SSP-3D
Sports Shape and Pose 3D
3D3d meshesImagesIntroduced 2020-09-21
SSP-3D is an evaluation dataset consisting of 311 images of sportspersons in tight-fitted clothes, with a variety of body shapes and poses. The images were collected from the Sports-1M dataset. SSP-3D is intended for use as a benchmark for body shape prediction methods. Pseudo-ground-truth 3D shape labels (using the SMPL body model) were obtained via multi-frame optimisation with shape consistency between frames, as described here.
Benchmarks
1 Image, 2*2 Stitchi/PVE-T-SC1 Image, 2*2 Stitchi/PVE-T1 Image, 2*2 Stitchi/mIOU3D/PVE-T-SC3D/PVE-T3D/mIOU3D Absolute Human Pose Estimation/PVE-T-SC3D Absolute Human Pose Estimation/PVE-T3D Absolute Human Pose Estimation/mIOU3D Human Pose Estimation/PVE-T-SC3D Human Pose Estimation/PVE-T3D Human Pose Estimation/mIOUPose Estimation/PVE-T-SCPose Estimation/PVE-TPose Estimation/mIOU