3D reconstruction

3D Image reconstruction using multiresolution hierarchies.

3D Image reconstruction

multiresolution hierarchies

Many tasks in geometry processing and physical simulation benefit from multiresolution hierarchies. In this project i aimed to re-implement and run this method to explore thi existing techniques such as surface mesh decimation, voxelization.The method constructs each next-coarsest level of the hierarchy using a sequence of decimation, flow. I also learnt using OPenGL,mesh with c++ and all the framework related to 3D analysis
The Images shows 3D reconstruction for bunny(nested cage method) and face reconstruction.
3D Face Reconstruction from a Single Image

3D face reconstruction is a fundamental Computer Vision problem of extraordinary difficulty. this paper uses a single image as input and i prove their effectiveness by running the code and using only one image(2D) that uses the CNN and facial landmark localization methods.
The image in the left shows, the paper output and method and the videos in the right presents my own re-implementation for Action recognition Dataset.