processing_timelens.py

# Modified from https://github.com/huggingface/transformers/blob/v4.57.1/src/transformers/models/qwen2_5_vl/processing_qwen2_5_vl.py
# Copyright 2025 The Qwen Team 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.

import numpy as np
import torch
from transformers import Qwen2_5_VLProcessor
from transformers.feature_extraction_utils import BatchFeature
from transformers.models.qwen2_5_vl.processing_qwen2_5_vl import (
    Qwen2_5_VLProcessorKwargs,
)


class TimeLensProcessor(Qwen2_5_VLProcessor):
    r"""
    Constructs a Qwen2.5-VL processor which wraps a Qwen2.5-VL image processor and a Qwen2 tokenizer into a single processor.
    [`Qwen2_5_VLProcessor`] offers all the functionalities of [`Qwen2VLImageProcessor`] and [`Qwen2TokenizerFast`]. See the
    [`~Qwen2_5_VLProcessor.__call__`] and [`~Qwen2_5_VLProcessor.decode`] for more information.
    Args:
        image_processor ([`Qwen2VLImageProcessor`], *optional*):
            The image processor is a required input.
        tokenizer ([`Qwen2TokenizerFast`], *optional*):
            The tokenizer is a required input.
        video_processor ([`Qwen2_5_VLVideoProcessor`], *optional*):
            The video processor is a required input.
        chat_template (`str`, *optional*): A Jinja template which will be used to convert lists of messages
            in a chat into a tokenizable string.
    """

    def __init__(
        self,
        image_processor=None,
        tokenizer=None,
        video_processor=None,
        chat_template=None,
        **kwargs,
    ):
        super().__init__(
            image_processor, tokenizer, video_processor, chat_template, **kwargs
        )
        # ============ [TimeLens] Modification BEGIN ============
        self.vision_start = (
            "<|vision_start|>"
            if not hasattr(tokenizer, "vision_start")
            else tokenizer.vision_start
        )
        self.vision_end = (
            "<|vision_end|>"
            if not hasattr(tokenizer, "vision_end")
            else tokenizer.vision_end
        )
        # ============ [TimeLens] Modification END ==============

    def __call__(
        self,
        images=None,
        text=None,
        videos=None,
        **kwargs,
    ) -> BatchFeature:
        """
        Main method to prepare for the model one or several sequences(s) and image(s). This method forwards the `text`
        and `kwargs` arguments to Qwen2TokenizerFast's [`~Qwen2TokenizerFast.__call__`] if `text` is not `None` to encode
        the text. To prepare the vision inputs, this method forwards the `vision_infos` and `kwargs` arguments to
        Qwen2VLImageProcessor's [`~Qwen2VLImageProcessor.__call__`] if `vision_infos` is not `None`.

        Args:
            images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `list[PIL.Image.Image]`, `list[np.ndarray]`, `list[torch.Tensor]`):
                The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch
                tensor. Both channels-first and channels-last formats are supported.
            text (`str`, `list[str]`, `list[list[str]]`):
                The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings
                (pretokenized string). If the sequences are provided as list of strings (pretokenized), you must set
                `is_split_into_words=True` (to lift the ambiguity with a batch of sequences).
            videos (`np.ndarray`, `torch.Tensor`, `list[np.ndarray]`, `list[torch.Tensor]`):
                The image or batch of videos to be prepared. Each video can be a 4D NumPy array or PyTorch
                tensor, or a nested list of 3D frames. Both channels-first and channels-last formats are supported.
            return_tensors (`str` or [`~utils.TensorType`], *optional*):
                If set, will return tensors of a particular framework. Acceptable values are:
                - `'tf'`: Return TensorFlow `tf.constant` objects.
                - `'pt'`: Return PyTorch `torch.Tensor` objects.
                - `'np'`: Return NumPy `np.ndarray` objects.
                - `'jax'`: Return JAX `jnp.ndarray` objects.

