Robots working alongside people need timely information about human posture, movement and occupied space to make appropriate decisions as interactions unfold. Bringing human mesh recovery onto the embedded device in real-time makes this information available locally during operation.
Human Mesh Recovery (HMR) is one of the leading features of Seeing Machines’ Human-Centred Physical AI Platform, launched in August 20261. Available through the platform, it reconstructs 3D persons from a single-view RGB or RGB-D input (comprising an RGB image and the corresponding depth map) in real-time on embedded systems.
Given a bounding box around a person, the system predicts articulated dense 3D pose, 3D body shape, and metric camera-frame placement achieving state-of-the-art 3D pose estimation accuracy with significantly lower latency (Figure 2) than frontier models such as Meta’s SAM-3D-Body2 and Fast-SAM-3D-Body3. Both HMR and HMR-Lite support RGB and RGB-D operation without increasing their respective parameter counts.
