Rafael Luiz Testa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Rafael Luiz Testa has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Ariane Machado-Lima
University of São Paulo
4 shared publications - Fátima Lourdes Santos Nunes
Each - USP
3 shared publications Fátima L. S. Nunes
School of Arts, Sciences and Humanities of the University of São Paulo, São Paulo, Brazil
2 shared publicationsAntonio Henrique Nunes Muniz
1 shared publicationAriane Machado Lima
1 shared publicationAriane Machado-Lima
School of Arts, Sciences and Humanities of the University of São Paulo, São Paulo, Brazil
1 shared publication
Latest Publications
- Open Access
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 170876-170894
Enhancing Temporal Coherence in Image-to-Video Facial Expression Synthesis: A Dual-Loss Framework for Smoother Generation
Facial expression synthesis for video sequences presents significant challenges in terms of maintaining temporal coherence while preserving identity and expression accuracy. Most current approaches primarily focus on individual frame quality. This results in videos with unnatural transitions, flickering artifacts, and inconsistent expression dynamics. This paper presents a novel method for improving temporal coherence in facial expression image-to-video synthesis. The proposed approach uses a dual-focus framework that explicitly models both pixel-level transitions and geometric consistency between consecutive frames. Our method includes two specialized loss functions: a frame difference loss and a landmark consistency loss. We also implement a two-stage training strategy and a frame-blending post-processing technique. Extensive testing on the MUG Facial Expression Database demonstrates that the enhanced-coherence approach outperforms baseline methods and competes with other approaches from the literature. The proposed method has the lowest Average Content Distance score of all the approaches tested. This indicates smoother frame transitions while maintaining high visual quality. Subjective evaluation with 140 participants confirms that the synthesized expressions are more realistic and expressive. Our approach outperformed real videos in recognition rates for certain emotion expressions, particularly disgust and sadness. These findings demonstrate the effectiveness of explicitly modeling temporal coherence in generating realistic facial expression videos. This has promising applications in therapy, entertainment, and human-computer interaction.
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 170876-170894
Open AccessEnhancing Temporal Coherence in Image-to-Video Facial Expression Synthesis: A Dual-Loss Framework for Smoother Generation
Facial expression synthesis for video sequences presents significant challenges in terms of maintaining temporal coherence while preserving identity and expression accuracy. Most current approaches primarily focus on individual frame quality. This results in videos with unnatural transitions, flickering artifacts, and inconsistent expression dynamics. This paper presents a novel method for improving temporal coherence in facial expression image-to-video synthesis. The proposed approach uses a dual-focus framework that explicitly models both pixel-level transitions and geometric consistency between consecutive frames. Our method includes two specialized loss functions: a frame difference loss and a landmark consistency loss. We also implement a two-stage training strategy and a frame-blending post-processing technique. Extensive testing on the MUG Facial Expression Database demonstrates that the enhanced-coherence approach outperforms baseline methods and competes with other approaches from the literature. The proposed method has the lowest Average Content Distance score of all the approaches tested. This indicates smoother frame transitions while maintaining high visual quality. Subjective evaluation with 140 participants confirms that the synthesized expressions are more realistic and expressive. Our approach outperformed real videos in recognition rates for certain emotion expressions, particularly disgust and sadness. These findings demonstrate the effectiveness of explicitly modeling temporal coherence in generating realistic facial expression videos. This has promising applications in therapy, entertainment, and human-computer interaction.
- 6 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 175-180
Does the Input Image Spatial Resolution Generate Different Synthetic Images? A Comparative Study of Facial Expression Synthesis Performance
Facial expression synthesis has gained significant attention in the image synthesis field. Generative Adversarial Network (GAN) models have recently gained popularity due to the high-quality synthetic images they produce. However, these models require complex network architectures that can take days to train, even with high-performance Graphics Processing Units (GPUs). Many efforts have been made to accelerate and compress such models, but little attention has been paid to the resolution of the images. This study aims to assess the impact of input/output spatial resolution on the resources needed for training a facial expression synthesis model, as well as on the quality of the results. Our results indicate that the produced images and videos had similar quality results measured through objective measures for the spatial resolution of 128 × 128, 256 × 256, and 480 × 480. Furthermore, we found that lower-resolution images could significantly reduce the time required to generate new facial expressions without compromising quality, as measured by objective measures.
- 6 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 175-180
Does the Input Image Spatial Resolution Generate Different Synthetic Images? A Comparative Study of Facial Expression Synthesis Performance
Facial expression synthesis has gained significant attention in the image synthesis field. Generative Adversarial Network (GAN) models have recently gained popularity due to the high-quality synthetic images they produce. However, these models require complex network architectures that can take days to train, even with high-performance Graphics Processing Units (GPUs). Many efforts have been made to accelerate and compress such models, but little attention has been paid to the resolution of the images. This study aims to assess the impact of input/output spatial resolution on the resources needed for training a facial expression synthesis model, as well as on the quality of the results. Our results indicate that the produced images and videos had similar quality results measured through objective measures for the spatial resolution of 128 × 128, 256 × 256, and 480 × 480. Furthermore, we found that lower-resolution images could significantly reduce the time required to generate new facial expressions without compromising quality, as measured by objective measures.
