Taiyuan, China
巨深 builds training data for embodied AI and related systems. We annotate datasets and capture synchronized, multi-camera recordings that partners can use as-is or to a supplied specification.
Two core lines of work, delivered to an agreed specification.
We label and review datasets for model training, following the taxonomy, quality rules, and acceptance criteria provided by the partner. Work can be scoped as a standalone annotation project or as a follow-on step after collection.
We run synchronized multi-camera capture for embodied-intelligence tasks, including first-person plus third-person setups and first-person plus wrist-mounted cameras. Collection is calibrated and time-aligned, and we can deliver in MCAP and LeRobot formats. Sample recordings are available to an agreed specification.
巨深 is a data company based in Taiyuan, Shanxi, China. We work with partners who need collection and annotation capacity for embodied AI and other data-intensive training work.
We can start from a partner’s written specification, provide sample recordings for review, and then scale collection across the environments listed above.