Skip to content

Latest commit

ย 

History

History
114 lines (102 loc) ยท 8.5 KB

DJ_SORA.md

File metadata and controls

114 lines (102 loc) ยท 8.5 KB

English | ไธญๆ–‡้กต้ข


Data is the key to the unprecedented development of large multi-modal models such as SORA. How to obtain and process data efficiently and scientifically faces new challenges! DJ-SORA aims to create a series of large-scale, high-quality open-source multi-modal data sets to assist the open-source community in data understanding and model training.

DJ-SORA is based on Data-Juicer (including hundreds of dedicated video, image, audio, text and other multi-modal data processing operators and tools) to form a series of systematic and reusable Multimodal "data recipes" for analyzing, cleaning, and generating large-scale, high-quality multimodal data.

This project is being actively updated and maintained. We eagerly invite you to participate and jointly create a more open and higher-quality multi-modal data ecosystem to unleash the unlimited potential of large models!

Overview

Motivation

  • SORA only briefly mentions using DALLE-3 to generate captions and can handle varying durations, resolutions and aspect ratios.
  • High-quality large-scale fine-grained data helps to densify data points, aiding models to better learn the conditional mapping of "text -> spacetime token", and solve a series of existing challenges in text-to-video models:
    • Smoothness of visual flow, with some generated videos exhibiting dropped frames and static states.
    • Text comprehension and fine-grained detail, where the produced results have a low match with the given prompts.
    • Generated content showing distortions and violations of physical laws, especially when entities are in motion.
    • Short video content, mostly around ~10 seconds, with little to no significant changes in scenes or backdrops.

Roadmap

Overview

Support high-performance loading and processing of video data

  • [โœ…] Parallelize data loading and storing:
    • [โœ…] lazy load with pyAV and ffmpeg
    • [โœ…] Multi-modal data path signature
  • [โœ…] Parallelization operator processing:
    • [โœ…] Support single machine multicore running
    • [โœ…] GPU utilization
    • [โœ…] Ray based multi-machine distributed running
    • [โœ…] Aliyun PAI-DLC & Slurm based multi-machine distributed running
  • [โœ…] Distributed scheduling optimization (OP-aware, automated load balancing) --> Aliyun PAI-DLC
  • [WIP] Low precision acceleration support for video related operators. (git tags: dj_op, dj_efficiency)
  • [WIP] SOTA model enhancement of existing video related operators. (git tags: dj_op, dj_sota_models)

Basic Operators (video spatio-temporal dimension)

  • Towards Data Quality
    • [โœ…] video_resolution_filter (targeted resolution)
    • [โœ…] video_aspect_ratio_filter (targeted aspect ratio)
    • [โœ…] video_duration_filter (targeted) duration)
    • [โœ…] video_motion_score_filter (video continuity dimension, calculating optical flow and removing statics and extreme dynamics)
    • [โœ…] video_ocr_area_ratio_filter (remove samples with text areas that are too large)
  • Towards Data Diversity & Quantity
    • [โœ…] video_resize_resolution_mapper (enhancement in resolution dimension)
    • [โœ…] video_resize_aspect_ratio_mapper (enhancement in aspect ratio dimension)
    • [โœ…] video_split_by_duration_mapper (enhancement in time dimension)
    • [โœ…] video_split_by_key_frame_mapper (enhancement in time dimension with key information focus)
    • [โœ…] video_split_by_scene_mapper (enhancement in time dimension with scene continuity focus)

Advanced Operators (fine-grained modal matching and data generation)

  • Towards Data Quality
    • [โœ…] video_frames_text_similarity_filter (enhancement in the spatiotemporal consistency dimension, calculating the matching score of key/specified frames and text)
  • Towards Diversity & Quantity
    • [โœ…] video_tagging_from_frames_mapper (with lightweight image-to-text models, spatial summary information from dense frames)
    • [โœ…] video_captioning_from_frames_mapper (heavier image-to-text models, generating more detailed spatial information from fewer frames)
    • [โœ…] video_tagging_from_audio_mapper (introducing audio classification/category and other meta information)
    • [โœ…] video_captioning_from_audio_mapper (incorporating voice/dialogue information; AudioCaption for environmental and global context)
    • [โœ…] video_captioning_from_video_mapper (video-to-text model, generating spacetime information from continuous frames)
    • [โœ…] video_captioning_from_summarizer_mapper (combining the above sub-abilities, using pure text large models for denoising and summarizing different types of caption information)
    • [WIP] video_interleaved_mapper (enhancement in ICL, temporal, and cross-modal dimensions), interleaved_modes include:
      • text_image_interleaved (placing captions and frames of the same video in temporal order)
      • text_audio_interleaved (placing ASR text and frames of the same video in temporal order)
      • text_image_audio_interleaved (alternating stitching of the above two types)

Advanced Operators (Video Content)

  • [โœ…] video_deduplicator (comparing hash values to deduplicate at the file sample level)
  • [โœ…] video_aesthetic_filter (performing aesthetic scoring filters after frame decomposition)
  • [โœ…] Compatibility with existing ffmpeg video commands
    • audio_ffmpeg_wrapped_mapper
    • video_ffmpeg_wrapped_mapper
  • [โœ…] Video content compliance and privacy protection operators (image, text, audio):
    • [โœ…] Mosaic
    • [โœ…] Copyright watermark
    • [โœ…] Face blurring
    • [โœ…] Violence and Adult Content
  • [TODO] (Beyond Interpolation) Enhancing data authenticity and density
    • Collisions, lighting, gravity, 3D, scene and phase transitions, depth of field, etc.
    • Filter-type operators: whether captions describe authenticity, relevance scoring/correctness of that description
    • Mapper-type operators: enhance textual descriptions of physical phenomena in video data
    • ...

DJ-SORA Data Recipes and Datasets

  • Support for unified loading and conversion of representative datasets (other-data <-> dj-data), facilitating DJ operator processing and dataset expansion.
    • [โœ…] Video-ChatGPT: 100K video-instruction data: {<question, answer, youtube_id>}
    • [โœ…] Youku-mPLUG-CN: 36TB video-caption data: {<caption, video_id>}
    • [โœ…] InternVid: 234M data sample: {<caption, youtube_id, start/end_time>}
    • [โœ…] MSR-VTT: 10K video-caption data: {<caption, video_id>}
    • [โœ…] ModelScope's datasets integration
    • [โœ…] VideoInstruct-100K, Panda70M, ......
  • Large-scale high-quality DJ-SORA dataset
    • [โœ…] (Data sandbox) Building and optimizing multimodal data recipes with DJ-video operators (which are also being continuously extended and improved).
    • [โœ…] Continuous expansion of data sources: open-datasets, Youku, web, ...
    • Large-scale analysis, cleaning, and generation of high-quality multimodal datasets based on DJ recipes (OpenVideos, ...)
      • [WIP] broad scenarios, high-dynamic
    • ...

DJ-SORA Data Validation and Model Training

  • Exploring and refining the collaborative development of multimodal data and model, establishing benchmarks and insights. paper
  • [WIP] Integration of SORA-like model training pipelines
  • [โœ…] (Model-Data sandbox) With relatively small models and the DJ-SORA dataset, exploring low-cost, transferable, and instructive data-model co-design, configurations and checkpoints.
  • [WIP] Training SORA-like models with DJ-SORA data on larger scales and in more scenarios to improve model performance.
  • ...