Please make sure you follow the instructions in INSTALL.md to build necessary dependencies and download our pretrained checkpoints.
To run the test code, you also need to download the mesh_release.zip in our NBA2K dataset and unzip it on your local machine.
You might need to change the following options in img_to_mesh/src/experiments/mesh/mesh_test.yaml
:
data_root_dir
: Set it to the mesh_release dataset path on your machine.vis_gpu
: Set it as your CUDA_VISIBLE_DEVICES.base_log_dir
: The root dir of the mesh log folder. If you follow the instruction of downloading pretrained checkpoints, you don't need to change it.log_name
: Name of the mesh log folder. If you follow the instruction of downloading pretrained checkpoints, you don't need to change it.pose_base_log_dir
: The root dir of the pose log folder. If you follow the instruction of downloading pretrained checkpoints, you don't need to change it.pose_log_name
: Name of the pose log folder. If you follow the instruction of downloading pretrained checkpoints, you don't need to change it.
You also need to change the following lines in img_to_mesh/src/experiments/mesh/mesh_run.sh
:
cfg_path=experiments/mesh/mesh_test.yaml
After you finish above steps, run the testing code:
cd img_to_mesh/src
bash experiments/mesh/mesh_run.sh
We report the performance on the released NBA2K Dataset. The results are slightly different from what we provide in the paper as the dataset is different.
MPVPE | MPVPE-PA | EMD | CD |
---|---|---|---|
65.991 | 44.622 | 0.073 | 2.870 |
MPVPE is mean per vertex position error in mm. MPVPE-PA is mean per vertex position error in mm after procrustes alignment. EMD is earth-mover distance. CD is chamfer distance scaled by 1000. EMD and CD are computed after Iterative Closest Point alignment.