- 7 September 2021
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 80 (30) , 36465-36489
Facial expression synthesis based on similar faces
- 7 September 2021
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 80 (30) , 36465-36489
Facial expression synthesis based on similar faces
- Open Access
- 8 November 2020
- Published by Universidade Federal da Paraíba in Comunicações em Informática
- Vol. 4 (2) , 39-42
Virtual Reality, Augmented Reality and Beyond: Laboratory of Computer Applications for Health Care at University of São Paulo
The Laboratory of Computer Applications for Health Care at University of São Paulo develop multidisciplinary research involving Computer Science areas and other knowledge fields such as Physiotherapy, Psychiatry, Radiology, and Cardiology. Most studies use concepts besides Computer Science, by aggregating knowledge to solve real problems. In this paper we present the main projects in Virtual Reality and Augmented Reality areas, highlighting serious games, computer-aided diagnosis and therapy, simulation and training. Besides the social and scientific impact inherent to the type of the developed research, some systems are available to health professionals in order to allow technology transfer and consequent technological and economic impact in daily activities of health tasks.
- 8 November 2020
- Published by Universidade Federal da Paraíba in Comunicações em Informática
- Vol. 4 (2) , 39-42
Open AccessVirtual Reality, Augmented Reality and Beyond: Laboratory of Computer Applications for Health Care at University of São Paulo
The Laboratory of Computer Applications for Health Care at University of São Paulo develop multidisciplinary research involving Computer Science areas and other knowledge fields such as Physiotherapy, Psychiatry, Radiology, and Cardiology. Most studies use concepts besides Computer Science, by aggregating knowledge to solve real problems. In this paper we present the main projects in Virtual Reality and Augmented Reality areas, highlighting serious games, computer-aided diagnosis and therapy, simulation and training. Besides the social and scientific impact inherent to the type of the developed research, some systems are available to health professionals in order to allow technology transfer and consequent technological and economic impact in daily activities of health tasks.
- Open Access
- 28 January 2019
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 51 (6) , 1-35
Synthesis of Facial Expressions in Photographs
The synthesis of facial expressions has applicationsin areas such as interactive games, biometrics systems, and training of people with disorders, among others. Although this is an area relatively well explored in the literature, there are no recent studies proposing to systematize an overview of research in the area. This systematic review analyzes the approaches to the synthesis of facial expressions in photographs, as well as important aspects of the synthesis process, such as preprocessing techniques, databases, and evaluation metrics. Forty-eight studies from three different scientific databases were analyzed. From these studies, we established an overview of the process, including all the stages used to synthesize expressions in facial images. We also analyze important aspects involved in these stages such as methods and techniques of each stage, databases, and evaluation metrics. We observed that machine learning approaches are the most widely used to synthesize expressions. Landmark identification, deformation, mapping, fusion, and training are common tasks considered in the approaches. We also found that few studies used metrics to evaluate the results, and most studies used public databases. Although the studies analyzed generated consistent and realistic results while preserving the identity of the subject, there are still research themes to be exploited.
- 28 January 2019
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 51 (6) , 1-35
Open AccessSynthesis of Facial Expressions in Photographs
The synthesis of facial expressions has applicationsin areas such as interactive games, biometrics systems, and training of people with disorders, among others. Although this is an area relatively well explored in the literature, there are no recent studies proposing to systematize an overview of research in the area. This systematic review analyzes the approaches to the synthesis of facial expressions in photographs, as well as important aspects of the synthesis process, such as preprocessing techniques, databases, and evaluation metrics. Forty-eight studies from three different scientific databases were analyzed. From these studies, we established an overview of the process, including all the stages used to synthesize expressions in facial images. We also analyze important aspects involved in these stages such as methods and techniques of each stage, databases, and evaluation metrics. We observed that machine learning approaches are the most widely used to synthesize expressions. Landmark identification, deformation, mapping, fusion, and training are common tasks considered in the approaches. We also found that few studies used metrics to evaluate the results, and most studies used public databases. Although the studies analyzed generated consistent and realistic results while preserving the identity of the subject, there are still research themes to be exploited.
- 1 October 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 297-304
Factors Influencing the Perception of Realism in Synthetic Facial Expressions
One way to synthesize facial expressions is to change an image to represent the desired emotion and it is useful in entertainment, diagnostic and psychiatric disorders therapy applications. Despite several existing approaches, there is little discussion about factors that contribute or hinder the perception of realism in synthetic facial expressions images. After presenting an approach for facial expressions synthesis through the deformation of facial features, this paper provides an evaluation by 155 volunteers regarding the realism of synthesized images. The proposed facial expression synthesis aims to generate new images using two source images (neutral and expressive face) and changing the expression in a target image (neutral face). The results suggest that assignment of realism depends on the type of image (real or synthetic). However, the synthesis presents images that can be considered realistic, especially for the expression of happiness. Finally, while factors such as color difference between subsequent regions and unnatural-sized facial features contribute to less realism, other factors such as the presence of wrinkles contribute to a greater assignment of realism to images.
