copilot/OpenSeeFace
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Added informal comparison to 3DDFA_V2.

Browse source
This commit is contained in:
Emiliana 2020-09-05 12:39:01 +02:00
parent 0d61b061a3
commit 415f0975c9
3 changed files with 4 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

BIN
Images/3DDFA_V2.jpg Normal file
View file

Binary file not shown.
Side by side

After

Width:  |  Height:  |  Size: 1.2 MiB

BIN
Images/3DDFA_V2_800.jpg Normal file
View file

Binary file not shown.
Side by side

After

Width:  |  Height:  |  Size: 371 KiB

4
README.md
View file

@ -103,6 +103,10 @@ Pytorch weights for use with `model.py` can be found [here](https://mega.nz/file
More samples: [Results3.png](https://raw.githubusercontent.com/emilianavt/OpenSeeFace/master/Images/Results3.png), [Results4.png](https://raw.githubusercontent.com/emilianavt/OpenSeeFace/master/Images/Results4.png)
Here's an informal [comparison](https://raw.githubusercontent.com/emilianavt/OpenSeeFace/master/Images/3DDFA_V2.png) with [3DDFA_V2](https://github.com/cleardusk/3DDFA_V2) on five arbitrary, non-cherry picked examples from my test data:
![3DDFA_V2_800.png](https://raw.githubusercontent.com/emilianavt/OpenSeeFace/master/Images/3DDFA_V2_800.png)
## Face detection
The landmark model is quite robust with respect to the size and orientation of the faces, so the custom face detection model gets away with rougher bounding boxes than other approaches. It has a favorable speed to accuracy ratio for the purposes of this project.