Datasets:

gouba2333
/

BoxComm-Dataset

video_id stringlengths 24 58	file_path stringlengths 64 98	fps float64 24 60	total_frames int64 2.48k 42.9k	language stringclasses 6 values	classified_segments listlengths 14 206
1_SZELECZKILila_LAM-Hoi-Ming-Sandy	/hdd/wangkw/dataset/shijs_processed/1_SZELECZKILila_LAM-Hoi-Ming-Sandy.mp4	50	32,841	en	[ { "text": "So here we go with our first bout of the World Boxing Championships here in Liverpool.", "start_frame": 101, "end_frame": 395, "start_time": 2.018, "end_time": 7.906, "class": 3, "words": [ { "word": "So", "start_time": 2.018, "end_time": 2.198, ...
10_FLOWERS-Carlos_ALIYEV-Sarkhan	/hdd/wangkw/dataset/shijs_processed/10_FLOWERS-Carlos_ALIYEV-Sarkhan.mp4	50	31,652	en	[{"text":"So here we go.","start_frame":114,"end_frame":146,"start_time":2.274,"end_time":2.915,"cla(...TRUNCATED)
100_MAMEDOV-Subhan_DE-LOS-SANTOS-Fredelin	/hdd/wangkw/dataset/shijs_processed/100_MAMEDOV-Subhan_DE-LOS-SANTOS-Fredelin.mp4	29.97003	19,780	en	[{"text":"party Azerbaijanis","start_frame":0,"end_frame":62,"start_time":0.002,"end_time":2.065,"cl(...TRUNCATED)
101_READSHAW-Reece_ATTRATTIVO-Salvatore	/hdd/wangkw/dataset/shijs_processed/101_READSHAW-Reece_ATTRATTIVO-Salvatore.mp4	29.97003	19,591	en	[{"text":"the man from Middlesbrough in England the 21 year old","start_frame":1,"end_frame":104,"st(...TRUNCATED)
102_McDONAGH-Martin-Christopher_NARENDER-Narender	/hdd/wangkw/dataset/shijs_processed/102_McDONAGH-Martin-Christopher_NARENDER-Narender.mp4	29.97003	19,666	en	[{"text":"something a little bit special when the very big boys meet.","start_frame":0,"end_frame":1(...TRUNCATED)
103_BANIULIS-Algirdas_WATTS-Kelvin	/hdd/wangkw/dataset/shijs_processed/103_BANIULIS-Algirdas_WATTS-Kelvin.mp4	29.97003	19,836	en	[{"text":"Banoelis, he wears all blue, the man from Kaunas.","start_frame":25,"end_frame":139,"start(...TRUNCATED)
104_BAYIKEWUZI-Danabieke_HERNANDEZ-MOREJON-Yordan-Alain	/hdd/wangkw/dataset/shijs_processed/104_BAYIKEWUZI-Danabieke_HERNANDEZ-MOREJON-Yordan-Alain.mp4	29.97003	16,182	en	[{"text":"Bulgaria in against Daniviecki Bekeuzi.","start_frame":20,"end_frame":131,"start_time":0.6(...TRUNCATED)
105_SHIHA-Omar-Mohamed_LOVCHYNSKYI-Dmytro	/hdd/wangkw/dataset/shijs_processed/105_SHIHA-Omar-Mohamed_LOVCHYNSKYI-Dmytro.mp4	29.97003	19,461	en	[{"text":"so we are underway in the men's super heavyweight","start_frame":136,"end_frame":207,"star(...TRUNCATED)
106_HAGAG-Ahmed_JU-Taeung	/hdd/wangkw/dataset/shijs_processed/106_HAGAG-Ahmed_JU-Taeung.mp4	29.97003	12,165	en	[{"text":"this is bout number 106 round of 32 in the men's super heavyweight 90 kilo plus","start_fr(...TRUNCATED)
107_MOURAD-Kadi_ZOKIROV-Jakhongir	/hdd/wangkw/dataset/shijs_processed/107_MOURAD-Kadi_ZOKIROV-Jakhongir.mp4	36	23,786	en	[{"text":"that they have found the right man to fill them in Zakirov in the red","start_frame":14,"e(...TRUNCATED)

End of preview. Expand in Data Studio

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Video Redistribution Notice

Since we do not own redistribution rights for the original broadcasts, we do not redistribute raw videos. We instead release original video URLs, timestamps, commentary annotations, category labels, punch-event labels, evaluation files, prompts, and reconstruction scripts for non-commercial academic research.

The official source URL table is provided at metadata/source_urls.csv.

Available reconstruction metadata:

metadata/source_urls.csv: official source URLs and match metadata.
metadata/timestamp.csv: original-video round boundary timestamps used for trimming.
metadata/timestamp_processed.csv: round boundary timestamps after processing/free-frame removal.

BoxComm-Dataset

BoxComm-Dataset is the official data release for BoxComm, a benchmark for category-aware boxing commentary generation and narration-rhythm evaluation.

Resources

Project Page: https://gouba2333.github.io/BoxComm
Paper: http://arxiv.org/abs/2604.04419
Code: https://github.com/gouba2333/BoxComm
Benchmark: https://huggingface.co/datasets/gouba2333/BoxComm

Overview

This dataset release is intended for training, analysis, and reproducible preprocessing. It contains the complete processed videos together with the released annotations and benchmark metadata.

Recommended structure:

BoxComm-Dataset/
├── train/
│   ├── videos/
│   ├── events/
│   └── asr/
├── eval/
│   ├── videos/
│   ├── events/
│   └── asr/
└── metadata/

The split convention is:

train: video id < 478
eval: video id >= 478

Each event directory should contain:

one skeleton .pkl file
one video_event_inference_3.json file

Each ASR JSON file should contain classified_segments.

What is included

processed match videos
event annotations
skeleton data
ASR with sentence segmentation
3-way commentary labels
split metadata

Intended uses

supervised fine-tuning for commentary generation
category-aware commentary evaluation
narration-rhythm analysis
multimodal sports video understanding research

Data preparation in the code repository

The official code repository provides:

scripts/prep_qwen3vl_sft_data.py
scripts/train_qwen3vl.py
scripts/infer_qwen3vl.py
scripts/eval_metrics.py
scripts/eval_streaming_cls_metrics.py

Repository: https://github.com/gouba2333/BoxComm

Licensing

The public release includes processed videos, ASR annotations, event JSON files, skeleton PKL files, and benchmark metadata for research use.

Citation

@article{wang2026boxcomm,
  title={BoxComm: Benchmarking Category-Aware Commentary Generation and Narration Rhythm in Boxing},
  author={Wang, Kaiwen and Zheng, Kaili and Deng, Rongrong and Shi, Yiming and Guo, Chenyi and Wu, Ji},
  journal={arXiv preprint arXiv:2604.04419},
  year={2026}
}

Downloads last month: 1,572

Paper for gouba2333/BoxComm-Dataset

BoxComm: Benchmarking Category-Aware Commentary Generation and Narration Rhythm in Boxing

Paper • 2604.04419 • Published Apr 6