image_processing_qwen2_vl_fast.py

from typing import Optional, Union

import torch
from torchvision.transforms.v2 import functional as F

from transformers.image_processing_base import BatchFeature
from transformers.image_processing_utils_fast import (
    BaseImageProcessorFast,
    DefaultFastImageProcessorKwargs,
)
from transformers.image_transforms import group_images_by_shape, reorder_images
from transformers.utils.constants import OPENAI_CLIP_MEAN, OPENAI_CLIP_STD
from transformers.image_utils import (
    ChannelDimension,
    ImageInput,
    PILImageResampling,
    SizeDict,
)
from transformers.processing_utils import Unpack
from transformers.utils.generic import TensorType
from transformers.utils.auto_docstring import auto_docstring
from transformers.utils import logging
from transformers.video_utils import VideoInput, make_batched_videos
from .image_processing_qwen2_vl import smart_resize


logger = logging.get_logger(__name__)


class Qwen2VLFastImageProcessorKwargs(DefaultFastImageProcessorKwargs):
    """
    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.
    focus_size (`int`, *optional*, defaults to 2):
        The focus size of the ZoomFocus.
    """

    min_pixels: Optional[int]
    max_pixels: Optional[int]
    patch_size: Optional[int]
    temporal_patch_size: Optional[int]
    merge_size: Optional[int]
    focus_size: Optional[int]


@auto_docstring
class ZFQwen2VLImageProcessorFast(BaseImageProcessorFast):
    do_resize = True
    resample = PILImageResampling.BICUBIC
    size = {"shortest_edge": 56 * 56, "longest_edge": 28 * 28 * 1280}
    do_rescale = True
    do_normalize = True
    image_mean = OPENAI_CLIP_MEAN
    image_std = OPENAI_CLIP_STD
    do_convert_rgb = True
    patch_size = 14
    temporal_patch_size = 2
    merge_size = 2
    focus_size = 2
    min_pixels = None
    max_pixels = None
    valid_kwargs = Qwen2VLFastImageProcessorKwargs
    model_input_names = ["pixel_values", "image_grid_thw", "pixel_values_videos", "video_grid_thw"]

    def __init__(self, **kwargs: Unpack[Qwen2VLFastImageProcessorKwargs]):
        size = kwargs.pop("size", None)
        min_pixels = kwargs.pop("min_pixels", None)
        max_pixels = kwargs.pop("max_pixels", None)
        # backward compatibility: override size with min_pixels and max_pixels if they are provided
        size = self.size if size is None else size
        if min_pixels is not None:
            size["shortest_edge"] = min_pixels # type: ignore
            size.pop("min_pixels", None)
        if max_pixels is not None:
            size["longest_edge"] = max_pixels # type: ignore
            size.pop("max_pixels", None)
        if "shortest_edge" not in size or "longest_edge" not in size:
            raise ValueError("size must contain 'shortest_edge' and 'longest_edge' keys.")

        super().__init__(size=size, min_pixels=min_pixels, max_pixels=max_pixels, **kwargs) # type: ignore

    def _further_process_kwargs( # type: ignore
        self,
        size: Optional[SizeDict] = None,
        min_pixels: Optional[int] = None,
        max_pixels: Optional[int] = None,
        **kwargs,
    ) -> dict:
        """
        Update kwargs that need further processing before being validated
        Can be overridden by subclasses to customize the processing of kwargs.
        """
        if min_pixels is not None and max_pixels is not None:
            size = {"shortest_edge": min_pixels, "longest_edge": max_pixels} # type: ignore
        elif 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"]
            max_pixels = size["longest_edge"]
        else:
            size = {**self.size} # type: ignore

        return super()._further_process_kwargs(size=size, min_pixels=min_pixels, max_pixels=max_pixels, **kwargs)

    @auto_docstring
    def preprocess( # type: ignore
        self,
        images: ImageInput,
        videos: Optional[VideoInput] = None,
        **kwargs: Unpack[Qwen2VLFastImageProcessorKwargs],
    ) -> BatchFeature:
        return super().preprocess(images, videos, **kwargs)

    def _preprocess_image_like_inputs( # type: ignore
        self,
        images: ImageInput,
        videos: VideoInput,
        do_convert_rgb: bool,
        input_data_format: ChannelDimension,
        device: Optional[Union[str, "torch.device"]] = None,
        **kwargs: Unpack[DefaultFastImageProcessorKwargs], # type: ignore
    ) -> BatchFeature:
        """
        Preprocess image-like inputs.
        To be overridden by subclasses when image-like inputs other than images should be processed.
        It can be used for segmentation maps, depth maps, etc.
        """
        # Prepare input images
        batch_feature = BatchFeature()
        if images is not None:
            images = self._prepare_image_like_inputs(
                images=images, do_convert_rgb=do_convert_rgb, input_data_format=input_data_format, device=device # type: ignore
            )
            batch_feature = self._preprocess(images, **kwargs) # type: ignore
        if videos is not None:
            logger.warning(
                "`Qwen2VLImageProcessorFast` 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`. "
            )
            # Can't change _prepare_images_structure to work with videos because it also needs to work with images.
            videos = make_batched_videos(videos) # type: ignore
            videos = [
                torch.stack(self._prepare_image_like_inputs(video, do_convert_rgb, input_data_format, device)) # type: ignore
                for video in videos
            ]
            video_outputs = self._preprocess(videos, **kwargs) # type: ignore
            batch_feature.update(
                {"pixel_values_videos": video_outputs.pixel_values, "video_grid_thw": video_outputs.image_grid_thw}
            )
        return batch_feature

