#Japan #Tokyo #Qlean Dataset #Visual Bank #AI Training Data

Visual Bank Unveils Qlean Dataset for AI Training

Visual Bank, based in Shibuya, Tokyo, has initiated the provision of the Qlean Dataset through its subsidiary, Amana Images. This dataset serves as an AI training solution designed to accommodate various research and commercial AI development needs. Recently, they launched a new segment of the dataset that includes not just single subjects but also diverse collections featuring multiple subjects. This aims to facilitate more realistic and generalized learning experiences for AI systems.

Overview of Qlean Dataset

The Qlean Dataset is part of a unique lineup of data offerings termed 'Data Recipes,' specifically crafted to support commercial use. The 'Data Recipes' feature a flexible combination of ready-to-use data materials, customizable to suit various applications, accuracy levels, and timelines. The dataset includes some annotated and unannotated images, and it can also be adjusted or expanded according to specific requirements.

Collaborations with notable partners like Chiba Lotte Marines and Toyo Keizai Inc., along with domestic and international networks, are pivotal in enhancing the dataset's offerings. This significant expansion aims to drastically reduce the burden of data collection and preparation in AI development environments, thereby accelerating the speed of AI solutions.

Key Features of the New Dataset

The newly launched dataset contains a wide range of full-body images of foreign individuals, captured in various settings—both indoors and outdoors, encompassing individuals from regions such as Africa, Sri Lanka, India, the Middle East, Europe, Australia, China, and Southeast Asia. Here are some of the detailed specifications:

• - Subjects Count: 843 individuals of diverse nationalities.
• - Image Count: 33,720 images.
• - Data Format: JPEG.
• - Data Size: Approximately 500GB, with individual image sizes ranging from 20MB to 30MB.

Use Case Scenarios

The diverse imagery provided in the Qlean Dataset opens up several practical applications:
1. Object Detection and Pose Estimation: Utilize multinational, full-body images for training algorithms to enhance detection and tracking capabilities in surveillance cameras and robotics.
2. Virtual Fitting in E-commerce: Generate 3D avatars that reflect a variety of body types and clothing styles for improved sizing recommendations, ultimately lowering return rates and enhancing purchase conversions in online retail.
3. Diverse Avatar Creation via Generative AI: Create advertisements and game characters that accurately represent a global market, all while minimizing biases in nationality and scene representation.
4. Autonomous Vehicle Applications: Utilize full-body images to improve pedestrian detection and behavior prediction, enhancing safety in vehicle control systems.
5. Healthcare Analytics: Apply the dataset for developing AI solutions in rehabilitation and posture correction, enabling quantitative effect measurement and early risk detection.
6. Bias Evaluation in AI: Evaluate facial recognition and classification algorithms for bias using a diverse dataset to comply with regulations like the EU AI Act.
7. AI Autofocus Enhancement: Improve camera and smartphone autofocus mechanisms by training on varying distances and poses of subjects, thus refining focus accuracy in challenging lighting conditions.

Commitment to Ethical Use

The Qlean Dataset abides by ethical standards, ensuring that consent for data collection and usage is obtained from all subjects. Compliance with privacy policies across different regions is paramount, allowing researchers and businesses to utilize the dataset confidently.

Unique Offerings for Custom Data

To accommodate special requirements, Qlean Dataset allows for the creation of custom datasets beyond the standard offerings in the Data Recipes lineup. This flexibility empowers clients to request tailored datasets that meet unique specifications.

For more information on Qlean Dataset, including access to the Data Recipes and customized requests, visit the official site: Qlean Dataset

About Visual Bank

Visual Bank operates as a startup focused on maximizing AI development capabilities through innovative data infrastructure. Their mission is to 'Unlock the Potential of All Data,' and they support the development of IP × AI use cases through their subsidiary, Amana Images, which offers various AI training datasets.

CEO: Masayuki Nagai
Location: 5-3-23 Kojimachi, Chiyoda, Tokyo 102-0083, Japan
Visual Bank Website: https://visual-bank.co.jp/
Amana Images Website: https://amanaimages.com/about/