的一样，sam包含了三个模块，image-encoder，prompt-encoder和两个融合的mask-decoder
    """
        SAM predicts object masks from an image and input prompts.

        Arguments:
          image_encoder (ImageEncoderViT): The backbone used to encode the
            image into image embeddings that allow for efficient mask prediction.
          prompt_encoder (PromptEncoder): Encodes various types of input prompts.
          mask_decoder (MaskDecoder): Predicts masks from the image embeddings
            and encoded prompts.
          pixel_mean (list(float)): Mean values for normalizing pixels in the input image.
          pixel_std (list(float)): Std values for normalizing pixels in the input image.
        """
    sam = Sam(
        image_encoder=ImageEncoderViT(
            depth=encoder_depth,
            embed_dim=encoder_embed_dim,
            img_size=image_size,
            mlp_ratio=4,
            norm_layer=partial(torch.nn.LayerNorm, eps=1e-6),
            num_heads=encoder_num_heads,
            patch_size=vit_patch_size,
            qkv_bias=True,
            use_rel_pos=True,
            global_attn_indexes=encoder_global_attn_indexes,
            window_size=14,
            out_chans=prompt_embed_dim,
        ),
        prompt_encoder=PromptEncoder(
            embed_dim=prompt_embed_dim,
            image_embedding_size=(image_embedding_size, image_embedding_size),
            input_image_size=(image_size, image_size),
            mask_in_chans=16,
        ),
        mask_decoder=MaskDecoder(
            num_multimask_outputs=3,
            transformer=TwoWayTransformer(
                depth=2,
                embedding_dim=prompt_embed_dim,
                mlp_dim=2048,
                num_heads=8,
            ),
            transformer_dim=prompt_embed_dim,
            iou_head_depth=3,
            iou_head_hidden_dim=256,
        ),
        pixel_mean=[123.675, 116.28, 103.53],
        pixel_std=[58.395, 57.12, 57.375],
    )
    sam.eval()
    if checkpoint is not None:
        with open(checkpoint, "rb") as f:
            state_dict = torch.load(f)
        sam.load_state_dict(state_dict)
    return sam

class Sam(nn.Module):
    mask_threshold: float = 0.0
    image_format: str = "RGB"

    def __init__(
        self,
        image_encoder: ImageEncoderViT,
        prompt_encoder: PromptEncoder,
        mask_decoder: MaskDecoder,
        pixel_mean: List[float] = [123.675, 116.28, 103.53],
        pixel_std: List[float] = [58.395, 57.12, 57.375],
    ) -> None:
        
        super().__init__()
        self.image_encoder = image_encoder
        self.prompt_encoder = prompt_encoder
        self.mask_decoder = mask_decoder
        self.register_buffer("pixel_mean", torch.Tensor(pixel_mean).view(-1, 1, 1), False)
        self.register_buffer("pixel_std", torch.Tensor(pixel_std).view(-1, 1, 1), False)

    @property
    def device(self) -> Any:
        return self.pixel_mean.device

    @torch.no_grad()
    def forward(
        self,
        batched_input: List[Dict[str, Any]],
        multimask_output: bool,
    ) -> List[Dict[str, torch.Tensor]]:
        """
        Predicts masks end-to-end from provided images and prompts.
        If prompts are not known in advance, using SamPredictor is
        recommended over calling the model directly.

        Arguments:
          batched_input (list(dict)): A list over input images, each a
            dictionary with the following keys. A prompt key can be
            excluded if it is not present.
              'image': The image as a torch tensor in 3xHxW format,
                already transformed for input to the model.
              'original_size': (tuple(int, int)) The original size of
                the image before transformation, as (H, W).
              'point_coords': (torch.Tensor) Batched point prompts for
                this image, with shape BxNx2. Already transformed to the
                input frame of the model.
              'point_labels': (torch.Tensor) Batched labels for point prompts,
                with shape BxN.
              'boxes': (torch.Tensor) Batched box inputs, with shape Bx4.
                Already transformed to the input frame of the model.
              'mask_inputs': (torch.Tensor) Batched mask inputs to the model,
                in the form Bx1xHxW.
          multimask_output (bool): Whether the model should predict multiple
            disambiguating masks, or return a single mask.

