Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression
2017IEEE Motivation 无论基本假设是什么，算法的输入和输出是什么，三维人脸重建在一般复杂管道中都需要解决非凸优化问题，无论是模型构建(训练期间)还是模型拟合 (在测试).例如使用标准L2损失将所有3D点连接为一个向量进行直接回归可能会导致学习困难，因为必须预测每个3D顶点的单个正确值；此外，这种方法需要将所有扫描插入到一个固定维数的向量中，当使用CNN对模型参数(如3DMM参数)
相关文章
- 1. 3D【4】人脸重建:Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric
- 2. 3D Face Reconstruction from a Single Image
- 3. Large Pose 3D Face Reconstruction 文章理解
- 4. Face Video Generation from a Single Image and Landmarks
- 5. (3D-HPE)Learning to Estimate 3D Human Pose and Shape from a Single Color Image
- 6. TVCG Collaborative Large-Scale Dense 3D Reconstruction with Online Inter-Agent Pose Optimisation
- 7. (3D-HPE) Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image
- 8. 论文解读：3D Hand Shape and Pose Estimation from a Singl RGB Image
- 9. Large-Scale Volumetric Scene Reconstruction using LiDAR
- 10. Image-based 3D Object Reconstruction
- 更多相关文章...
- • RSS
元素 - RSS 教程 - • ASP.NET Image 控件 - ASP.NET 教程
- • 漫谈MySQL的锁机制
- • 互联网组织的未来：剖析GitHub员工的任性之源
相关标签/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
欢迎关注本站公众号,获取更多信息
相关文章
- 1. 3D【4】人脸重建:Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric
- 2. 3D Face Reconstruction from a Single Image
- 3. Large Pose 3D Face Reconstruction 文章理解
- 4. Face Video Generation from a Single Image and Landmarks
- 5. (3D-HPE)Learning to Estimate 3D Human Pose and Shape from a Single Color Image
- 6. TVCG Collaborative Large-Scale Dense 3D Reconstruction with Online Inter-Agent Pose Optimisation
- 7. (3D-HPE) Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image
- 8. 论文解读：3D Hand Shape and Pose Estimation from a Singl RGB Image
- 9. Large-Scale Volumetric Scene Reconstruction using LiDAR
- 10. Image-based 3D Object Reconstruction