        Returns:
            [`BatchFeature`]: A [`BatchFeature`] with the following fields:

            - **input_ids** -- List of token ids to be fed to a model. Returned when `text` is not `None`.
            - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when
              `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not
              `None`).
            - **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`.
            - **pixel_values_videos** -- Pixel values of videos to be fed to a model. Returned when `videos` is not `None`.
            - **image_grid_thw** -- List of image 3D grid in LLM. Returned when `images` is not `None`.
            - **video_grid_thw** -- List of video 3D grid in LLM. Returned when `videos` is not `None`.
            - **second_per_grid_ts** -- List of video seconds per time grid. Returned when `videos` is not `None`.
        """
        output_kwargs = self._merge_kwargs(
            Qwen2_5_VLProcessorKwargs,
            tokenizer_init_kwargs=self.tokenizer.init_kwargs,
            **kwargs,
        )

        image_inputs = videos_inputs = {}
        if images is not None:
            image_inputs = self.image_processor(
                images=images, **output_kwargs["images_kwargs"]
            )
            image_grid_thw = image_inputs["image_grid_thw"]

        if videos is not None:
            # ============ [TimeLens] Modification BEGIN ============
            # videos is a list of (video_tensor, metadata) tuples
            videos, metadata = [v[0] for v in videos], [v[1] for v in videos]
            # Duplicate frames at even indices
            for cur_video_tensor in videos:
                cur_video_tensor[1::2] = cur_video_tensor[::2]
            # Calculate sampled timestamps for each video
            frames_timestamps = [
                [
                    idx / cur_metadata["fps"]
                    for idx in cur_metadata["frames_indices"][::2]
                ]
                for cur_metadata in metadata
            ]

            videos_inputs = self.video_processor(
                videos=videos, **output_kwargs["videos_kwargs"]
            )
            video_grid_thw = videos_inputs["video_grid_thw"]
            # ============ [TimeLens] Modification END ==============

        if not isinstance(text, list):
            text = [text]

        text = text.copy()  # below lines change text in-place
        if images is not None:
            merge_length = self.image_processor.merge_size**2
            index = 0
            for i in range(len(text)):
                while self.image_token in text[i]:
                    num_image_tokens = image_grid_thw[index].prod() // merge_length
                    text[i] = text[i].replace(
                        self.image_token, "<|placeholder|>" * num_image_tokens, 1
                    )
                    index += 1
                text[i] = text[i].replace("<|placeholder|>", self.image_token)

        if videos is not None:
            merge_length = self.video_processor.merge_size**2
            index = 0
            # ============ [TimeLens] Modification BEGIN ============
            for i in range(len(text)):
                while self.video_token in text[i]:
                    cur_video_tokens = ""
                    num_tokens_per_frame = (
                        video_grid_thw[index][1:].prod() // merge_length
                    )
                    per_frame_tokens = (
                        self.vision_start
                        + "<|placeholder|>" * num_tokens_per_frame
                        + self.vision_end
                    )
                    for cur_frames_timestamp in frames_timestamps[index]:
                        cur_video_tokens += (
                            f"{cur_frames_timestamp:.1f}s: " + per_frame_tokens
                        )

                    text[i] = text[i].replace(
                        self.vision_start + self.video_token + self.vision_end,
                        cur_video_tokens,
                        1,
                    )
                    index += 1
                text[i] = text[i].replace("<|placeholder|>", self.image_token)
            # modeling_qwen2_5_vl.py calls `.item()` on image_grid_thw to convert t, h, w from tensor to int, so we create image_grid_thw as Tensor to be compatible with `.item()` call
            image_grid_thw = torch.tensor(
                [
                    [1, grid_h, grid_w]
                    for grid_t, grid_h, grid_w in video_grid_thw
                    for _ in range(grid_t)
                ],
                dtype=torch.long,
            )

            image_inputs = {
                "pixel_values": videos_inputs[
                    "pixel_values_videos"
                ],  # [grid_t * grid_h * grid_w, channel * temporal_patch_size * patch_size * patch_size] = [num_patches, dim]
                "image_grid_thw": image_grid_thw,
            }
            videos_inputs = {}
            # ============ [TimeLens] Modification END ==============

        return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None)
        return_mm_token_type_ids = output_kwargs["text_kwargs"].pop(
            "return_mm_token_type_ids", None
        )
        text_inputs = self.tokenizer(text, **output_kwargs["text_kwargs"])
        self._check_special_mm_tokens(text, text_inputs, modalities=["image", "video"])

        if return_mm_token_type_ids:
            array_ids = np.array(text_inputs["input_ids"])
            mm_token_type_ids = np.zeros_like(text_inputs["input_ids"])
            mm_token_type_ids[array_ids == self.image_token_id] = 1
            text_inputs["mm_token_type_ids"] = mm_token_type_ids.tolist()

        return BatchFeature(
            data={**text_inputs, **image_inputs, **videos_inputs},
            tensor_type=return_tensors,
        )