Please use this identifier to cite or link to this item:
Title: Towards facial expression robustness in multi-scale wild environments
Authors: Freire Obregón, David Sebastián 
Hernández Sosa, José Daniel 
Santana Jaria, Oliverio Jesús 
Lorenzo Navarro, José Javier 
Castrillón Santana, Modesto Fernando 
UNESCO Clasification: 120304 Inteligencia artificial
Keywords: Facial expressions
Multi-scale resolution
Sequence-based approach
Issue Date: 2023
Publisher: Springer 
Project: Interaccióny Re-Identificación de Personas Mediante Machine Learning, Deep Learningy Análisis de Datos Multimodal: Hacia Una Comunicación Más Natural en la Robótica Social 
Journal: Lecture Notes in Computer Science 
Conference: 22th International Conference on Image Analysis and Processing (ICIAP 2023)
Abstract: Facial expressions are dynamic processes that evolve over temporal segments, including onset, apex, offset, and neutral. However, previous works on automatic facial expression analysis have mainly focused on the recognition of discrete emotions, neglecting the continuous nature of these processes. Additionally, facial images captured from videos in the wild often have varying resolutions due to fixed-lens cameras. To address these problems, our objective is to develop a robust facial expression recognition classifier that provides good performance in such challenging environments. We evaluated several state-of-the-art models on labeled and unlabeled collections and analyzed their performance at different scales. To improve performance, we filtered the probabilities provided by each classifier and demonstrated that this improves decision-making consistency by more than 10%, leading to accuracy improvement. Finally, we combined the models’ backbones into a temporal-sequence classifier, leveraging this consistency-performance trade-off and achieving an additional improvement of 9.6%.
ISBN: 978-3-031-43147-0
ISSN: 0302-9743
DOI: 10.1007/978-3-031-43148-7_16
Source: ICIAP 2023: Image Analysis and Processing. Lecture Notes in Computer Science, [ISSN 0302-9743], vol 14233, p. 184–195 (September,2023)
Appears in Collections:Actas de congresos
Show full item record

Google ScholarTM




Export metadata

Items in accedaCRIS are protected by copyright, with all rights reserved, unless otherwise indicated.