提出了DeepHuman,体到体(volume to volume)的3D人体重建模 将从SMPL模型生成的 稠密语义表示 作为额外的输入,来让重建的物体表面更清晰,甚至还能重建不可见的部分 通过体积特征变换(volumetric feature transformation),把不同尺度的图像特征融合起来,得以恢复精确的表面;然后通过标准提炼网络(normal refinement network)进...
cd path/to/deephuman git clone https://github.com/ZhengZerong/DeepHuman cd DeepHuman virtualenv deephuman_env source deephuman_env/bin/activate pip install -r requirements.txt Build voxelizer: cd voxelizer mkdir build & cd build cmake .. make Change the path configuration at LINE 23 of main_pr...
Deep Learning powered models of humans for virtual clinical trials. - GitHub - deepchem/deephuman: Deep Learning powered models of humans for virtual clinical trials.
论文链接:http://www.liuyebin.com/deephuman/assets/DeepHuman.pdf 摘要 我们提出深度人体,一个图像导向立体到立体的转换CNN对3D人体重建从一个单张RGB图像。为减少干扰涉及在表面几何重建,即使对不可见的区域也可以进行重建,我们提出和利用一个稠密的语义展示形成从SMPL模型作为一个外部的输入。一个...
@InProceedings{Zheng2019DeepHuman, author = {Zheng, Zerong and Yu, Tao and Wei, Yixuan and Dai, Qionghai and Liu, Yebin}, title = {DeepHuman: 3D Human Reconstruction From a Single Image}, booktitle = {The IEEE International Conference on Computer Vision (ICCV)}, month = {October}, year...