Akai Sports Video Captions V1: 1,119 Expert-Reviewed Video Clips

1,119 sports video clips are paired with English captions authored and reviewed by expert labelers from Akai Space Labs. The dataset is designed for training and evaluating multimodal models. It was last updated on June 14, 2026.

Use Cases

• Train video captioning models based on the expert-reviewed English captions.
• Benchmark cross-modal retrieval systems using the paired video and text data.
• Fine-tune multimodal foundation models on a domain-specific sports video corpus.

Strengths

• Dataset contains 1,119 individual video clips.
• Captions were authored and reviewed by expert labelers, suggesting quality control.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Akai Space Labs

Collection Method

Video was contributed by contributors; captions were authored and reviewed by Akai labelers.

Freshness

Last updated 2026-06-14 19:07:02; freshness should be verified.