VideoX-Fun/scripts/wan2.1/image_processing_qwen2_vl.py

# coding=utf-8
# Copyright 2024 The Qwen team, Alibaba Group and the HuggingFace Inc. team. All rights reserved.
#
# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
# and OPT implementations in this library. It has been modified from its
# original forms to accommodate minor architectural differences compared
# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
#
# 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.
"""Image processor class for Qwen2-VL."""

import math
from typing import Optional, Union

import numpy as np

from ...image_processing_utils import BaseImageProcessor, BatchFeature
from ...image_transforms import (
    convert_to_rgb,
    resize,
)
from ...image_utils import (
    OPENAI_CLIP_MEAN,
    OPENAI_CLIP_STD,
    ChannelDimension,
    ImageInput,
    PILImageResampling,
    get_image_size,
    infer_channel_dimension_format,
    is_scaled_image,
    make_list_of_images,
    to_numpy_array,
    valid_images,
    validate_preprocess_arguments,
)
from ...utils import TensorType, logging
# image_processing_qwen2_vl.py 文件的开头

# ... (文件原有的import语句，例如 import json, os, etc.)

# =========================================================================================
# START: Manual patch for older transformers versions
# =========================================================================================
import collections
from typing import List, Union, TypeVar

try:
    import numpy as np
except ImportError:
    np = None

try:
    import torch
except ImportError:
    torch = None

try:
    from PIL import Image
except ImportError:
    Image = None

VideoInput = Union[List[ImageInput], List[List[ImageInput]]]
T = TypeVar("T")

def to_channel_dimension_format(
    image: Union[np.ndarray, torch.Tensor],  # 修改: 接受 torch.Tensor
    channel_dim: Union[str, ChannelDimension],
    input_channel_dim: Optional[Union[str, ChannelDimension]] = None,
) -> Union[np.ndarray, torch.Tensor]: # 修改: 返回相应类型
    """
    Converts `image` to the channel dimension format specified by `channel_dim`.
    This version is modified to handle both NumPy arrays and PyTorch Tensors to preserve gradients.
    """
    # 检查输入类型，并选择相应的处理方式
    is_torch_tensor = isinstance(image, torch.Tensor)

    if not isinstance(image, (np.ndarray, torch.Tensor)):
        raise TypeError(f"Input image must be of type np.ndarray or torch.Tensor, got {type(image)}")

    if input_channel_dim is None:
        input_channel_dim = infer_channel_dimension_format(image)

    target_channel_dim = ChannelDimension(channel_dim)
    if input_channel_dim == target_channel_dim:
        return image

    # 根据目标格式进行维度重排
    if target_channel_dim == ChannelDimension.FIRST:
        # 目标: (C, H, W)
        if is_torch_tensor:
            image = image.permute(2, 0, 1)  # PyTorch 张量处理，保留梯度
        else:
            image = image.transpose((2, 0, 1))  # NumPy 数组处理
    elif target_channel_dim == ChannelDimension.LAST:
        # 目标: (H, W, C)
        if is_torch_tensor:
            image = image.permute(1, 2, 0)  # PyTorch 张量处理，保留梯度
        else:
            image = image.transpose((1, 2, 0))  # NumPy 数组处理
    else:
        raise ValueError(f"Unsupported channel dimension format: {channel_dim}")

    return image

def make_batched_videos(videos: Union[T, List[T]]) -> List[T]:
    """Ensure that the input videos are in a batched list format."""
    if not isinstance(videos, list) or (videos and not isinstance(videos[0], collections.abc.Sized)):
        return [videos]
    return videos

# 为了让代码能找到 make_flat_list_of_images，也把它加进来
def make_flat_list_of_images(images: Union[T, List[T]]) -> List[T]:
    """Ensure that the input images are in a flat list format."""
    if not isinstance(images, list):
        return [images]
    if not images or isinstance(images[0], (list, tuple)):
        return [item for sublist in images for item in sublist]
    return images
# =========================================================================================
# END: Manual patch
# =========================================================================================


# ... (文件原有的其他代码，例如 class Qwen2VLImageProcessor(...))
# ...

logger = logging.get_logger(__name__)


def smart_resize(
    height: int, width: int, factor: int = 28, min_pixels: int = 56 * 56, max_pixels: int = 14 * 14 * 4 * 1280
):
    """Rescales the image so that the following conditions are met:

