Skip to content

Latest commit

 

History

History
6 lines (4 loc) · 383 Bytes

README.md

File metadata and controls

6 lines (4 loc) · 383 Bytes

FocusnetAblationTests

Various Ablation tests that eventually led to the optimal configuration of the Focusnet (ISBI2019) architecture

The experiments are done for 3D face landmarking using their 2D representations (Depth maps). Please request for the paper/see the paper for the tests tried.

The work deals with regressing pixel locations on 2D grids using attention mechanisms.