3D【4】人臉重建:Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric
該論文主要是用CNN直接重建3D人臉，而不是預測3DMM模型的參數，網絡結構是漏斗網絡（ HG）。論文提出了三種方案，第一種是直接從圖片重建，第二種是加入了人臉特徵點，第三種是多任務（重建+人臉特徵點預測）。效果最好的是第二種方法。 值得注意的是，該論文不是直接預測頂點的x、y、z座標。而是將mesh轉成voxel，變成一個192*192*200的矩陣。這樣就比較適合CNN。我們先看看mesh和v
相關文章
- 1. Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression
- 2. 3D Face Reconstruction from a Single Image
- 3. Large Pose 3D Face Reconstruction 文章理解
- 4. 3D【3】人臉重建:Image Based 3D Face Reconstruction: A Survey閱讀筆記
- 5. 人臉識別-3D:Learning from Millions of 3D Scans for Large-scale 3D Face Recognition
- 6. (3D-HPE)Learning to Estimate 3D Human Pose and Shape from a Single Color Image
- 7. 3D人臉重建：《Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network》
- 8. 3D【11】人臉重建：Disentangling Features in 3D Face Shapes for Joint Face Reconstruction and Recognition
- 9. Face Video Generation from a Single Image and Landmarks
- 10. 3D【9】人臉重建:Joint3D Face Reconstruction and Dense Alignment with Position Map Regression
- 更多相關文章...
- • ASP.NET Image 控件 - ASP.NET 教程
- • MySQL DISTINCT：去重（過濾重複數據） - MySQL教程
- • 再有人問你分佈式事務，把這篇扔給他
- • 適用於PHP初學者的學習線路和建議
相關標籤/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。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression
- 2. 3D Face Reconstruction from a Single Image
- 3. Large Pose 3D Face Reconstruction 文章理解
- 4. 3D【3】人臉重建:Image Based 3D Face Reconstruction: A Survey閱讀筆記
- 5. 人臉識別-3D:Learning from Millions of 3D Scans for Large-scale 3D Face Recognition
- 6. (3D-HPE)Learning to Estimate 3D Human Pose and Shape from a Single Color Image
- 7. 3D人臉重建：《Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network》
- 8. 3D【11】人臉重建：Disentangling Features in 3D Face Shapes for Joint Face Reconstruction and Recognition
- 9. Face Video Generation from a Single Image and Landmarks
- 10. 3D【9】人臉重建:Joint3D Face Reconstruction and Dense Alignment with Position Map Regression