    1. Both dimensions (height and width) are divisible by 'factor'.

    2. The total number of pixels is within the range ['min_pixels', 'max_pixels'].

    3. The aspect ratio of the image is maintained as closely as possible.

    """
    if max(height, width) / min(height, width) > 200:
        raise ValueError(
            f"absolute aspect ratio must be smaller than 200, got {max(height, width) / min(height, width)}"
        )
    h_bar = round(height / factor) * factor
    w_bar = round(width / factor) * factor
    if h_bar * w_bar > max_pixels:
        beta = math.sqrt((height * width) / max_pixels)
        h_bar = max(factor, math.floor(height / beta / factor) * factor)
        w_bar = max(factor, math.floor(width / beta / factor) * factor)
    elif h_bar * w_bar < min_pixels:
        beta = math.sqrt(min_pixels / (height * width))
        h_bar = math.ceil(height * beta / factor) * factor
        w_bar = math.ceil(width * beta / factor) * factor
    return h_bar, w_bar


# new_qwen2_vl_image_processor.py

class Qwen2VLImageProcessor(BaseImageProcessor):
    r"""
    Constructs a Qwen2-VL image processor that dynamically resizes images based on the original images.

    Args:
        do_resize (`bool`, *optional*, defaults to `True`):
            Whether to resize the image's (height, width) dimensions.
        size (`dict[str, int]`, *optional*, defaults to `{"shortest_edge": 56 * 56, "longest_edge": 28 * 28 * 1280}`):
            Size of the image after resizing. `shortest_edge` and `longest_edge` keys must be present.
        resample (`PILImageResampling`, *optional*, defaults to `Resampling.BICUBIC`):
            Resampling filter to use when resizing the image.
        do_rescale (`bool`, *optional*, defaults to `True`):
            Whether to rescale the image by the specified scale `rescale_factor`.
        rescale_factor (`int` or `float`, *optional*, defaults to `1/255`):
            Scale factor to use if rescaling the image.
        do_normalize (`bool`, *optional*, defaults to `True`):
            Whether to normalize the image.
        image_mean (`float` or `list[float]`, *optional*, defaults to `[0.48145466, 0.4578275, 0.40821073]`):
            Mean to use if normalizing the image. This is a float or list of floats for each channel in the image.
        image_std (`float` or `list[float]`, *optional*, defaults to `[0.26862954, 0.26130258, 0.27577711]`):
            Standard deviation to use if normalizing the image. This is a float or list of floats for each channel in the image.
        do_convert_rgb (`bool`, *optional*, defaults to `True`):
            Whether to convert the image to RGB.
        min_pixels (`int`, *optional*, defaults to `56 * 56`):
            The min pixels of the image to resize the image.
        max_pixels (`int`, *optional*, defaults to `28 * 28 * 1280`):
            The max pixels of the image to resize the image.
        patch_size (`int`, *optional*, defaults to 14):
            The spatial patch size of the vision encoder.
        temporal_patch_size (`int`, *optional*, defaults to 2):
            The temporal patch size of the vision encoder.
        merge_size (`int`, *optional*, defaults to 2):
            The merge size of the vision encoder to llm encoder.
    """

    model_input_names = ["pixel_values", "image_grid_thw", "pixel_values_videos", "video_grid_thw"]

    def __init__(
        self,
        do_resize: bool = True,
        size: Optional[dict[str, int]] = None,
        resample: PILImageResampling = PILImageResampling.BICUBIC,
        do_rescale: bool = True,
        rescale_factor: Union[int, float] = 1 / 255,
        do_normalize: bool = True,
        image_mean: Optional[Union[float, list[float]]] = None,
        image_std: Optional[Union[float, list[float]]] = None,
        do_convert_rgb: bool = True,
        min_pixels: Optional[int] = None,
        max_pixels: Optional[int] = None,
        patch_size: int = 14,
        temporal_patch_size: int = 2,
        merge_size: int = 2,
        **kwargs,
    ) -> None:
        super().__init__(**kwargs)
        if size is not None and ("shortest_edge" not in size or "longest_edge" not in size):
            raise ValueError("size must contain 'shortest_edge' and 'longest_edge' keys.")
        else:
            size = {"shortest_edge": 56 * 56, "longest_edge": 28 * 28 * 1280}
        # backward compatibility: override size with min_pixels and max_pixels if they are provided
        if min_pixels is not None:
            size["shortest_edge"] = min_pixels
        if max_pixels is not None:
            size["longest_edge"] = max_pixels
        self.min_pixels = size["shortest_edge"]
        self.max_pixels = size["longest_edge"]
        self.size = size

