Beyond Static Features for Temporally Consistent 3D Human Pose and Shape from a Video

Despite the recent success of single image-based 3D human pose and shapeestimation methods, recovering temporally consistent and smooth 3D human motionfrom a video is still challenging. Several video-based methods have beenproposed; however, they fail to resolve the single image-based methods'temporal inconsistency issue due to a strong dependency on a static feature ofthe current frame. In this regard, we present a temporally consistent meshrecovery system (TCMR). It effectively focuses on the past and future frames'temporal information without being dominated by the current static feature. OurTCMR significantly outperforms previous video-based methods in temporalconsistency with better per-frame 3D pose and shape accuracy. We also releasethe codes. For the demo video, see https://youtu.be/WB3nTnSQDII. For the codes,see https://github.com/hongsukchoi/TCMR_RELEASE.