Ciencias Exactas y Ciencias de la Salud
Permanent URI for this collectionhttps://hdl.handle.net/11285/551014
Pertenecen a esta colección Tesis y Trabajos de grado de los Doctorados correspondientes a las Escuelas de Ingeniería y Ciencias así como a Medicina y Ciencias de la Salud.
Browse
Search Results
- A Comprehensive study into digital human faces. emphatic perception using an improved machine vision model for a democratized facial tracking using a facial action code system(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-05-30) Vilchis Zapata, Carlos Leonel; González Mendoza, Miguel; emipsanchez; Rudomín Goldberg, Isaac J.; Ochoa Ruiz, Gilberto; Navarro Tuch, Sergio; Escuela de Ingenierias y Ciencias; Campus Monterrey; Chang Fernández, LeonardoThis thesis explores the flourishing field of digital humans, which has gained prominence in recent years as embodied conversational agents with diverse applications. The challenge lies in achieving realism in graphics, empathic responses, and accurate facial expression replication. While computer facial animation has made strides in creating adaptable, real-time systems, democratized solutions for facial motion capture are limited by cost and accessibility. This research presents a framework integrating artificial intelligence techniques for real-time facial tracking. It is hypothesized that by combining the Facial Action Coding System with machine learning, digital human realism and empathic responses can be enhanced. Key questions and objectives address the feasibility of open-source real-time facial tracking, the impact of integration between the Facial Action Coding System and artificial intelligence, and the balance between photo-realistic quality and expressive nuances. Contributions encompass a general facial capture pipeline proposal, an open-source application, an evaluation model for empathic responses, and a comparative analysis of accessible facial performance capture solutions. Parallel research findings include a protocol for genuine expression capture, insights into emotional regulation in virtual environments, and an evaluation of lightweight backbone models for facial reconstruction. Overall, the research carried out during this thesis holds the potential for improving realism and empathy in digital humans, offering valuable insights and setting the stage for future advancements in the field.
- Enhancing video-based human action recognition: leveraging knowledge distillation for improved training efficiency and flexibility(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-05) Camarena Trinidad, Luis Fernando; González Mendoza, Miguel; 123361; https://orcid.org/0000-0001-6451-9109; emipsanchez; Ochoa Ruiz, Gilberto; Pérez Suárez, Ariel; Marín Hernández, Antonio; School of Engineering and Science; Campus Estado de México; Chang Fernández, LeonardoArtificial Intelligence (AI) stands out for its transformative potential, revolutionizing sectors from healthcare and transport to e-commerce and industrial maintenance. A core task of AI applications is to be able to understand human behavior in videos, which is the foundation in areas like surveillance, content monitoring, patient care, and gaming. Training a model to recognize human actions implies a highly complex computational process in which modern strategies use a knowledge transfer approach to reduce computational complexity. However, they come with challenges, especially in flexibility and efficiency. Existing solutions are limited in functionality, relying heavily on pretrained model architectures, which can restrict their applicability in diverse scenarios. Our research, titled ”Enhancing Video-Based Human Action Recognition: Leverag- ing Knowledge Distillation for Improved Training Efficiency and Flexibility”, proposes a framework that uses knowledge distillation (KD) to guide the training of self-supervised models. This framework has significant practical implications, as it improves classification accuracy, accelerates model convergence, and increases model flexibility under regular and limited data scenarios. We tested our method on the UCF101 dataset, varying the balanced proportions from 100 % to 2 %, and measured their performance at different training stages. Our results show that our approach outperforms traditional training methods, maintaining classification accuracy while improving the convergence rate. In addition to the efficiency of the model training, our methods enable cross-architecture adaptability, allowing model customization for various applications. In data-scarce environments, KD maintains its robustness, proving invaluable for applications where gathering extensive labeled data is challenging or expensive.
- Closing the gap on affordable real-time very low resolution face recognition for automated video surveillance(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2022-12-05) Luévano García, Luis Santiago; González Mendoza, Miguel; puemcuervo; Ochoa Ruiz, Gilberto; Méndez Vázquez, Heydi; Martínez Díaz, Yoanna; School of Engineering and Sciences; Campus Estado de México; Chang Fernández, LeonardoPublic and private security is a worldwide problem where efficient and automated video-surveillance technologies have a lot of potential. In an emerging country like Mexico, a functional real-time automated video-surveillance system will have a very positive social, economic, and technological impact. Proposing an open framework for face recognition at very low resolutions which public and private institutions could implement and take advantage of, will ultimately benefit our society and contribute to the state of the art in terms of efficacy and efficiency. Currently, efficient face recognition for automated video surveillance is not present within the reach of public institutions and much less so for the smallest business establishments, such as convenience stores and small offices. To make an impact in this area, the scientific problem that we are focusing on solving is the one of effectively and efficiently extracting robust facial features from Very Low Resolution face images from surveillance footage, to perform the appropriate subspace projection, and perform the posterior face identification using a dataset reference, in order to improve in efficiency terms. In this thesis, we propose solving this problem using our novel method, BinaryFaceNet, with state-of-the-art training methodology and advancements in the Binary Neural Network (BNN) and Lightweight Convolutional Neural Network (CNN) literature. The implementation of our method makes accurate and real-time face recognition available for affordable ARM-based embedded devices, with limited identification and verification performance penalties while achieving an inference performance of less than 90\% latency against state-of-the-art BNNs. We finally discuss the feasibility of implementing BNN technology on extremely limited hardware, the compromises made to achieve maximum efficiency, training stable ultra-compact binarized models, and provide future work directions to complement this proposal. Finally, in our concluding remarks, we summarize the research work done and the research outcomes during the tenure of this thesis project.
- Formación automática de ontologías de conceptos encontrados en documentos no estructurados(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2022-07-02) Rivera González, Martha Idalid; González Mendoza, Miguel; emimmayorquin; Cuevas Arenas, Alma; Gibrán Cansino, Héctor; Luévano García, Luis; Escuela de Ingeniería y Ciencias; Campus Estado de México; Guzmán, AdolfoEn las últimas décadas han sido desarrollados numerosos enfoques, métodos y técnicas para el desarrollo en la búsqueda del nuevo conocimiento, las ontologías es una prueba de ello, que permiten estructurar los datos, donde la información es almacenada en nodos y relaciones, agregando con ello, la aplicación de métodos, que permite realizar búsqueda de conocimientos de información no estructurada, tales como páginas web, documentos, artículos, palabras entre otros. El presente trabajo lleva a cabo una búsqueda en una gran cantidad de documentos, iniciando con un tema o concepto “semilla”, usando técnicas de minería de datos (minería de texto) se detectan palabras y frases temáticas que son candidatos a ser conceptos (se usará desambiguación para pasarlos a conceptos aunque generalmente en campos especializados, cada palabra o frase temática tiene un significado único). Las relaciones entre algunos de los conceptos se hallarán mediante la creación de un modelo que permitirá la formación del nuevo conocimiento para obtener las ontologías. La tarea de generar nuevo conocimiento no es nada trivial ya que se requiere de la elaboración de dos algoritmos el primero llamado AS-WT que da como resultado un conjunto de 6 palabras conjuntas, lematizadas y categorizadas y el segundo el AS-WT-ON da como resultado triadas compuestas por sustantivo-verbo-sustantivo para dar como resultado la generación de ontologías