Sports Videos Dataset: Curated Collection for Generative AI and Action Analysis

MIT

Description

Sports Videos Dataset: Curated Collection for Generative AI and Action Analysis

Overview

This dataset features 1,119 short sports videos, each paired with detailed text descriptions in a video-text modality. Focused on diverse movements and actions across sports categories, it was curated through a structured process to ensure relevance and clarity.

Ideal applications include:

  • Generative AI (GenAI) tasks: video-to-text generation, text-to-video synthesis, multimodal content creation.
  • Action recognition, sports analytics, and related tasks in computer vision and NLP.

Dataset Details

Video Specifications

  • Duration: ~30 seconds (optimized for efficient processing).
  • Quality: Minimal background clutter, no cinematic filters, clear full-body movements.
  • Camera Angle: Primarily front-facing to highlight form and actions.

Annotations

  • Content: Text descriptions detailing actions, movements, and context for paired learning.
  • Size: 1,119 videos collected from dedicated contributors.
  • Format: Standard formats for ML pipeline integration.

Categories

Organized into non-overlapping categories for broad coverage:

  1. Field Sports – Dynamic team actions (e.g., football, cricket, hockey) in clean angles.
  2. Gym Workouts & Strength Training (Solo) – Solo exercises like weightlifting and yoga, emphasizing form.
  3. Indoor Court Games – Active play (e.g., badminton, volleyball) with continuous movement.
  4. Track & Field / Athletics – Events like running and jumping, capturing full motion.
  5. Martial Arts, Boxing & Combat Sports – Real training and drills in combat styles.
  6. Adventure & Extreme Sports – High-energy activities (e.g., parkour, climbing) with clear movements.
  7. Water Sports – Aquatic pursuits (e.g., swimming, diving) prioritizing athlete visibility.
  8. Cross-Category Enhancer – Warmups, stretching, and posture clips to supplement others.

Potential Applications

Generative AI Focus

  • Fine-tuning models for video captioning.
  • Generating descriptive narratives from visuals.
  • Text-based video retrieval.
  • Creating synthetic sports content via multimodal learning.

Other Uses

  • Action detection in computer vision.
  • Performance analysis and injury prevention in sports analytics.
  • Benchmarking for research on real-world datasets.

Inspired by established datasets, this resource empowers GenAI innovation in sports-related AI.

Dataset Information

Number of Datapoints

1119

Created By

Teamakai

Timestamps

Created At

August 18th, 2025 8:51 AM

Last Updated

August 19th, 2025 1:50 PM

Sr. No.VideoDescription
1

The video captures a dynamic women's 400-meter hurdles race at the World Athletics Championships, showcasing athletes in mid-race as they expertly nav...

2

The video captures a dynamic baton exchange during the men's 4x400m relay race, set inside a brightly lit stadium with "World Athletics Relay" brandin...

3

The video captures the electrifying atmosphere of a men's relay final at the IAAF World Championships in London, held in a large stadium bustling with...

4

In an outdoor track-and-field stadium, the video captures a male athlete engaged in a triple jump event. The scene unfolds under overcast skies, evide...

5

The video captures a male athlete executing a discus throw using a rotational start technique within a concrete throwing circle on a sports field. The...

6

The video captures a male athlete engaged in a hurdles training drill on a standard athletics track, emphasizing rhythm and efficient clearance over m...

7

The video captures an athlete performing a triple jump at an outdoor sports complex, showcasing the technique's distinct phases. Initially, the athlet...

8

The video captures a shot put event on a red synthetic track, featuring an athlete executing the rotational or "spin" technique. Dressed in a light-co...

9

The video captures a training drill for a track and field hurdles event, focusing on the preparatory phase where the athlete, dressed in sportswear, e...

10

The video captures an exhilarating track-and-field sprinting event held under bright stadium floodlights during an evening session, ensuring clear vis...

PRODUCTS

AI Engine

Enterprise solutions

De-AI Workflows

Akai Marketplace

Akailon

COMPANY

About Us

Careers

Security

Privacy

RESOURCES

Partners

Contact

Blog

GUIDE

Akailon Guide

Get started with labeling

Partner with us

Marketplace

Onboarding Enterprise

SOCIALS

Copyright © 2025 Akai Space Labs. All rights reserved.

Terms & Privacy policies