        Returns:
          (list(dict)): A list over input images, where each element is
            as dictionary with the following keys.
              'masks': (torch.Tensor) Batched binary mask predictions,
                with shape BxCxHxW, where B is the number of input promts,
                C is determiend by multimask_output, and (H, W) is the
                original size of the image.
              'iou_predictions': (torch.Tensor) The model's predictions
                of mask quality, in shape BxC.
              'low_res_logits': (torch.Tensor) Low resolution logits with
                shape BxCxHxW, where H=W=256. Can be passed as mask input
                to subsequent iterations of prediction.
        """
        # 对图像做预处理，然后推叠成B，C，W，H的形状
        # 使用image_encoder对图像进行编码
        input_images = torch.stack([self.preprocess(x["image"]) for x in batched_input], dim=0)
        image_embeddings = self.image_encoder(input_images)

        outputs = []
        for image_record, curr_embedding in zip(batched_input, image_embeddings):
            if "point_coords" in image_record:
                points = (image_record["point_coords"], image_record["point_labels"])
            else:
                points = None
            sparse_embeddings, dense_embeddings = self.prompt_encoder(
                points=points,
                boxes=image_record.get("boxes", None),
                masks=image_record.get("mask_inputs", None),
            )
            low_res_masks, iou_predictions = self.mask_decoder(
                image_embeddings=curr_embedding.unsqueeze(0),
                image_pe=self.prompt_encoder.get_dense_pe(),
                sparse_prompt_embeddings=sparse_embeddings,
                dense_prompt_embeddings=dense_embeddings,
                multimask_output=multimask_output,
            )
            masks = self.postprocess_masks(
                low_res_masks,
                input_size=image_record["image"].shape[-2:],
                original_size=image_record["original_size"],
            )
            masks = masks > self.mask_threshold
            outputs.append(
                {
                    "masks": masks,
                    "iou_predictions": iou_predictions,
                    "low_res_logits": low_res_masks,
                }
            )
        return outputs

    def postprocess_masks(
        self,
        masks: torch.Tensor,
        input_size: Tuple[int, ...],
        original_size: Tuple[int, ...],
    ) -> torch.Tensor:
        """
        Remove padding and upscale masks to the original image size.

        Arguments:
          masks (torch.Tensor): Batched masks from the mask_decoder,
            in BxCxHxW format.
          input_size (tuple(int, int)): The size of the image input to the
            model, in (H, W) format. Used to remove padding.
          original_size (tuple(int, int)): The original size of the image
            before resizing for input to the model, in (H, W) format.

        Returns:
          (torch.Tensor): Batched masks in BxCxHxW format, where (H, W)
            is given by original_size.
        """
        masks = F.interpolate(
            masks,
            (self.image_encoder.img_size, self.image_encoder.img_size),
            mode="bilinear",
            align_corners=False,
        )
        masks = masks[..., : input_size[0], : input_size[1]]
        masks = F.interpolate(masks, original_size, mode="bilinear", align_corners=False)
        return masks

    def preprocess(self, x: torch.Tensor) -> torch.Tensor:
        """Normalize pixel values and pad to a square input."""
        # Normalize colors
        # 自然图像imagenet的归一化
        x = (x - self.pixel_mean) / self.pixel_std

        # Pad
        # 是否需要pad
        h, w = x.shape[-2:]
        padh = self.image_encoder.img_size - h
        padw = self.image_encoder.img_size - w
        x = F.pad(x, (0, padw, 0, padh))
        return x