    def _preprocess( # type: ignore
        self,
        images: list["torch.Tensor"],
        do_resize: bool,
        size: SizeDict,
        interpolation: Optional["F.InterpolationMode"],
        do_rescale: bool,
        rescale_factor: float,
        do_normalize: bool,
        image_mean: Optional[Union[float, list[float]]],
        image_std: Optional[Union[float, list[float]]],
        patch_size: int,
        temporal_patch_size: int,
        merge_size: int,
        focus_size: int,
        disable_grouping: Optional[bool],
        return_tensors: Optional[Union[str, TensorType]],
        **kwargs,
    ):
        # Group images by size for batched resizing
        grouped_images, grouped_images_index = group_images_by_shape(images, disable_grouping=disable_grouping) # type: ignore
        resized_images_grouped = {}
        for shape, stacked_images in grouped_images.items():
            height, width = stacked_images.shape[-2:] # type: ignore
            if do_resize:
                resized_height, resized_width = smart_resize(
                    height,
                    width,
                    factor=patch_size * merge_size * focus_size,
                    min_pixels=size["shortest_edge"],
                    max_pixels=size["longest_edge"],
                )
                stacked_images = self.resize(
                    image=stacked_images, # type: ignore
                    size=SizeDict(height=resized_height, width=resized_width),
                    interpolation=interpolation,
                )
            resized_images_grouped[shape] = stacked_images
        resized_images = reorder_images(resized_images_grouped, grouped_images_index)

        # Group images by size for further processing
        # Needed in case do_resize is False, or resize returns images with different sizes
        grouped_images, grouped_images_index = group_images_by_shape(resized_images, disable_grouping=disable_grouping) # type: ignore
        processed_images_grouped = {}
        processed_grids = {}
        for shape, stacked_images in grouped_images.items():
            resized_height, resized_width = stacked_images.shape[-2:] # type: ignore
            # Fused rescale and normalize
            patches = self.rescale_and_normalize(
                stacked_images, do_rescale, rescale_factor, do_normalize, image_mean, image_std # type: ignore
            )
            if patches.ndim == 4:
                # add a temporal dimension if we have images
                patches = patches.unsqueeze(1)
            if patches.shape[1] % temporal_patch_size != 0:
                repeats = patches[:, -1:].repeat(1, temporal_patch_size - 1, 1, 1, 1)
                patches = torch.cat([patches, repeats], dim=1)
            batch_size, grid_t, channel = patches.shape[:3]
            grid_t = grid_t // temporal_patch_size
            grid_h, grid_w = resized_height // patch_size, resized_width // patch_size

            patches = patches.view(
                batch_size,
                grid_t,
                temporal_patch_size,
                channel,
                grid_h // merge_size,
                merge_size,
                patch_size,
                grid_w // merge_size,
                merge_size,
                patch_size,
            )
            # Reorder dimensions to group grid and patch information for subsequent flattening.
            # (batch, grid_t, grid_h, grid_w, merge_h, merge_w, channel, temp_patch_size, patch_h, patch_w)
            patches = patches.permute(0, 1, 4, 7, 5, 8, 3, 2, 6, 9)
            flatten_patches = patches.reshape(
                batch_size,
                grid_t * grid_h * grid_w,
                channel * temporal_patch_size * patch_size * patch_size,
            )

            processed_images_grouped[shape] = flatten_patches
            processed_grids[shape] = [[grid_t, grid_h, grid_w]] * batch_size

        processed_images = reorder_images(processed_images_grouped, grouped_images_index)
        processed_grids = reorder_images(processed_grids, grouped_images_index)
        pixel_values = torch.cat(processed_images, dim=0) # type: ignore
        image_grid_thw = torch.tensor(processed_grids)

        return BatchFeature(
            data={"pixel_values": pixel_values, "image_grid_thw": image_grid_thw}, tensor_type=return_tensors
        )

    def get_number_of_image_patches(self, height: int, width: int, images_kwargs=None):
        """
        A utility that returns number of image patches for a given image size.

        Note: Do not remove this method! It is used by vLLM to infer the number of patches and placeholders
        without an image input.

        Args:
            height (`int`):
                Height of the input image.
            width (`int`):
                Width of the input image.
            images_kwargs (`dict`, *optional*)
                Any kwargs to override defaults of the image processor.
        Returns:
            `int`: Number of image patches per image.
        """
        min_pixels = images_kwargs["min_pixels"] if "min_pixels" in images_kwargs else self.size["shortest_edge"] # type: ignore
        max_pixels = images_kwargs["max_pixels"] if "max_pixels" in images_kwargs else self.size["longest_edge"] # type: ignore
        patch_size = images_kwargs.get("patch_size", self.patch_size) # type: ignore
        merge_size = images_kwargs.get("merge_size", self.merge_size) # type: ignore
        focus_size = images_kwargs.get("focus_size", self.focus_size) # type: ignore

        factor = patch_size * merge_size * focus_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__ = ["ZFQwen2VLImageProcessorFast"]