        self.do_resize = do_resize
        self.resample = resample
        self.do_rescale = do_rescale
        self.rescale_factor = rescale_factor
        self.do_normalize = do_normalize
        self.image_mean = image_mean if image_mean is not None else OPENAI_CLIP_MEAN
        self.image_std = image_std if image_std is not None else OPENAI_CLIP_STD

        self.patch_size = patch_size
        self.temporal_patch_size = temporal_patch_size
        self.merge_size = merge_size
        self.do_convert_rgb = do_convert_rgb

    def _preprocess(
        self,
        images: Union[ImageInput, VideoInput],
        do_resize: Optional[bool] = None,
        size: Optional[dict[str, int]] = None,
        resample: PILImageResampling = None,
        do_rescale: Optional[bool] = None,
        rescale_factor: Optional[float] = None,
        do_normalize: Optional[bool] = None,
        image_mean: Optional[Union[float, list[float]]] = None,
        image_std: Optional[Union[float, list[float]]] = None,
        patch_size: Optional[int] = None,
        temporal_patch_size: Optional[int] = None,
        merge_size: Optional[int] = None,
        do_convert_rgb: Optional[bool] = None,
        data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
        input_data_format: Optional[Union[str, ChannelDimension]] = None,
    ):
        images = make_list_of_images(images)
        do_rescale = False
        do_resize = False
        do_normalize = False

        # <--- MODIFIED: Check if input is a torch tensor to decide the processing path
        is_torch_tensor = isinstance(images[0], torch.Tensor)
        
        if do_convert_rgb and not is_torch_tensor:
            # For non-tensor inputs, convert to RGB using PIL/Numpy logic
            images = [convert_to_rgb(image) for image in images]

        # <--- MODIFIED: Avoid converting to numpy if the input is already a tensor
        if not is_torch_tensor:
            # All transformations expect numpy arrays for the original path.
            images = [to_numpy_array(image) for image in images]
        
        # if do_rescale and is_scaled_image(images[0]):
        #     logger.warning_once(
        #         "It looks like you are trying to rescale already rescaled images. If the input"
        #         " images have pixel values between 0 and 1, set `do_rescale=False` to avoid rescaling them again."
        #     )
        if input_data_format is None:
            input_data_format = infer_channel_dimension_format(images[0])

        height, width = get_image_size(images[0], channel_dim=input_data_format)
        resized_height, resized_width = height, width
        processed_images = []
        for image in images:
            if do_resize:
                resized_height, resized_width = smart_resize(
                    height,
                    width,
                    factor=patch_size * merge_size,
                    min_pixels=size["shortest_edge"],
                    max_pixels=size["longest_edge"],
                )
                image = resize(
                    image, size=(resized_height, resized_width), resample=resample, input_data_format=input_data_format
                )

            if do_rescale:
                image = self.rescale(image, scale=rescale_factor, input_data_format=input_data_format)

            if do_normalize:
                image = self.normalize(
                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
                )

            image = to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
            processed_images.append(image)

        # <--- MODIFIED: Handle tensor and numpy patching separately
        if is_torch_tensor:
            patches = torch.stack(processed_images)
            if data_format == ChannelDimension.LAST:
                patches = patches.permute(0, 3, 1, 2)
            
            if patches.shape[0] % temporal_patch_size != 0:
                num_repeats = temporal_patch_size - (patches.shape[0] % temporal_patch_size)
                repeats = patches[-1].unsqueeze(0).repeat(num_repeats, 1, 1, 1)
                patches = torch.cat([patches, repeats], dim=0)

            channel = patches.shape[1]
            grid_t = patches.shape[0] // temporal_patch_size
            grid_h, grid_w = resized_height // patch_size, resized_width // patch_size
            
