metadata

license: agpl-3.0
language:
  - en
tags:
  - code

Volleyball Video Analytics Dataset

Overview

This dataset is designed to support AI tools for analyzing volleyball games. It contains both raw footage (before processing) and short video snippets with 6 consecutive plays including downtime. GitHub Link

Planned Features 📋

Auto-Editor: Automatically remove downtime between plays for faster video review.
Score Tracking: Timestamp key moments and keep an accurate score throughout the video.
Action-Specific Replays: Replay only the most relevant portions of the video based on specific actions.
Court Usage Analysis: Understand where the ball is contacted most often during play.
Player-Specific Statistics: Collect and present performance metrics for individual players.

Our goal is to make video analysis easier for players, coaches, and teams by automating performance review through AI.

Dataset Structure

raw/: Contains unprocessed, full-length videos.
snippets/: Contains short clips, each representing 3 consecutive plays including downtime between plays.
metadata.json: Will contain every snippet and its source, type, date, camera_type, and snippet_index.
train.txt: List of train snippet file names
val.txt: List of validation snippet file names
test.txt: List of test snippet file names

Snippet Metadata

Each snippet includes:

Type: Game type (indoor, beach, etc.)
Date: Game date (e.g., YYYYMMDD)
Camera type: Normal or wide-angle
Snippet index: A unique number incremented for each new snippet

Example File Name

YouTube_Indoor_20241012_Wide_001.mp4

We welcome contributions! You can submit volleyball game footage, which would improve the tool performance on your videos!

Thanks to everyone contributing videos and labels, and to the volleyball and AI communities for their support!