OpenFace/matlab_version/experiments_menpo

Scripts for evaluating CE-CLM, and CLNF methods on the Menpo dataset - https://ibug.doc.ic.ac.uk/resources/2nd-facial-landmark-tracking-competition-menpo-ben/

To run the models on the data using models trained on 300W + MultiPIE on all Menpo training data:
- Script_CECLM_cross_data.m
- Script_CLNF_cross_data_general.m

To run CE-CLM model trained on 300W + MultiPIE + Menpo on menpo validation data:
- Script CECLM_menpo_valid.m

To visualize the results of the models against baselines (using inner face markup - 49, and full face markup - 68):
Display_ceclm_results_49.m
Display_ceclm_results_68.m

To construct the error table:
Construct_error_table.m

For a fair comparison all the models and baselines are initialized using the bounding boxes generated by using an MTCNN face detector (https://github.com/kpzhang93/MTCNN_face_detection_alignment).

Scripts to reproduce the CE-CLM submission to the Menpo challenge:
Script_CECLM_menpo_test_frontal.m
Script_CECLM_menpo_test_profile.m