#Japan #AI Research #Minato #Qlean Dataset #Visual Bank

Visual Bank Unveils Innovative Qlean Dataset

Visual Bank Co., Ltd., based in Minato Ward, Tokyo, has recently expanded its offerings with the addition of a new dataset named Qlean Dataset. This dataset focuses on figure skating, comprising a collection of videos and images featuring both male and female skaters. The launch aims to bolster AI research and development, especially in commercial applications and various academic fields.

Overview of the New Figure Skating Dataset

The newly launched Figure Skating Video and Image Dataset includes:

• - Contents: Images and videos capturing singles and pairs in figure skating performances.
• - Subject Attributes: The dataset features two Japanese skaters in their 20s—one male and one female.
• - Shooting Locations: The images were filmed indoors at a skating rink, utilizing various angles to capture detailed movements.

Specific Actions Captured

• - Single Skating: Includes jumps, step sequences, spins, and more.
• - Pair Skating: Features lifts, spin combinations, death spirals, and step sequences.

Dataset Capacities

• - Video Format: The dataset comprises a total of 6 videos, with a combined size of 13.17 GB, in MP4 format, documenting nearly 11 minutes and 50 seconds of performance.
• - Image Format: The image collection contains 17,124 photos, amounting to 116.04 GB, saved in JPEG format.

Further details of the video and image datasets can be found here:

• - Video Sample
• - Image Sample

Use Cases for the New Dataset

The new figure skating dataset is expected to serve various research and development purposes, including:

1. Development of Pose Estimation Models

The dataset captures complex aerial movements such as jumps and spins, which can aid in the training of skeleton recognition and pose estimation algorithms. This has valuable applications in sports science and rehabilitation research focusing on body control analysis.

2. Behavior Recognition and Interaction Analysis for Pairs

By documenting pair-specific movements from multiple angles, the dataset allows for simultaneous analysis of the two skaters’ actions. This can enhance AI models focused on recognizing cooperative movements and analyzing interactions between humans.

3. Development of AI Judging Systems

The dataset provides detailed features such as the number of rotations in jumps and the stability of lifts, which can be leveraged for developing automated AI judging systems and evaluation tools tailored to figure skating's judging criteria.

4. Recreation of Performances in VR/AR and Metaverse Environments

The diverse angles captured allow for lifelike performance recreation within virtual reality spaces, benefitting applications in virtual viewing experiences, educational content, and competition simulations in metaverse settings.

5. Robotics and Motion Learning with Artistic Consideration

The unique balance, gliding, and rotational movements in figure skating serve as fresh training material for robotics motion imitation, particularly in research focused on the dual requirements of artistry and mechanics.

6. Development of High-Speed Autofocus and Object Tracking AI

Data from the dataset, detailing the fast and intricate movements of skaters, is ideal for studying and validating autofocus and object tracking algorithms. Camera and imaging equipment manufacturers can utilize this dataset to enhance next-gen autofocus performance.

Unique Features of the Qlean Dataset

Visual Bank’s Qlean Dataset provides several distinct advantages:

• - Compliance with Research and Commercial Use Cases: All subjects featured have consented to data use, ensuring compliance with privacy policies across various nations, enabling confidence in utilizing this data for research and commercial purposes.
• - Fast and Efficient ROI Maximization: The dataset is accessible through the proprietary “AI Data Recipe” approach, which allows companies to secure the necessary data quickly while minimizing initial investment.
• - Customization Capabilities: If there are specific data needs not met within the existing lineup, the Qlean Dataset can be tailored and constructed to meet individual requirements.

For inquiries or further information, you can visit the Qlean Dataset Contact Page or access the service site at Qlean Dataset Service.

Supporting Academia

In an effort to integrate academia into AI advancements, Visual Bank has initiated a free dataset provision program targeting universities, research institutions, and non-profit tech development teams. This program encompasses a wide array of over 80 different types of data, exceeding 500,000 samples, addressing the common issue of insufficient access to high-quality, rights-clear training data. Visual Bank, recognized as a GENIAC-selected company, aims to contribute solutions to this challenge. More details can be found at the Qlean Dataset Academia Support Program.

About Visual Bank

Visual Bank is a startup with a mission to unlock the potential of data across its applications. Apart from providing AI-supported tools for creators and its Qlean Dataset for AI development, it has a subsidiary called Amana Images, which focuses on dataset development. Additionally, Visual Bank is part of the national research program GENIAC, accelerating its initiative towards societal implementation.

CEO: Masayuki Nagai
Address: 6F C-Cube Minami Aoyama Bldg., 7-1-7 Minami Aoyama, Minato-ku, Tokyo
Official Website: Visual Bank
Amana Images: Amana Images