import torch
import torch.nn as nn
import torch.nn.functional as F

from typing import Optional, Tuple, Type

from .common import LayerNorm2d, MLPBlock


# This class and its supporting functions below lightly adapted from the ViTDet backbone available at: https://github.com/facebookresearch/detectron2/blob/main/detectron2/modeling/backbone/vit.py # noqa
class ImageEncoderViT(nn.Module):
    def __init__(
        self,
        img_size: int = 1024,
        patch_size: int = 16,
        in_chans: int = 3,
        embed_dim: int = 768,
        depth: int = 12,
        num_heads: int = 12,
        mlp_ratio: float = 4.0,
        out_chans: int = 256,
        qkv_bias: bool = True,
        norm_layer: Type[nn.Module] = nn.LayerNorm,
        act_layer: Type[nn.Module] = nn.GELU,
        use_abs_pos: bool = True,
        use_rel_pos: bool = False,
        rel_pos_zero_init: bool = True,
        window_size: int = 0,
        global_attn_indexes: Tuple[int, ...] = (),
    ) -> None:
        """
        Args:
            img_size (int): Input image size.
            patch_size (int): Patch size.
            in_chans (int): Number of input image channels.
            embed_dim (int): Patch embedding dimension.
            depth (int): Depth of ViT.
            num_heads (int): Number of attention heads in each ViT block.
            mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
            qkv_bias (bool): If True, add a learnable bias to query, key, value.
            norm_layer (nn.Module): Normalization layer.
            act_layer (nn.Module): Activation layer.
            use_abs_pos (bool): If True, use absolute positional embeddings.
            use_rel_pos (bool): If True, add relative positional embeddings to the attention map.
            rel_pos_zero_init (bool): If True, zero initialize relative positional parameters.
            window_size (int): Window size for window attention blocks.
            global_attn_indexes (list): Indexes for blocks using global attention.
        """
        super().__init__()
        self.img_size = img_size
        # 这是一种划分patch的方法
        self.patch_embed = PatchEmbed(
            kernel_size=(patch_size, patch_size),
            stride=(patch_size, patch_size),
            in_chans=in_chans,
            embed_dim=embed_dim,
        )

        self.pos_embed: Optional[nn.Parameter] = None
        # 这个pos_embedding是可训练的？并且每一个embedding都有吗？
        if use_abs_pos:
            # Initialize absolute positional embedding with pretrain image size.
            self.pos_embed = nn.Parameter(
                torch.zeros(1, img_size // patch_size, img_size // patch_size, embed_dim)
            )

        self.blocks = nn.ModuleList()
        for i in range(depth):
            # 每一个block，里面包含了多头注意力，还有全连接层等等
            block = Block(
                dim=embed_dim,
                num_heads=num_heads,
                mlp_ratio=mlp_ratio,
                qkv_bias=qkv_bias,
                norm_layer=norm_layer,
                act_layer=act_layer,
                use_rel_pos=use_rel_pos,
                rel_pos_zero_init=rel_pos_zero_init,
                window_size=window_size if i not in global_attn_indexes else 0,
                input_size=(img_size // patch_size, img_size // patch_size),
            )
            self.blocks.append(block)
        # 就是输出层了，这里的out_channs就是为了和prompt_embedding维度对齐，变成256
        self.neck = nn.Sequential(
            nn.Conv2d(
                embed_dim,
                out_chans,
                kernel_size=1,
                bias=False,
            ),
            LayerNorm2d(out_chans),
            nn.Conv2d(
                out_chans,
                out_chans,
                kernel_size=3,
                padding=1,
                bias=False,
            ),
            LayerNorm2d(out_chans),
        )

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        x = self.patch_embed(x)
        if self.pos_embed is not None:
            x = x + self.pos_embed

        for blk in self.blocks:
            x = blk(x)

        x = self.neck(x.permute(0, 3, 1, 2))

        return x


class Block(nn.Module):
    """Transformer blocks with support of window attention and residual propagation blocks"""

    def __init__(
        self,
        dim: int,
        num_heads: int,
        mlp_ratio: float = 4.0,
        qkv_bias: bool = True,
        norm_layer: Type[nn.Module] = nn.LayerNorm,
        act_layer: Type[nn.Module] = nn.GELU,
        use_rel_pos: bool = False,
        rel_pos_zero_init: bool = True,
        window_size: int = 0,
        input_size: Optional[Tuple[int, int]] = None,
    ) -> None:
        """
        Args:
            dim (int): Number of input channels.
            num_heads (int): Number of attention heads in each ViT block.
            mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
            qkv_bias (bool): If True, add a learnable bias to query, key, value.
            norm_layer (nn.Module): Normalization layer.
            act_layer (nn.Module): Activation layer.
            use_rel_pos (bool): If True, add relative positional embeddings to the attention map.
            rel_pos_zero_init (bool): If True, zero initialize relative positional parameters.
            window_size (int): Window size for window attention blocks. If it equals 0, then
                use global attention.
            input_size (int or None): Input resolution for calculating the relative positional
                parameter size.
        """
        super().__init__()
        # 归一化曾选择layer norm
        self.norm1 = norm_layer(dim)
        # attentionmo模块
        self.attn = Attention(
            dim,
            num_heads=num_heads,
            qkv_bias=qkv_bias,
            use_rel_pos=use_rel_pos,
            rel_pos_zero_init=rel_pos_zero_init,
            input_size=input_size if window_size == 0 else (window_size, window_size),
        )
        # 第二个归一化层
        self.norm2 = norm_layer(dim)
        self.mlp = MLPBlock(embedding_dim=dim, mlp_dim=int(dim * mlp_ratio), act=act_layer)

