raymond ren comments

Results 13 comments of


                                            raymond ren

训练报错，RuntimeError: The size of tensor a (3) must match the size of tensor b (0) at non-singleton dimension 3 求解答

解决了吗，老大

训练报错，RuntimeError: The size of tensor a (3) must match the size of tensor b (0) at non-singleton dimension 3 求解答

我解决了这个问题，应该是eye_blink_dim设置问题，找到egs\datasets\May\lm3d_radnerf_sr.yaml修改如下 eye_blink_dim: 8 #2 把2改为8，就ok；其实不用它原来的checkpoints是不存在这个问题的

训练时 raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format 求指导

应该是eye_blink_dim设置问题，找到egs\datasets\May\lm3d_radnerf_sr.yaml修改如下 eye_blink_dim: 8 https://github.com/yerfor/GeneFacePlusPlus/pull/2 把2改为8，就ok；其实不用它原来的checkpoints是不存在这个问题的

ModuleNotFoundError: No module named '_raymarching_face'

hi, find the setup.py under the directory, change as following: cmdclass={ 'build_ext': BuildExtension.with_options(use_ninja=False),

Unable to extract image

hi,guy change extract_segment_imgs.py as following and try again: def generate_segment_imgs_job(img_name, segmap, img, model): out_img_name = segmap_name = img_name.replace("/gt_imgs/", "/segmaps/").replace(".jpg", ".png") # 存成jpg的话，pixel value会有误差 try: os.makedirs(os.path.dirname(out_img_name), exist_ok=True) except: pass encoded_segmap =...

<class 'ValueError'> Found array with 0 sample(s) (shape=(0, 2)) while a minimum of 1 is required by NearestNeighbors.

没有人看一下吗，我用的是windows powershell，有没有大佬帮忙看看

<class 'ValueError'> Found array with 0 sample(s) (shape=(0, 2)) while a minimum of 1 is required by NearestNeighbors.

knn 中这里出错， bg_xys = np.stack(np.nonzero(~bc_pixs)).transpose() fg_xys = np.stack(np.nonzero(bc_pixs)).transpose() nbrs = NearestNeighbors(n_neighbors=1, algorithm='kd_tree').fit(fg_xys)

<class 'ValueError'> Found array with 0 sample(s) (shape=(0, 2)) while a minimum of 1 is required by NearestNeighbors.

knn 中这里出错， bg_xys = np.stack(np.nonzero(~bc_pixs)).transpose() fg_xys = np.stack(np.nonzero(bc_pixs)).transpose() nbrs = NearestNeighbors(n_neighbors=1, algorithm='kd_tree').fit(fg_xys)

<class 'ValueError'> Found array with 0 sample(s) (shape=(0, 2)) while a minimum of 1 is required by NearestNeighbors.

这个问题已经解决了

训练时一直停在Sanity Val

去安装一下rysnc试试