| dc.contributor | CARLOS ALBERTO REYES GARCIA | |
| dc.contributor | Hayde Peregrina Barreto | |
| dc.creator | Eduardo Morales-Vargas | |
| dc.date | 2017-07 | |
| dc.date.accessioned | 2023-07-25T16:25:34Z | |
| dc.date.available | 2023-07-25T16:25:34Z | |
| dc.identifier | http://inaoe.repositorioinstitucional.mx/jspui/handle/1009/2414 | |
| dc.identifier.uri | https://repositorioslatinoamericanos.uchile.cl/handle/2250/7807590 | |
| dc.description | Facial expression recognition is related to the automatic identification of the overt manifestation of affective states of a subject by computational means and has applications in security, human computer interaction among others. This work focuses on the design of a model for the recognition of the seven basic facial expressions: anger, contempt, disgust, fear, happiness, sadness and surprise. Facial expressions description in terms of action units is used as depart point. Fuzzy models are used in order to maintain a relation between the facial muscle appearance and the fuzzily associated facial expressions. The proposed method, model facial expressions using granular fuzzy models, finding automatically fuzzy rules that can describe the output class with a low number of rules. The reason of use a model with a low number of rules lies in the necessity of a simple model that do not loose the ability to explain why is making a decision. First, heuristic guided affine transformations align facial landmarks of the neutral and the target expression. Second, features are extracted describing face movements in terms of changes in orientation (angle and magnitude) of distinctive facial areas. Third, the full featured representation is embedded into a compact one by means of pooling. Finally, a Sugeno-type adaptive Neuro Fuzzy Inference System is used for each action unit to generate a description of the movements in the face that identifies the facial expression present in an image sequence. For evaluating the method the CK+ database is used, it contains 327 labeled frontal image sequences from 123 healthy subjects in which one of the seven basic facial expressions is represented. Each sequence begins with a subject in a neutrally affective state and ends with a facial expression. The proposed model discriminates facial expressions with mean accuracy of 89.04±0.91% with a maximum accuracy of 91.41±28%. Further, distinctly to current solutions the model can also describe why is reaching such decision. | |
| dc.format | application/pdf | |
| dc.language | eng | |
| dc.publisher | Instituto Nacional de Astrofísica, Óptica y Electrónica | |
| dc.relation | citation:Morales Vargas, Eduardo, (2017), Granular fuzzy model with hyperboxes for facial expression recognition, Tesis de Maestría, Instituto Nacional de Astrofísica, Óptica y Electrónica. | |
| dc.rights | info:eu-repo/semantics/openAccess | |
| dc.rights | http://creativecommons.org/licenses/by-nc-nd/4.0 | |
| dc.subject | info:eu-repo/classification/Reconocimiento de expresiones faciales/Facial expression recognition | |
| dc.subject | info:eu-repo/classification/Modelos difusos explicables/Fuzzy explainable models | |
| dc.subject | info:eu-repo/classification/Sistema de codificación de acciones faciales/Facial action coding system | |
| dc.subject | info:eu-repo/classification/cti/1 | |
| dc.subject | info:eu-repo/classification/cti/12 | |
| dc.subject | info:eu-repo/classification/cti/1203 | |
| dc.subject | info:eu-repo/classification/cti/120304 | |
| dc.subject | info:eu-repo/classification/cti/120304 | |
| dc.title | Granular fuzzy model with hyperboxes for facial expression recognition | |
| dc.type | info:eu-repo/semantics/masterThesis | |
| dc.type | info:eu-repo/semantics/acceptedVersion | |
| dc.audience | students | |
| dc.audience | researchers | |
| dc.audience | generalPublic | |