        self.window_size = window_size

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        shortcut = x
        x = self.norm1(x)
        # Window partition
        # 这个比较关键，解释了之前的问题：window_size=window_size if i not in global_attn_indexes else 0
        # sam的window_size设置成了14
        # 当block的数量不在global_attn_indexes的时候，windows_size=14，否则就是0
        if self.window_size > 0:
            H, W = x.shape[1], x.shape[2]
            x, pad_hw = window_partition(x, self.window_size)
        # 所以这里的attention是在
        x = self.attn(x)
        # Reverse window partition
        if self.window_size > 0:
            x = window_unpartition(x, self.window_size, pad_hw, (H, W))
        # 两个残差
        x = shortcut + x
        x = x + self.mlp(self.norm2(x))

        return x


class Attention(nn.Module):
    """Multi-head Attention block with relative position embeddings."""

    def __init__(
        self,
        dim: int,
        num_heads: int = 8,
        qkv_bias: bool = True,
        use_rel_pos: bool = False,
        rel_pos_zero_init: bool = True,
        input_size: Optional[Tuple[int, int]] = None,
    ) -> None:
        """
        Args:
            dim (int): Number of input channels.
            num_heads (int): Number of attention heads.
            qkv_bias (bool:  If True, add a learnable bias to query, key, value.
            rel_pos (bool): If True, add relative positional embeddings to the attention map.
            rel_pos_zero_init (bool): If True, zero initialize relative positional parameters.
            input_size (int or None): Input resolution for calculating the relative positional
                parameter size.
        """
        super().__init__()
        self.num_heads = num_heads
        head_dim = dim // num_heads
        self.scale = head_dim**-0.5

        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
        self.proj = nn.Linear(dim, dim)

        self.use_rel_pos = use_rel_pos
        if self.use_rel_pos:
            assert (
                input_size is not None
            ), "Input size must be provided if using relative positional encoding."
            # initialize relative positional embeddings
            self.rel_pos_h = nn.Parameter(torch.zeros(2 * input_size[0] - 1, head_dim))
            self.rel_pos_w = nn.Parameter(torch.zeros(2 * input_size[1] - 1, head_dim))

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        B, H, W, _ = x.shape
        # qkv with shape (3, B, nHead, H * W, C)
        qkv = self.qkv(x).reshape(B, H * W, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)
        # q, k, v with shape (B * nHead, H * W, C)
        # 因为之前做了那个window partition，导致这里的H和W从原来的64x64变成了14x14,
        # 这里的14其实是对64做了一个采样得到的。
        q, k, v = qkv.reshape(3, B * self.num_heads, H * W, -1).unbind(0)
        attn = (q * self.scale) @ k.transpose(-2, -1)
        # 跳过
        if self.use_rel_pos:
            attn = add_decomposed_rel_pos(attn, q, self.rel_pos_h, self.rel_pos_w, (H, W), (H, W))

        attn = attn.softmax(dim=-1)
        x = (attn @ v).view(B, self.num_heads, H, W, -1).permute(0, 2, 3, 1, 4).reshape(B, H, W, -1)
        x = self.proj(x)

        return x


def window_partition(x: torch.Tensor, window_size: int) -> Tuple[torch.Tensor, Tuple[int, int]]:
    """
    Partition into non-overlapping windows with padding if needed.
    Args:
        x (tensor): input tokens with [B, H, W, C].
        window_size (int): window size.

