Shaofei Wang
Shaofei Wang
Hi, thanks for your interest in our work! The optimization code does SMPL+D fitting at this line: https://github.com/taconite/PTF/blob/a8789c9f752aea2944c2a75e04cc2aa21c7e4a00/smpl_registration/fit_SMPLD_PTFs.py#L498 So you can feed your raw scan to the optimization script and...
init_smpl_meshes其实就是前面fit SMPL步骤输出的mesh，它们的作用regularize你的SMPL+D优化过程： https://github.com/taconite/PTF/blob/a8789c9f752aea2944c2a75e04cc2aa21c7e4a00/smpl_registration/fit_SMPLD_PTFs.py#L138 这里的laplacian loss表明，你优化过后的SMPL+D mesh (th_smpl_meshes)相对于你的初始SMPL mesh（init_smpl_meshes）的形变应当尽量小。 所以优化的完整流程应该是：1、先优化SMPL parameter，得到SMPL mesh；2、用第一步得到SMPL parameter和SMPL mesh作为初始化，再去优化SMPL+D的parameter。第一步的SMPL优化很关键，如果不准的话，第二步的结果也不会好
The baselines (neural body, animatable nerf) use resized 512x512 images as inputs, please see here for example https://github.com/zju3dv/neuralbody/blob/333026fc12f33d5e732008e7094b442a0095a1e2/configs/zju_mocap_exp/latent_xyzc_313.yaml#L76 This is also evidenced by the fact that the rendering results they...
> ``` > File "/codes/arah-release/im2mesh/data/zju_mocap.py", line 414, in __getitem__ > fg_inds = np.random.choice(valid_inds.shape[0], size=self.num_fg_samples, replace=False) > File "mtrand.pyx", line 965, in numpy.random.mtrand.RandomState.choice > ValueError: Cannot take a larger sample than...
> That means SMPL result was wrong. But the SMPL result isn't wrong And below is the that SMPL result.  > No, this means the camera parameters are wrong...
The `center_img` at line 268 specifies the center of data augmentation - which is just resizing for now. The `crop_new` function handles data augmentation and ideally, it should support flipping/rotation/scaling;...
The folder should be empty
It's some dummy logging directory created with test.py - the reason is that when running test.py for the demo, I do not use WandbLogger as the logger for PyTorch lightning,...
Hi, Bingoang Thanks for your interest! Regarding your questions: 1) mIoU is already computed in the validation function. Per-vertex error is computed in the registration code. 2) For outer/inner chamfer...
No. There is no segmentation augmentation. Your case is most likely the result of a misalignment between the 2D bounding box (from which we sample camera rays) and the actual...