Jinhui Ye1*,
Zihan Wang2*,
Haosen Sun2,
Keshigeyan Chandrasegaran1,
Zane Durante1,
Cristobal Eyzaguirre1,
Yonatan Bisk3,
Juan Carlos Niebles1,
Ehsan Adeli1,
Li Fei-Fei1,
Jiajun Wu1,
Manling Li2
Stanford University1, Northwestern University2, Carnegie Mellon University3
CVPR 2025 . Dataset is part of the T* project
🌎Website |
🧑💻Code |
📄arXiv |
🏆 Leaderboard (Coming Soon)
#### Dataset Sample ```python { 'video_id': 'b6ae365a-dd70-42c4-90d6-e0351778d991', 'vclip_id': '6338b73e-393f-4d37-b278-68703b45908c', 'question_id': 10, 'question': 'What nail did I pull out?', 'answer': 'E', 'frame_indexes_vclip': [5036, 5232], # the keyframe indexes in the vclip 'frame_indexes_video': [5036, 5232], # the keyframe indexes in the video 'choices': { 'A': 'The nail from the front wheel fender', 'B': 'The nail from the motorcycle battery compartment', 'C': 'The nail from the left side of the motorcycle seat', 'D': 'The nail from the rearview mirror mount', 'E': 'The nail on the right side of the motorcycle exhaust pipe' }, 'video_metadata': { 'CLIP-reference-interval-vclip': [180.0, 240.0], # Time interval of the "vclip" that is considered to be important by CLIP. this is calculated by (CLIP-reference-interval-video - vclip-interval-in-video[0]) 'CLIP-reference-interval-video': [180.0, 240.0], # Time interval of the "video" that is considered to be important by CLIP. This is originally from the **Ego4D dataset**, used in our work for annotators to quickly locate in the video. 'vclip_interval_in_video': [0.0, 480.06667277018227], # the vclip start and end second, i.e., for [a, b], the vclip starts at the a second of the video, ends at the b second of the video 'frame_count': 14155, # Total number of frames in the video 'frame_rate': 30.0, # Frame rate of the video 'duration': 471.8333435058594, # Duration of the video that are valid and unbroken, in seconds 'resolution': '454x256', # Original resolution of the video 'frame_dimensions': None, # Frame dimensions (if available) 'codec': 'N/A', # Codec used for the video (if available) 'bitrate': 0, # Bitrate of the video (if available) 'frame_dimensions_resized': [340, 256], # Resized frame dimensions 'resolution_resized': '340x256', # Resized resolution 'video_id': 'b6ae365a-dd70-42c4-90d6-e0351778d991' # Unique video identifier } } ``` #### Dataset exploration add hyperlink to demo #### Dataset Usage ```python from datasets import load_dataset dataset = load_dataset("LVHaystack/LongVideoHaystack") print(dataset['train']) ``` ```bash Dataset({ features: ['video_id', 'vclip_id', 'question', 'options', 'answer', 'frame_indexes_video', 'frame_indexes_vclip', 'video_metadata'], num_rows: 12892 }) ``` #### Download and Process Video Source TODO: We plan to provide a script of how to download a subset from [Ego4d](https://ego4d-data.org/) and process them. Below is download part adapted from their [official guide](https://github.com/facebookresearch/Ego4d/tree/main/ego4d/cli) and we will add video2clip script soon. ```bash pip install ego4d ego4d --output_directory=your_path/videos/ \ --datasets full_scale annotations \ --metadata \ --video_uid_file video_uids.txt # python process_videos_to_clips.py # TODO ``` Please find [video_uids.txt](https://huggingface.co/datasets/LVHaystack/LongVideoHaystack/blob/main/video_uids.txt) in our repo, or you can generate it by: ```python import datasets metadata = datasets.load_dataset("LVHaystack/LongVideoHaystack-metadata")["metadata"] with open("video_uids.txt", "w") as file: for video_id in list(set(metadata['video_id'])): file.write(video_id + " ") ``` To follow evaluation for LongVideoBench in our paper, please find script to transform LongVideoBench to LongVideoHaystack format in `transform_longvideobench.py`. #### Dataset Statistics Summary | **Metric** | **Total** | **Train** | **Val** | **Test** | **Test_Tiny** | |-------------------------------|-----------|-----------|---------|----------|---------------| |🎥 Video Statistics | | | | | | | Total Videos | **988** | **858** | **71** | **53** | **13** | | Total Video Duration (hr) | 420 | 370 | 27 | 24 | 4.3 | | Avg. Video Duration (min) | 26 | 26 | 23 | 28 | 20 | |🎞️ Clip Statistics | | | | | | | Total Video Clips | **1,324** | **1,141** | **89** | **80** | **17** | | Total Clip Duration (hr) | 180 | 160 | 12 | 11 | 2.2 | | Avg. Clip Duration (sec) | 490 | 490 | 480 | 470 | 460 | |🖼️ Frame Statistics | | | | | | | Total Frames (k) | **45,716**| **40,150**| **2,958**| **2,637**| **467** | | Avg. Frames per Video (k) | 46 | 47 | 42 | 50 | 36 | | Ratio of Keyframe / Frame (‰) | 0.62 | 0.59 | 0.69 | 0.78 | 0.89 | |❓ QA Statistics | | | | | | | Total QA Pairs | **15,092**| **12,892**| **1,000**| **1,000**| **200** | | Avg. QA Pairs per Video | 15 | 15 | 14 | 19 | 15 | | Avg. QA Pairs per Clip | 11 | 11 | 11 | 13 | 12 | | Avg. Keyframes per Question | 1.9 | 1.8 | 2.0 | 2.1 | 2.1 | #### Evaluation scripts Please refer to [./eval.py](https://huggingface.co/datasets/LVHaystack/LongVideoHaystack/blob/main/eval.py). #### Contact - Jinhui Ye: jinhuiyes@gmail.com - Zihan Wang: zihanw@u.northwestern.edu (datasets) - Haosen Sun: haosensun2026@u.northwestern.edu - Keshigeyan Chandrasegaran: keshik@stanford.edu - Anabella Aisaro: anabellaisaro2025@u.northwestern.edu - Manling Li: manling.li@northwestern.edu #### Citation ```bibtex @misc{tstar, title={Re-thinking Temporal Search for Long-Form Video Understanding}, author={Jinhui Ye and Zihan Wang and Haosen Sun and Keshigeyan Chandrasegaran and Zane Durante and Cristobal Eyzaguirre and Yonatan Bisk and Juan Carlos Niebles and Ehsan Adeli and Li Fei-Fei and Jiajun Wu and Manling Li}, year={2025}, eprint={2504.02259}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2504.02259}, } ``` Website template borrowed from [HourVideo](https://huggingface.co/datasets/HourVideo/HourVideo).