SM4RT addresses monocular 4D reconstruction by modeling motion as structured geometry rather than independent point-wise displacement. Its Structure-of-Motion representation decomposes scene dynamics into a compact set of motion bases, each expressed as a temporal sequence of 6D twists in SE(3). Time-shared per-pixel assignment weights recover dense motion while encouraging points on the same object to follow a shared rigid-body trajectory. The proposed parallel motion geometry encoder and decoder jointly infer 3D geometry, world-coordinate motion, and kinematic structure from monocular RGB video in one forward pass. The abstract reports strong motion reconstruction performance, but provides no numerical results.
No heat snapshots are available in the last 24 hours.