    Returns:
        windows: windows after partition with [B * num_windows, window_size, window_size, C].
        (Hp, Wp): padded height and width before partition
    本来是Bx64x64的尺寸，经过变换后变成了Bx25，14，14，C的尺寸
    """
    B, H, W, C = x.shape
    # 这里的W和H表示64，given image size=1024 and patch_size=16
    # (14-64%14)%14=6?
    pad_h = (window_size - H % window_size) % window_size
    pad_w = (window_size - W % window_size) % window_size
    if pad_h > 0 or pad_w > 0:
        x = F.pad(x, (0, 0, 0, pad_w, 0, pad_h))
    # Hp和Wp都是70,
    Hp, Wp = H + pad_h, W + pad_w
    # Hp // window_size=5
    #B,5, 14, 5, 14, C
    x = x.view(B, Hp // window_size, window_size, Wp // window_size, window_size, C)
    #B,5, 5 14,, 14, C
    windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)
    return windows, (Hp, Wp)


def window_unpartition(
    windows: torch.Tensor, window_size: int, pad_hw: Tuple[int, int], hw: Tuple[int, int]
) -> torch.Tensor:
    """
    Window unpartition into original sequences and removing padding.
    Args:
        x (tensor): input tokens with [B * num_windows, window_size, window_size, C].
        window_size (int): window size.
        pad_hw (Tuple): padded height and width (Hp, Wp).
        hw (Tuple): original height and width (H, W) before padding.

    Returns:
        x: unpartitioned sequences with [B, H, W, C].
    """
    Hp, Wp = pad_hw
    H, W = hw
    B = windows.shape[0] // (Hp * Wp // window_size // window_size)
    x = windows.view(B, Hp // window_size, Wp // window_size, window_size, window_size, -1)
    x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, Hp, Wp, -1)

    if Hp > H or Wp > W:
        x = x[:, :H, :W, :].contiguous()
    return x
class PatchEmbed(nn.Module):
    """
    Image to Patch Embedding.
    """

    def __init__(
        self,
        kernel_size: Tuple[int, int] = (16, 16),
        stride: Tuple[int, int] = (16, 16),
        padding: Tuple[int, int] = (0, 0),
        in_chans: int = 3,
        embed_dim: int = 768,
    ) -> None:
        """
        Args:
            kernel_size (Tuple): kernel size of the projection layer.
            stride (Tuple): stride of the projection layer.
            padding (Tuple): padding size of the projection layer.
            in_chans (int): Number of input image channels.
            embed_dim (int):  embed_dim (int): Patch embedding dimension.
        """
        super().__init__()

        self.proj = nn.Conv2d(
            in_chans, embed_dim, kernel_size=kernel_size, stride=stride, padding=padding
        )

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        x = self.proj(x)
        # B C H W -> B H W C
        x = x.permute(0, 2, 3, 1)
        return x

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.

# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.

import numpy as np
import torch
from torch import nn

from typing import Any, Optional, Tuple, Type

from .common import LayerNorm2d


class PromptEncoder(nn.Module):
    def __init__(
        self,
        embed_dim: int,
        image_embedding_size: Tuple[int, int],
        input_image_size: Tuple[int, int],
        mask_in_chans: int,
        activation: Type[nn.Module] = nn.GELU,
    ) -> None:
        """
        Encodes prompts for input to SAM's mask decoder.

        Arguments:
          embed_dim (int): The prompts' embedding dimension
          image_embedding_size (tuple(int, int)): The spatial size of the
            image embedding, as (H, W).
          input_image_size (int): The padded size of the image as input
            to the image encoder, as (H, W).
          mask_in_chans (int): The number of hidden channels used for
            encoding input masks.
          activation (nn.Module): The activation to use when encoding
            input masks.
        """
        super().__init__()
        self.embed_dim = embed_dim
        self.input_image_size = input_image_size
        self.image_embedding_size = image_embedding_size
        self.pe_layer = PositionEmbeddingRandom(embed_dim // 2)

        self.num_point_embeddings: int = 4  # pos/neg point + 2 box corners
        point_embeddings = [nn.Embedding(1, embed_dim) for i in range(self.num_point_embeddings)]
        self.point_embeddings = nn.ModuleList(point_embeddings)
        self.not_a_point_embed = nn.Embedding(1, embed_dim)

        self.mask_input_size = (4 * image_embedding_size[0], 4 * image_embedding_size[1])
        self.mask_downscaling = nn.Sequential(
            nn.Conv2d(1, mask_in_chans // 4, kernel_size=2, stride=2),
            LayerNorm2d(mask_in_chans // 4),
            activation(),
            nn.Conv2d(mask_in_chans // 4, mask_in_chans, kernel_size=2, stride=2),
            LayerNorm2d(mask_in_chans),
            activation(),
            nn.Conv2d(mask_in_chans, embed_dim, kernel_size=1),
        )
        self.no_mask_embed = nn.Embedding(1, embed_dim)

    def get_dense_pe(self) -> torch.Tensor:
        """
        Returns the positional encoding used to encode point prompts,
        applied to a dense set of points the shape of the image encoding.