            patches = patches.view(
                grid_t,
                temporal_patch_size,
                channel,
                grid_h // merge_size,
                merge_size,
                patch_size,
                grid_w // merge_size,
                merge_size,
                patch_size,
            )
            patches = patches.permute(0, 3, 6, 4, 7, 2, 1, 5, 8)
            flatten_patches = patches.reshape(
                grid_t * grid_h * grid_w, channel * temporal_patch_size * patch_size * patch_size
            )
        else:
            # Original numpy-based logic
            patches = np.array(processed_images)
            if data_format == ChannelDimension.LAST:
                patches = patches.transpose(0, 3, 1, 2)
            if patches.shape[0] % temporal_patch_size != 0:
                repeats = np.repeat(
                    patches[-1][np.newaxis], temporal_patch_size - (patches.shape[0] % temporal_patch_size), axis=0
                )
                patches = np.concatenate([patches, repeats], axis=0)
            
            channel = patches.shape[1]
            grid_t = patches.shape[0] // temporal_patch_size
            grid_h, grid_w = resized_height // patch_size, resized_width // patch_size
            
            patches = patches.reshape(
                grid_t,
                temporal_patch_size,
                channel,
                grid_h // merge_size,
                merge_size,
                patch_size,
                grid_w // merge_size,
                merge_size,
                patch_size,
            )
            patches = patches.transpose(0, 3, 6, 4, 7, 2, 1, 5, 8)
            flatten_patches = patches.reshape(
                grid_t * grid_h * grid_w, channel * temporal_patch_size * patch_size * patch_size
            )

        return flatten_patches, (grid_t, grid_h, grid_w)


    def preprocess(
        self,
        images: ImageInput,
        videos: VideoInput = None,
        do_resize: Optional[bool] = None,
        size: Optional[dict[str, int]] = None,
        min_pixels: Optional[int] = None,
        max_pixels: Optional[int] = None,
        resample: PILImageResampling = None,
        do_rescale: Optional[bool] = None,
        rescale_factor: Optional[float] = None,
        do_normalize: Optional[bool] = None,
        image_mean: Optional[Union[float, list[float]]] = None,
        image_std: Optional[Union[float, list[float]]] = None,
        patch_size: Optional[int] = None,
        temporal_patch_size: Optional[int] = None,
        merge_size: Optional[int] = None,
        do_convert_rgb: Optional[bool] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
        input_data_format: Optional[Union[str, ChannelDimension]] = None,
    ):
        min_pixels = min_pixels if min_pixels is not None else self.min_pixels
        max_pixels = max_pixels if max_pixels is not None else self.max_pixels

        if size is not None:
            if "shortest_edge" not in size or "longest_edge" not in size:
                raise ValueError("size must contain 'shortest_edge' and 'longest_edge' keys.")
            min_pixels = size["shortest_edge"]
        elif min_pixels is not None and max_pixels is not None:
            size = {"shortest_edge": min_pixels, "longest_edge": max_pixels}
        else:
            size = {**self.size}

        do_resize = do_resize if do_resize is not None else self.do_resize
        resample = resample if resample is not None else self.resample
        do_rescale = do_rescale if do_rescale is not None else self.do_rescale
        rescale_factor = rescale_factor if rescale_factor is not None else self.rescale_factor
        do_normalize = do_normalize if do_normalize is not None else self.do_normalize
        image_mean = image_mean if image_mean is not None else self.image_mean
        image_std = image_std if image_std is not None else self.image_std
        patch_size = patch_size if patch_size is not None else self.patch_size
        temporal_patch_size = temporal_patch_size if temporal_patch_size is not None else self.temporal_patch_size
        merge_size = merge_size if merge_size is not None else self.merge_size
        do_convert_rgb = do_convert_rgb if do_convert_rgb is not None else self.do_convert_rgb

        if images is not None:
            images = make_list_of_images(images)

        if images is not None and not valid_images(images):
            raise ValueError(
                "Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
                "torch.Tensor, tf.Tensor or jax.ndarray."
            )

        validate_preprocess_arguments(
            rescale_factor=rescale_factor,
            do_normalize=do_normalize,
            image_mean=image_mean,
            image_std=image_std,
            do_resize=do_resize,
            size=size,
            resample=resample,
        )