- 1 October 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 297-304
Factors Influencing the Perception of Realism in Synthetic Facial Expressions
One way to synthesize facial expressions is to change an image to represent the desired emotion and it is useful in entertainment, diagnostic and psychiatric disorders therapy applications. Despite several existing approaches, there is little discussion about factors that contribute or hinder the perception of realism in synthetic facial expressions images. After presenting an approach for facial expressions synthesis through the deformation of facial features, this paper provides an evaluation by 155 volunteers regarding the realism of synthesized images. The proposed facial expression synthesis aims to generate new images using two source images (neutral and expressive face) and changing the expression in a target image (neutral face). The results suggest that assignment of realism depends on the type of image (real or synthetic). However, the synthesis presents images that can be considered realistic, especially for the expression of happiness. Finally, while factors such as color difference between subsequent regions and unnatural-sized facial features contribute to less realism, other factors such as the presence of wrinkles contribute to a greater assignment of realism to images.
- 1 June 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 304-309
Generating Facial Emotions for Diagnosis and Training
The ability to process and identify facial emotions is an essential factor for an individuals social interaction. There are certain psychiatric disorders that can limit an individuals ability to recognize emotions in facial expressions. This problem could be confronted by making use of computational techniques in order to develop learning environments for the diagnosis, evaluation and training in identifying facial emotions. This paper presents an approach that uses image processing techniques, formal languages, anthropometry and Facial Action Coding System (FACS) to generate caricatures that represent facial movements related to neutral, satisfaction, sadness, anger, disgust, fear and surprise emotions. The rules that define the emotions were determined using an AND-OR graph to enable generating these images in a flexible manner. An evaluation conducted with healthy volunteers showed that some emotions are more easily recognized, while for other emotions the caricatures need to be further improved. This is a promising approach, since the parameters used provide flexibility to define the emotional intensity that must be represented.
- 1 June 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 304-309
Generating Facial Emotions for Diagnosis and Training
The ability to process and identify facial emotions is an essential factor for an individuals social interaction. There are certain psychiatric disorders that can limit an individuals ability to recognize emotions in facial expressions. This problem could be confronted by making use of computational techniques in order to develop learning environments for the diagnosis, evaluation and training in identifying facial emotions. This paper presents an approach that uses image processing techniques, formal languages, anthropometry and Facial Action Coding System (FACS) to generate caricatures that represent facial movements related to neutral, satisfaction, sadness, anger, disgust, fear and surprise emotions. The rules that define the emotions were determined using an AND-OR graph to enable generating these images in a flexible manner. An evaluation conducted with healthy volunteers showed that some emotions are more easily recognized, while for other emotions the caricatures need to be further improved. This is a promising approach, since the parameters used provide flexibility to define the emotional intensity that must be represented.
- 25 November 2013
- Published by Sociedade Brasileira de Computação
- Vol. 2 (1)
Sistema de Coleta de Dados para Avaliação da Aprendizagem em Ambientes Virtuais de Aprendizagem Tridimensionais
As tecnologias de Realidade Virtual e Realidade Aumentada favorecem o desenvolvimento de interfaces humano-computador mais interativas e imersivas. Tais interfaces têm aumentado o interesse dos pesquisadores em conceber Ambientes Virtuais de Aprendizagem Tridimensionais (AVAs 3D) em diferentes contextos, mas coletar dados para avaliar o aprendizado usando esses ambientes ainda é um desafio. Este artigo apresenta o Sistema de Avaliação de Aprendizagem (SAAp), que inclui instrumentos para avaliação diagnóstica e registro das interações do usuário no ambiente virtual. Ao analisar os dados coletados, o avaliador pode medir como e quanto os AVAs 3D colaboram com o processo de aquisição de conhecimento. Os resultados mostraram que o SAAp é eficiente para facilitar a coleta de dados e permite monitorar as interações que ocorrem automaticamente, sem interferir no código fonte da aplicação.
- 25 November 2013
- Published by Sociedade Brasileira de Computação
- Vol. 2 (1)
Sistema de Coleta de Dados para Avaliação da Aprendizagem em Ambientes Virtuais de Aprendizagem Tridimensionais
As tecnologias de Realidade Virtual e Realidade Aumentada favorecem o desenvolvimento de interfaces humano-computador mais interativas e imersivas. Tais interfaces têm aumentado o interesse dos pesquisadores em conceber Ambientes Virtuais de Aprendizagem Tridimensionais (AVAs 3D) em diferentes contextos, mas coletar dados para avaliar o aprendizado usando esses ambientes ainda é um desafio. Este artigo apresenta o Sistema de Avaliação de Aprendizagem (SAAp), que inclui instrumentos para avaliação diagnóstica e registro das interações do usuário no ambiente virtual. Ao analisar os dados coletados, o avaliador pode medir como e quanto os AVAs 3D colaboram com o processo de aquisição de conhecimento. Os resultados mostraram que o SAAp é eficiente para facilitar a coleta de dados e permite monitorar as interações que ocorrem automaticamente, sem interferir no código fonte da aplicação.