        Returns:
          torch.Tensor: Positional encoding with shape
            1x(embed_dim)x(embedding_h)x(embedding_w)
        """
        return self.pe_layer(self.image_embedding_size).unsqueeze(0)

    def _embed_points(
        self,
        points: torch.Tensor,
        labels: torch.Tensor,
        pad: bool,
    ) -> torch.Tensor:
        """Embeds point prompts."""
        
        points = points + 0.5  # Shift to center of pixel
        if pad:
            padding_point = torch.zeros((points.shape[0], 1, 2), device=points.device)
            padding_label = -torch.ones((labels.shape[0], 1), device=labels.device)
            points = torch.cat([points, padding_point], dim=1)
            labels = torch.cat([labels, padding_label], dim=1)
        point_embedding = self.pe_layer.forward_with_coords(points, self.input_image_size)
        point_embedding[labels == -1] = 0.0
        point_embedding[labels == -1] += self.not_a_point_embed.weight
        point_embedding[labels == 0] += self.point_embeddings[0].weight
        point_embedding[labels == 1] += self.point_embeddings[1].weight
        return point_embedding

    def _embed_boxes(self, boxes: torch.Tensor) -> torch.Tensor:
        """Embeds box prompts."""
        boxes = boxes + 0.5  # Shift to center of pixel
        coords = boxes.reshape(-1, 2, 2)
        corner_embedding = self.pe_layer.forward_with_coords(coords, self.input_image_size)
        corner_embedding[:, 0, :] += self.point_embeddings[2].weight
        corner_embedding[:, 1, :] += self.point_embeddings[3].weight
        return corner_embedding

    def _embed_masks(self, masks: torch.Tensor) -> torch.Tensor:
        """Embeds mask inputs."""
        mask_embedding = self.mask_downscaling(masks)
        return mask_embedding

    def _get_batch_size(
        self,
        points: Optional[Tuple[torch.Tensor, torch.Tensor]],
        boxes: Optional[torch.Tensor],
        masks: Optional[torch.Tensor],
    ) -> int:
        """
        Gets the batch size of the output given the batch size of the input prompts.
        """
        if points is not None:
            return points[0].shape[0]
        elif boxes is not None:
            return boxes.shape[0]
        elif masks is not None:
            return masks.shape[0]
        else:
            return 1

    def _get_device(self) -> torch.device:
        return self.point_embeddings[0].weight.device

    def forward(
        self,
        points: Optional[Tuple[torch.Tensor, torch.Tensor]],
        boxes: Optional[torch.Tensor],
        masks: Optional[torch.Tensor],
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        """
        Embeds different types of prompts, returning both sparse and dense
        embeddings.

        Arguments:
          points (tuple(torch.Tensor, torch.Tensor) or none): point coordinates
            and labels to embed.
          boxes (torch.Tensor or none): boxes to embed
          masks (torch.Tensor or none): masks to embed

        Returns:
          torch.Tensor: sparse embeddings for the points and boxes, with shape
            BxNx(embed_dim), where N is determined by the number of input points
            and boxes.
          torch.Tensor: dense embeddings for the masks, in the shape
            Bx(embed_dim)x(embed_H)x(embed_W)
        """
        bs = self._get_batch_size(points, boxes, masks)
        sparse_embeddings = torch.empty((bs, 0, self.embed_dim), device=self._get_device())
        if points is not None:
            coords, labels = points
            # coords就是坐标，labels就是正样本还是负样本
            point_embeddings = self._embed_points(coords, labels, pad=(boxes is None))
            sparse_embeddings = torch.cat([sparse_embeddings, point_embeddings], dim=1)
        if boxes is not None:
            box_embeddings = self._embed_boxes(boxes)
            sparse_embeddings = torch.cat([sparse_embeddings, box_embeddings], dim=1)

