# coding=utf-8 # Copyright 2025 The Qwen Team 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. """video processor class for Qwen3-VL https://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen3_vl/video_processing_qwen3_vl.py""" import math import numpy as np import torch from typing import Callable, Optional, Union from transformers.models.qwen2_vl.image_processing_qwen2_vl import smart_resize from transformers.feature_extraction_utils import BatchFeature from transformers.image_utils import ChannelDimension, PILImageResampling, SizeDict, get_image_size from transformers.processing_utils import Unpack, VideosKwargs from transformers.utils import TensorType, add_start_docstrings, logging, is_torchvision_v2_available from transformers.video_processing_utils import BASE_VIDEO_PROCESSOR_DOCSTRING, BaseVideoProcessor from transformers.video_utils import ( VideoInput, group_videos_by_shape, reorder_videos, is_valid_video, make_batched_videos, ) from .videochat3_utils import VideoChat3VideoMetadata if is_torchvision_v2_available(): from torchvision.transforms.v2 import functional as F else: from torchvision.transforms import functional as F logger = logging.get_logger(__name__) def smart_video_resize( num_frames: int, height: int, width: int, temporal_factor: int = 1, factor: int = 28, frame_min_pixels: int = 16 * 28 * 28 * 4, frame_max_pixels: int = 1024 * 28 * 28 * 4, video_max_total_pixels: int = 5000 * 28 * 28 * 4, ): assert temporal_factor == 1, "temporal_factor must be 1 for videochat3!" if num_frames < temporal_factor: raise ValueError(f"t:{num_frames} must be larger than temporal_factor:{temporal_factor}") if height < factor or width < factor: raise ValueError(f"height:{height} or width:{width} must be larger than factor:{factor}") elif 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, w_bar = smart_resize(height, width, factor, frame_min_pixels, frame_max_pixels) t_bar = round(num_frames / temporal_factor) * temporal_factor if t_bar * h_bar * w_bar > video_max_total_pixels: beta = math.sqrt((num_frames * height * width) / video_max_total_pixels) h_bar = max(factor, math.floor(height / beta / factor) * factor) w_bar = max(factor, math.floor(width / beta / factor) * factor) return h_bar, w_bar class VideoChat3VideoProcessorInitKwargs(VideosKwargs): patch_size: Optional[int] temporal_patch_size: Optional[int] merge_size: Optional[int] min_frames: Optional[int] max_frames: Optional[int] @add_start_docstrings( "Constructs a fast Qwen3-VL image processor that dynamically resizes videos based on the original videos.", BASE_VIDEO_PROCESSOR_DOCSTRING, """ patch_size (`int`, *optional*, defaults to 16): The spacial 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. """, ) class VideoChat3VideoProcessor(BaseVideoProcessor): resample = PILImageResampling.BICUBIC size = {"shortest_edge": 128 * 32 * 32, "longest_edge": 768 * 32 * 32} video_max_total_pixels = 1000000000 * 32 * 32 image_mean = [0.5, 0.5, 0.5] image_std = [0.5, 0.5, 0.5] do_resize = True do_rescale = True do_normalize = True do_convert_rgb = True patch_size = 14 temporal_patch_size = 1 merge_size = 2 temporal_merge_size = 4 fps = 2 min_frames = 4 max_frames = 1024 do_sample_frames = True valid_kwargs = VideoChat3VideoProcessorInitKwargs model_input_names = ["pixel_values_videos", "video_grid_thw"] def __init__(self, **kwargs: Unpack[VideoChat3VideoProcessorInitKwargs]): super().__init__(**kwargs) if self.size is not None and ( self.size.get("shortest_edge", None) is None or self.size.get("longest_edge", None) is None ): raise ValueError("size must contain 'shortest_edge' and 'longest_edge' keys.") def _further_process_kwargs( self, size: Optional[SizeDict] = None, **kwargs, ) -> dict: """ Update kwargs that need further processing before being validated Can be overridden by subclasses to customize the processing of 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.") return super()._further_process_kwargs(size=size, **kwargs) def get_num_sampled_frames( self, metadata: VideoChat3VideoMetadata, num_frames: Optional[int] = None, fps: Optional[Union[int, float]] = None ): if fps is not None and num_frames is not None: raise ValueError("`num_frames` and `fps` are mutually exclusive arguments, please use only one!") if metadata.clip_start_time