        data = {}
        if images is not None:
            pixel_values, vision_grid_thws = [], []
            for image in images:
                patches, image_grid_thw = self._preprocess(
                    image,
                    do_resize=do_resize,
                    size=size,
                    resample=resample,
                    do_rescale=do_rescale,
                    rescale_factor=rescale_factor,
                    do_normalize=do_normalize,
                    image_mean=image_mean,
                    image_std=image_std,
                    patch_size=patch_size,
                    temporal_patch_size=temporal_patch_size,
                    merge_size=merge_size,
                    data_format=data_format,
                    do_convert_rgb=do_convert_rgb,
                    input_data_format=input_data_format,
                )
                # <--- MODIFIED: Append the raw tensor/array instead of extending
                pixel_values.append(patches)
                vision_grid_thws.append(image_grid_thw)
            
            # <--- MODIFIED: Conditionally concatenate tensors or stack numpy arrays
            is_processed_tensor = isinstance(pixel_values[0], torch.Tensor)
            if is_processed_tensor:
                pixel_values = torch.cat(pixel_values, dim=0)
                vision_grid_thws = torch.tensor(vision_grid_thws, dtype=torch.long)
            else:
                # The original `extend` followed by `np.array` is equivalent to concatenating
                # along the first axis if each patch set is 2D.
                all_patches = []
                for p in pixel_values:
                    all_patches.extend(p)
                pixel_values = np.array(all_patches)
                vision_grid_thws = np.array(vision_grid_thws)

            data.update({"pixel_values": pixel_values, "image_grid_thw": vision_grid_thws})
        
        # This part for videos remains unchanged as it's deprecated.
        if videos is not None:
            logger.warning(
                "`Qwen2VLImageProcessor` works only with image inputs and doesn't process videos anymore. "
                "This is a deprecated behavior and will be removed in v5.0. "
                "Your videos should be forwarded to `Qwen2VLVideoProcessor`. "
            )
            videos = make_batched_videos(videos)
            # pixel_values_videos_list = []
            # vision_grid_thws_videos_list = []
            pixel_values_videos, vision_grid_thws_videos = [], []
            for video_frames in videos:
                patches, video_grid_thw = self._preprocess(
                    video_frames,
                    do_resize=do_resize,
                    size=size,
                    resample=resample,
                    do_rescale=do_rescale,
                    rescale_factor=rescale_factor,
                    do_normalize=do_normalize,
                    image_mean=image_mean,
                    image_std=image_std,
                    patch_size=patch_size,
                    temporal_patch_size=temporal_patch_size,
                    merge_size=merge_size,
                    data_format=data_format,
                    do_convert_rgb=do_convert_rgb,
                    input_data_format=input_data_format,
                )

                pixel_values_videos.append(patches)
                vision_grid_thws_videos.append(torch.tensor(video_grid_thw))

            if pixel_values_videos:
                final_pixel_values_videos = torch.cat(pixel_values_videos, dim=0)
                final_vision_grid_thws_videos = torch.stack(vision_grid_thws_videos, dim=0)
            else:
                final_pixel_values_videos = torch.empty(0)
                final_vision_grid_thws_videos = torch.empty(0)

            data.update({"pixel_values_videos": final_pixel_values_videos,"video_grid_thw": final_vision_grid_thws_videos,})
            # import pdb
            # pdb.set_trace()

        return data
    
    # get_number_of_image_patches method does not need modification as it only performs calculations.
    def get_number_of_image_patches(self, height: int, width: int, images_kwargs=None):
        if images_kwargs is None:
            images_kwargs = {}
            
        min_pixels = images_kwargs.get("min_pixels", None) or self.size["shortest_edge"]
        max_pixels = images_kwargs.get("max_pixels", None) or self.size["longest_edge"]
        patch_size = images_kwargs.get("patch_size", None) or self.patch_size
        merge_size = images_kwargs.get("merge_size", None) or self.merge_size

        factor = patch_size * merge_size
        resized_height, resized_width = smart_resize(
            height, width, factor, min_pixels=min_pixels, max_pixels=max_pixels
        )
        grid_h, grid_w = resized_height // patch_size, resized_width // patch_size
        return grid_h * grid_w


__all__ = ["Qwen2VLImageProcessor"]