        if masks is not None:
            dense_embeddings = self._embed_masks(masks)
        else:
            dense_embeddings = self.no_mask_embed.weight.reshape(1, -1, 1, 1).expand(
                bs, -1, self.image_embedding_size[0], self.image_embedding_size[1]
            )

        return sparse_embeddings, dense_embeddings


class PositionEmbeddingRandom(nn.Module):
    """
    Positional encoding using random spatial frequencies.
    """

    def __init__(self, num_pos_feats: int = 64, scale: Optional[float] = None) -> None:
        super().__init__()
        if scale is None or scale <= 0.0:
            scale = 1.0
        self.register_buffer(
            "positional_encoding_gaussian_matrix",
            scale * torch.randn((2, num_pos_feats)),
        )

    def _pe_encoding(self, coords: torch.Tensor) -> torch.Tensor:
        """Positionally encode points that are normalized to [0,1]."""
        # assuming coords are in [0, 1]^2 square and have d_1 x ... x d_n x 2 shape
        coords = 2 * coords - 1
        coords = coords @ self.positional_encoding_gaussian_matrix
        coords = 2 * np.pi * coords
        # outputs d_1 x ... x d_n x C shape
        return torch.cat([torch.sin(coords), torch.cos(coords)], dim=-1)

    def forward(self, size: Tuple[int, int]) -> torch.Tensor:
        """Generate positional encoding for a grid of the specified size."""
        h, w = size
        device: Any = self.positional_encoding_gaussian_matrix.device
        grid = torch.ones((h, w), device=device, dtype=torch.float32)
        y_embed = grid.cumsum(dim=0) - 0.5
        x_embed = grid.cumsum(dim=1) - 0.5
        y_embed = y_embed / h
        x_embed = x_embed / w

        pe = self._pe_encoding(torch.stack([x_embed, y_embed], dim=-1))
        return pe.permute(2, 0, 1)  # C x H x W

    def forward_with_coords(
        self, coords_input: torch.Tensor, image_size: Tuple[int, int]
    ) -> torch.Tensor:
        """Positionally encode points that are not normalized to [0,1]."""
        coords = coords_input.clone()
        coords[:, :, 0] = coords[:, :, 0] / image_size[1]
        coords[:, :, 1] = coords[:, :, 1] / image_size[0]
        return self._pe_encoding(coords.to(torch.float))  # B x N x C

import torch
from torch import nn
from torch.nn import functional as F

from typing import List, Tuple, Type

from .common import LayerNorm2d


class MaskDecoder(nn.Module):
    def __init__(
        self,
        *,
        transformer_dim: int,
        transformer: nn.Module,
        num_multimask_outputs: int = 3,
        activation: Type[nn.Module] = nn.GELU,
        iou_head_depth: int = 3,
        iou_head_hidden_dim: int = 256,
    ) -> None:
        """
        Predicts masks given an image and prompt embeddings, using a
        tranformer architecture.

        Arguments:
          transformer_dim (int): the channel dimension of the transformer
          transformer (nn.Module): the transformer used to predict masks
          num_multimask_outputs (int): the number of masks to predict
            when disambiguating masks
          activation (nn.Module): the type of activation to use when
            upscaling masks
          iou_head_depth (int): the depth of the MLP used to predict
            mask quality
          iou_head_hidden_dim (int): the hidden dimension of the MLP
            used to predict mask quality
        """
        super().__init__()
        self.transformer_dim = transformer_dim
        self.transformer = transformer

        self.num_multimask_outputs = num_multimask_outputs

        self.iou_token = nn.Embedding(1, transformer_dim)
        self.num_mask_tokens = num_multimask_outputs + 1
        self.mask_tokens = nn.Embedding(self.num_mask_tokens, transformer_dim)

        self.output_upscaling = nn.Sequential(
            nn.ConvTranspose2d(transformer_dim, transformer_dim // 4, kernel_size=2, stride=2),
            LayerNorm2d(transformer_dim // 4),
            activation(),
            nn.ConvTranspose2d(transformer_dim // 4, transformer_dim // 8, kernel_size=2, stride=2),
            activation(),
        )
        self.output_hypernetworks_mlps = nn.ModuleList(
            [
                MLP(transformer_dim, transformer_dim, transformer_dim // 8, 3)
                for i in range(self.num_mask_tokens)
            ]
        )

        self.iou_prediction_head = MLP(
            transformer_dim, iou_head_hidden_dim, self.num_mask_tokens, iou_head_depth
        )

    def forward(
        self,
        image_embeddings: torch.Tensor,
        image_pe: torch.Tensor,
        sparse_prompt_embeddings: torch.Tensor,
        dense_prompt_embeddings: torch.Tensor,
        multimask_output: bool,
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        """
        Predict masks given image and prompt embeddings.