is not None and metadata.clip_end_time is not None: total_num_frames = int((metadata.clip_end_time - metadata.clip_start_time) * metadata.fps) else: total_num_frames = metadata.total_num_frames sample_fps = fps if fps is not None else self.fps # If num_frames is not given but fps is, calculate num_frames from fps if num_frames is None and fps is not None: if metadata.fps is None: raise ValueError("`fps` is not provided in video metadata.") num_sampled_frames = int(total_num_frames / metadata.fps * sample_fps) num_sampled_frames = min(min(max(num_sampled_frames, self.min_frames), self.max_frames), total_num_frames) elif num_frames is not None: num_sampled_frames = min(min(max(num_frames, self.min_frames), self.max_frames), total_num_frames) else: raise ValueError("`num_frames` and `fps` are not provided for sampling frames.") return num_sampled_frames def sample_frames( self, metadata: VideoChat3VideoMetadata, num_frames: Optional[int] = None, fps: Optional[Union[int, float]] = None, **kwargs, ): """ Default sampling function which uniformly samples the desired number of frames between 0 and total number of frames. If `fps` is passed along with metadata, `fps` frames per second are sampled uniformty. Arguments `num_frames` and `fps` are mutually exclusive. Args: video (`torch.Tensor`): Video that need to be sampled. metadata (`VideoChat3VideoMetadata`): Metadata of the video containing information about total duration, fps and total number of frames. num_frames (`int`, *optional*): Maximum number of frames to sample. Defaults to `self.num_frames`. fps (`int` or `float`, *optional*): Target frames to sample per second. Defaults to `self.fps`. Returns: torch.Tensor: Sampled video frames. """ num_sampled_frames = self.get_num_sampled_frames(metadata, num_frames, fps) if metadata.clip_start_time is not None and metadata.clip_end_time is not None: start_idx = int(metadata.clip_start_time * metadata.fps) end_idx = int(metadata.clip_end_time * metadata.fps) assert end_idx <= metadata.total_num_frames, f"end_idx:{end_idx} must be less than or equal to total_num_frames:{metadata.total_num_frames} (实际上合法值是小于， 如果等于则勉强接受)" # 确保索引不超过 total_num_frames - 1 end_idx = min(end_idx, metadata.total_num_frames - 1) indices = np.linspace(start_idx, end_idx, num_sampled_frames).round().astype(int) else: indices = np.linspace(0, metadata.total_num_frames - 1, num_sampled_frames).round().astype(int) return indices def _decode_and_sample_videos( self, videos: VideoInput, video_metadata: Union[VideoChat3VideoMetadata, dict], do_sample_frames: Optional[bool] = None, sample_indices_fn: Optional[Callable] = None, ) -> list["torch.Tensor"]: """ Decode input videos and sample frames if needed. """ videos = make_batched_videos(videos) # 自定义处理video_metadata，避免使用make_batched_metadata if video_metadata is None: video_metadata = [None] * len(videos) elif isinstance(video_metadata, (VideoChat3VideoMetadata, dict)): video_metadata = [video_metadata] elif isinstance(video_metadata, list): # 确保每个元素都是VideoChat3VideoMetadata或dict processed_metadata = [] for metadata in video_metadata: if isinstance(metadata, dict): # 如果是dict，转换为VideoChat3VideoMetadata processed_metadata.append(VideoChat3VideoMetadata(**metadata)) elif isinstance(metadata, VideoChat3VideoMetadata): processed_metadata.append(metadata) else: # 如果是其他类型，尝试转换 processed_metadata.append(VideoChat3VideoMetadata(**metadata.__dict__)) video_metadata = processed_metadata _is_valid_video = is_valid_video(videos[0]) # Only sample frames if an array video is passed, otherwise first decode -> then sample if _is_valid_video and do_sample_frames: sampled_videos = [] for video, metadata in zip(videos, video_metadata): indices = sample_indices_fn(metadata=metadata) metadata.frames_indices = indices # NOTE: @Lixinhao, for _calculate_timestamps! sampled_videos.append(video[indices]) videos = sampled_videos elif not _is_valid_video: if isinstance(videos[0], list): # Videos sometimes are passed as a list of image URLs, especially through templates videos = [ torch.stack([F.pil_to_tensor(image) for image in images], dim=0) for images in self.fetch_images(videos) ] if do_sample_frames: sampled_videos = [] for video, metadata in zip(videos, video_metadata): indices = sample_indices_fn(metadata=metadata) metadata.frames_indices = indices # NOTE: @Lixinhao, for _calculate_timestamps! sampled_videos.append(video[indices]) videos = sampled_videos else: videos = [ torch.stack([F.pil_to_tensor(image) for image in images], dim=0) for images in self.fetch_images(videos) ] else: # 使用父类的fetch_videos方法，但不传递sample_indices_fn videos, metadata_list = super().fetch_videos(videos, sample_indices_fn=None) # 将VideoMetadata转换为VideoChat3VideoMetadata video_metadata = [] for metadata in metadata_list: if metadata is None: # 如果metadata是None，跳过 continue elif isinstance(metadata, VideoChat3VideoMetadata): video_metadata.append(metadata) else: # 转换为VideoChat3VideoMetadata video_metadata.append(VideoChat3VideoMetadata( total_num_frames=metadata.total_num_frames, fps=metadata.fps, width=metadata.width, height=metadata.height, duration=metadata.duration, video_backend=metadata.video_backend, frames_indices=metadata.frames_indices, video_start_time=0.0, clip_start_time=None, clip_end_time=None )) # 如果需要采样帧，使用我们自己的sample_indices_fn if do_sample_frames and sample_indices_fn is not None: sampled_videos = [] for video, metadata in zip(videos, video_metadata): indices = sample_indices_fn(metadata=metadata) metadata.frames_indices = indices sampled_videos.append(video[indices]) videos = sampled_videos return videos, video_metadata def _preprocess( self, videos: list[torch.Tensor], do_convert_rgb: bool = True, do_resize: bool = True, size: Optional[SizeDict] = None, interpolation: PILImageResampling = PILImageResampling.BICUBIC, do_rescale: bool = True, rescale_factor: float = 1 / 255.0, do_normalize: bool = True, 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, return_tensors: Optional[Union[str, TensorType]] = None, **kwargs, ): grouped_videos, grouped_videos_index = group_videos_by_shape(videos) resized_videos_grouped = {} for shape, stacked_videos in grouped_videos.items(): B, T, C, H, W = stacked_videos.shape num_frames, height, width = T, H, W if do_resize: resized_height, resized_width = smart_video_resize( num_frames=num_frames, height=height, width=width, temporal_factor=temporal_patch_size, factor=patch_size * merge_size, frame_min_pixels=size.shortest_edge, frame_max_pixels=size.longest_edge, video_max_total_pixels=self.video_max_total_pixels, ) stacked_videos = stacked_videos.view(B * T, C, H, W) stacked_videos = self.resize( stacked_videos, size=SizeDict(height=resized_height, width=resized_width), interpolation=interpolation, ) stacked_videos = stacked_videos.view(B, T, C, resized_height, resized_width) resized_videos_grouped[shape] = stacked_videos resized_videos = reorder_videos(resized_videos_grouped, grouped_videos_index) # Group videos by size for further processing # Needed in case do_resize is False, or resize returns videos with different sizes grouped_videos, grouped_videos_index = group_videos_by_shape(resized_videos) processed_videos_grouped = {} processed_grids = {} for shape, stacked_videos in grouped_videos.items(): resized_height, resized_width = get_image_size(stacked_videos[0], channel_dim=ChannelDimension.FIRST) # Fused rescale and normalize stacked_videos = self.rescale_and_normalize( stacked_videos, do_rescale, rescale_factor, do_normalize, image_mean, image_std ) patches = stacked_videos # Check that videos have `num_frames` divisible by `temporal_patch_size` 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, ) 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_videos_grouped[shape] = flatten_patches processed_grids[shape] = [[grid_t, grid_h, grid_w]] * batch_size processed_videos = reorder_videos(processed_videos_grouped, grouped_videos_index) processed_grids = reorder_videos(processed_grids, grouped_videos_index) pixel_values_videos = torch.cat(processed_videos, dim=0) video_grid_thw = torch.tensor(processed_grids) data = { "pixel_values_videos": pixel_values_videos, "video_grid_thw": video_grid_thw, } return BatchFeature(data=data, tensor_type=return_tensors) def get_number_of_video_tokens(self, num_frames: int, height: int, width: int, videos_kwargs=None): if num_frames % self.temporal_merge_size != 0: num_clips = num_frames // self.temporal_merge_size + 1 else: num_clips = num_frames // self.temporal_merge_size return num_clips * height * width // self.merge_size**2 __all__ = ["VideoChat3VideoProcessor"]