0.5.0

Compatibility update for menpo 0.8.x

Add a pretrained AAM

Adds a pretrained AAM available at menpofit.aam.load_balanced_frontal_face_fitter. It will be automatically downloaded from the menpo website if it does not already exist.

0.4.0

This is a major new release of menpofit with numerous bug fixes and improvements. It includes complete documentation for the first time, and now depends on menpo v0.7.0.

Fix warping bug for HolisticAAM

Don't warp existing landmarks on images when building the shape normalized images.

Massive Changes!

This update represents an enormous set of changes that involve completely changing how menpofit builds models. Unfortunately, due to current time restrictions, we were unable to properly document nor extensively test all of the models. However, upon the release of menpobench we will endeavor to add proper documentation and use menpobench to perform quantitative testing of all of the models.

Fix SDM pickling bug

SDM can now be pickled.

0.2.0

menpofit 0.2.0 brings menpofit up to date with menpo 0.5.0 and adds support for the new IPython widgets introduced in IPython 3.

A much more significant improvement to menpofit will be coming later this summer.

0.1.0

This release coincides with the official release of Menpo v0.4.0. Mostly, this fixes visualization within menpofit for the widgets.

0.1.0 alpha 2

• AAMs, CLMs and SDMs are now pickleable (use menpo.io.export_pickle() to save)
• Faster dW_dp calculation for AAMs (thanks @trigeorgis)
• Fixes for widgets

The initial release for ACM

This is the first release of the menpofit package to the public and currently represents an alpha stage build.

This repository contains all the code for fitting and training Active Appearance Models, Supervised Descent Method and Constrained Local Models.