        Arguments:
          image_embeddings (torch.Tensor): the embeddings from the image encoder
          image_pe (torch.Tensor): positional encoding with the shape of image_embeddings
          sparse_prompt_embeddings (torch.Tensor): the embeddings of the points and boxes
          dense_prompt_embeddings (torch.Tensor): the embeddings of the mask inputs
          multimask_output (bool): Whether to return multiple masks or a single
            mask.

        Returns:
          torch.Tensor: batched predicted masks
          torch.Tensor: batched predictions of mask quality
        """
        masks, iou_pred = self.predict_masks(
            image_embeddings=image_embeddings,
            image_pe=image_pe,
            sparse_prompt_embeddings=sparse_prompt_embeddings,
            dense_prompt_embeddings=dense_prompt_embeddings,
        )

        # Select the correct mask or masks for outptu
        if multimask_output:
            mask_slice = slice(1, None)
        else:
            mask_slice = slice(0, 1)
        masks = masks[:, mask_slice, :, :]
        iou_pred = iou_pred[:, mask_slice]

        # Prepare output
        return masks, iou_pred

    def predict_masks(
        self,
        image_embeddings: torch.Tensor,
        image_pe: torch.Tensor,
        sparse_prompt_embeddings: torch.Tensor,
        dense_prompt_embeddings: torch.Tensor,
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        """Predicts masks. See 'forward' for more details."""
        # Concatenate output tokens
        output_tokens = torch.cat([self.iou_token.weight, self.mask_tokens.weight], dim=0)
        output_tokens = output_tokens.unsqueeze(0).expand(sparse_prompt_embeddings.size(0), -1, -1)
        tokens = torch.cat((output_tokens, sparse_prompt_embeddings), dim=1)

        # Expand per-image data in batch direction to be per-mask
        src = torch.repeat_interleave(image_embeddings, tokens.shape[0], dim=0)
        src = src + dense_prompt_embeddings
        pos_src = torch.repeat_interleave(image_pe, tokens.shape[0], dim=0)
        b, c, h, w = src.shape

        # Run the transformer
        hs, src = self.transformer(src, pos_src, tokens)
        iou_token_out = hs[:, 0, :]
        mask_tokens_out = hs[:, 1 : (1 + self.num_mask_tokens), :]

        # Upscale mask embeddings and predict masks using the mask tokens
        src = src.transpose(1, 2).view(b, c, h, w)
        upscaled_embedding = self.output_upscaling(src)
        hyper_in_list: List[torch.Tensor] = []
        for i in range(self.num_mask_tokens):
            hyper_in_list.append(self.output_hypernetworks_mlps[i](mask_tokens_out[:, i, :]))
        hyper_in = torch.stack(hyper_in_list, dim=1)
        b, c, h, w = upscaled_embedding.shape
        masks = (hyper_in @ upscaled_embedding.view(b, c, h * w)).view(b, -1, h, w)

        # Generate mask quality predictions
        iou_pred = self.iou_prediction_head(iou_token_out)

        return masks, iou_pred


# Lightly adapted from
# https://github.com/facebookresearch/MaskFormer/blob/main/mask_former/modeling/transformer/transformer_predictor.py # noqa
class MLP(nn.Module):
    def __init__(
        self,
        input_dim: int,
        hidden_dim: int,
        output_dim: int,
        num_layers: int,
        sigmoid_output: bool = False,
    ) -> None:
        super().__init__()
        self.num_layers = num_layers
        h = [hidden_dim] * (num_layers - 1)
        self.layers = nn.ModuleList(
            nn.Linear(n, k) for n, k in zip([input_dim] + h, h + [output_dim])
        )
        self.sigmoid_output = sigmoid_output

    def forward(self, x):
        for i, layer in enumerate(self.layers):
            x = F.relu(layer(x)) if i < self.num_layers - 1 else layer(x)
        if self.sigmoid_output:
            x = F.sigmoid(x)
        return x