Introducing the Qlean Dataset: Diverse Japanese Portrait Images for AI Innovation

Qlean Dataset: Expanding Data Solutions for AI Development

Visual Bank Inc., based in Minato, Tokyo, is pioneering the provision of AI training data solutions through its subsidiary, Amana Images. Their recent addition of the 'Diverse Japanese Portrait Image Dataset' to the Qlean Dataset lineup marks a significant expansion of their AI development toolkit. This innovative dataset is especially tailored to meet the needs of various research and commercial AI projects.

The Qlean Dataset is designed for commercial use, offering a collection of original data that is versatile and readily available for immediate application. The dataset can be customized according to specific requirements, ensuring flexibility in how data is aggregated and utilized. Some data is already annotated, while others can be modified based on user needs. Visual Bank is also enhancing its offerings by collaborating with partners like the Chiba Lotte Marines and Toyo Keizai Inc., leveraging both domestic and international networks.

Overview of the Diverse Japanese Portrait Image Dataset

This newly launched dataset offers:

- Content: Images of people in various situations and environments
- Format: JPEG images
- Total Files: 10,000 images
- Meta information: Included
- Pixel Dimensions: Minimum 2,300px on the longer edge (e.g., 640×427)
- Sample Detail URL: Sample Dataset

Use Cases for the New Dataset

1. Development of Person Recognition and Action Classification AI: The dataset features a variety of occupations and situations such as healthcare, office settings, outdoor work, and customer service. These scenarios are ideal for training AI models that accurately identify human behavior and posture. The dataset supports industrial AI development, enabling tasks like person detection and action analysis.

2. Research on AI Models by Occupation and Environment: The dataset includes numerous images capturing distinct characteristics of various professions such as medical staff, office workers, and food service personnel. This resource is invaluable for developing AI systems that recognize specific occupational attributes and for evaluating sector-specific AI models.

3. Training on Pose Estimation and Motion Analysis AI: With diverse human movements depicted—ranging from standing poses to actions during conversations—this dataset serves as training data for systems focusing on pose estimation and motion analysis. It’s particularly useful in fields such as manufacturing, healthcare, and education, where safety management and motion optimization are key.

4. Emotion and Facial Expression Recognition AI Research: The dataset contains images reflecting a range of emotions, including natural smiles and thoughtful expressions. These scenes are essential for training AI systems in emotion recognition and facial expression analysis, contributing to advancements in customer service AI and educational support systems.

5. Development of Generative AI and Digital Humans: Utilizing high-resolution images that encompass diverse professions and environments, the dataset is also applicable for generating realistic images or avatars through generative AI. This can aid in creating domestic digital human projects along with advertisements and educational AI content.

Features of the Qlean Dataset

The Qlean Dataset supports both research and commercial use by ensuring all images come with necessary consent forms from subjects. This commitment to privacy guarantees compliance with various international privacy policies, making it a reliable choice for serious developers.

Distinctively, the dataset is offered through the 'AI Data Recipe', enabling efficient data acquisition while maximizing return on investment. For data that is not currently available in the 'AI Data Recipe', Visual Bank can create customized datasets tailored to unique requirements.

Support for Academia

As part of its commitment to academic advancement, Visual Bank has launched a program providing free access to datasets for universities, research institutions, and non-profit tech development teams. Over 80 types of data, including images, audio, video, and text, are available—totalling more than 500,000 data points—to aid in overcoming the challenge of accessing high-quality, rights-cleared training data. More details can be found on their Academia Support Program site.

About Visual Bank

Visual Bank is a startup dedicated to creating and providing next-generation data infrastructure aimed at harnessing the vast potential of data. Their mission, 'Unlocking the Potential of All Data', encompasses various initiatives, including the AI-assisted tool 'THE PEN' for manga artists and the development service for training datasets, 'Qlean Dataset'. The company is proud to be part of the national research program 'GENIAC', which accelerates their efforts towards social implementation.

CEO: Masayuki Nagai
Location: 6F C-Cube Minami Aoyama Building, 7-1-7 Minami Aoyama, Minato, Tokyo 107-0062
Company URL: Visual Bank Inc.
Amana Images URL: Amana Images