Real-world data from Southeast Asia.

Human Ground collects specialized datasets across Southeast Asia for AI teams worldwide. Collection spans language, voice, workplace activity and human interaction, supported by local recruitment and delivery, with participants’ consent and compensation.

What we collect

Language, voice and specialist data

Speech recordings, conversations, regional languages and dialects, and expert contributions for training and evaluating AI systems. Projects are organized around specific languages, participant profiles and subject areas.

Physical-world data

Video and sensor data from factories, construction sites, hotels, kitchens and other real environments, supporting robotics, computer vision and broader AI applications.

Human touch

Our human-touch dataset combines skilled practitioners’ movements, contact force and participants’ reports of intensity and comfort. The first release, HG-100, is in development.

How it works

End-to-end collection covers recruitment, permissions, coordination and quality checks against agreed specifications. Collection scales to 1,000 hours per week within 2–3 weeks of contract signing.

Contact

Human Ground was founded by David Garai.