Menpo (profile) Results

This jupyter notebook summarises the results of the active appearance models which were trained using the menpo Python Package. For more information regarding the training of these models please refer to:

Section 4 of the Review Paper
The Active Appearance Model base class contained within aam.py
The Menpo specific subclass of the base menpoAAM.py
The holisitc AAM and patch AAM models described within menpo_aam_profile_hol.py and menpo_aam_profile_patch.py

This notebook will contained the cumulative error distribution curves produced by the models as well as a selection from literature.

Resource Utilisation

Model	Time (hh:mm:ss)	Avg # CPUs Used	Memory Used (GB)
hAAM	00:28:56	4.76	5.4
pAAM	00:36:44	6.09	5.4

Test Set

A brief description of the test set used during the experiment. Note that the same test set was used for both holistic and patch experiments.



In [1]:

    
import os
with open('menpo_aam_profile_hol_testset.txt', 'r') as f:
    test_set_filenames = f.read().split('\n')

print("The test set comprised of %i images\n" % len(test_set_filenames))
print("Listing the first 10 images:")

# Print summary
test_set_filenames = [os.path.basename(filename) for filename in test_set_filenames]
print("\t".join(test_set_filenames[:10]) + "...")









    



The test set comprised of 690 images

Listing the first 10 images:
aflw__face_43527.jpg	aflw__face_47700.jpg	aflw__face_41359.jpg	aflw__face_46016.jpg	aflw__face_41722.jpg	aflw__face_42627.jpg	aflw__face_44132.jpg	aflw__face_47664.jpg	aflw__face_40758.jpg	aflw__face_49788.jpg...

Cumulative Error Distribution Curves

The following chart shows the normalised error distribution as produced by the holisitic and patch AAMs vs that described by Chen et al. in 2017¹, He et al. in 2017² and Zadeh et al. in 2017³

Chen, X., Zhou, E., Mo, Y., Liu, J., & Cao, Z. (2017). Delving Deep Into Coarse-To-Fine Framework for Facial Landmark Localization. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops.
He, Z., Kan, M., Zhang, J., Chen, X., & Shan, S. (2017). A Fully End-to-End Cascaded CNN for Facial Landmark Detection. In 2017 12th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2017) (pp. 200–207). IEEE. http://doi.org/10.1109/FG.2017.33
Zadeh, A., Baltrusaitis, T., & Morency, L.-P. (2017). Convolutional Experts Constrained Local Model for Facial Landmark Detection. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops.



In [9]:

    
import numpy as np

# Import and configure matplotlib
import matplotlib
%matplotlib inline
matplotlib.rcParams['figure.figsize'] = (20, 10)
matplotlib.rcParams['font.size'] = 14
import matplotlib.pyplot as plt


# These files contains the CED data from experimentation & literature
FILES = ['menpo_aam_profile_patch.npy', 'menpo_aam_profile_hol.npy', 'menpo_profile_chen_2017.npy',
         'menpo_profile_he_2017.npy', 'menpo_profile_zadeh_2017.npy']
LABELS = ['pAAM', 'hAAM', 'Chen et al. 2017', 'He et al. 2017',
          'Zadeh et al. 2017']
COLOURS = ['b', 'g', 'r', 'c', 'y']

fig, (ax1, ax2) = plt.subplots(1, 2, sharey=True)

for filename, label, col in zip(FILES, LABELS, COLOURS):
    
    # Load the data
    data = np.load(filename)
    
    # Normalise
    data[1] /= np.max(data[1])
    
    # Plot hAAM on second axis
    if label == 'hAAM':
        ax = ax2
    else:
        ax = ax1
    # Plot
    ax.plot(data[0], data[1], label=label, c=col)


# Add details to plot
ax1.legend(loc='lower right', prop={'size': 13});
ax2.legend(loc='lower right', prop={'size': 13});
ax1.grid();
ax2.grid();
ax1.set_xlabel('Point-to-point error normalised by face diagonal');
ax1.set_ylabel('Proportion of Images');
ax2.set_xlabel('Point-to-point error normalised by face diagonal (hAAM)');
plt.tight_layout()