A problem about test our own face images

[weiyt2014]

Now, the code is normal running.But I found your databases include the file" .pts".Because our databases don't have the file ".pts",Can I use your test_model.m to realise face-alignment? If poosible,how to realise ? Thanks very much!

[jwyang]

Congratulations! If possible, hope you can share your results with others.

of course you can save the final shape into pts file in your own format. It's just a simple I/O process.

At 2014-10-21 11:33:47, "weiyt2014" [email protected] wrote:

Now, the code is normal running.But I found your databases include the file" .pts".Because our databases don't have the file ".pts",Can I use your test_model.m to realise face-alignment? If poosible,how to realise ? Thanks very much!

— Reply to this email directly or view it on GitHub.

[jwyang]

Hi, @weiyt2014 , another issue should be pointed out is that you need train the face alignment model for to adapt to your own face detector. As you know, the current codes you used for face alignment used the bounding box of groundtruth shape as input. It may be not applicable to your detector. To address such problem, I suggest three solutions:

1. use your own detector to initialize the bounding box of face alignment;

2. transfer the bounding boxes obtained from your detector to those of groundtruth shape;

3. train a detector which is expeted to output the bounding boxes as those of groundtruth shape.

For me, I chose the first solution.

[ghost]

@jwyang Hi I have trained using the database (afw) and generated LBFRegModel.mat for testing an image of mine which does not have the ground truth points . how do i obtain the landmark points and observe the output (face alignment ) i.e rotation translation and scaling

thanks in advance

[jwyang]

@sunilrao43 , Hello, you can extract the testing codes which do not rely on groundtruth landmarks inputs. Alternatively, you can make a fake pts file for your own images, so that it matches the test codes.