ken881015

Just for displaying error information in percentage format... 

@liqidong I think it's weird, too. In "normal" mode, loss was defined below https://github.com/choyingw/SynergyNet/blob/072c577ec2add613f87c81cfc882c72c9a9a87c6/loss_definition.py#L37 So, It reasonable to think that the order of pose, shape, exp output by PointNet(reverse direction)...

@choyingw I think that I found another typo in released code. But I still want to discuss with you.  If you want to reverse the order of y axis...

I think the ckpt assigned by main.py is already prepared in this repo (and it's trained!!!). https://github.com/cleardusk/3DDFA/blob/master/models/phase1_wpdc_vdc.pth.tar And if you can run main.py, it obvious that the program can find...

@Zvyozdo4ka I don't know whether it is possible, but maybe it just because your images for testing are too similar? Here are some way I suggest you: - Can you...

@ZHJNCUT you have to use these 7 parameters to reconstruct the transform matrix (it's a homogeneous matrix 4x4 element, and only 3x4 is adjustable) which include the scaling, rotation, and...

Because for alignment, you have to not only **rotate** the face in 3D space, but also have to translate the face from **3D space** to **image coordinate**.

- Thanks for the reply. - The pic I show is **reannotated** version of benchmark. However, what surprises me is that it still has some bad annotations (such as pair(1,1),...

Do you guys make sure the **resolution** and **ckpt version** are assigned correctly (should match)? Here is the command I use. It seems the you are get the output from...

Are our commit version different? Mine has the last layer for modulate the channel from 513 to 512...