Daniel Cervantes Ambriz

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Research Subjects

Research Interests

Daniel Cervantes Ambriz has a total of 20 co-authors. Here are the top co-authors by number of shared publications:

Everardo Efrén Granda Gutiérrez
Universidad Autónoma del Estado de México, Centro Universitario UAEM Atlacomulco
3 shared publications
Roberto Alejo
Division of Graduate Studies and Research, National Technological of Mexico, Campus Toluca, Metepec 52149, Mexico
3 shared publications
Carlos Mauricio Castorena Lara
1 shared publication
Daniel Villanueva-Vásquez
Division of Graduate Studies and Research, National Technological of Mexico, Campus Toluca, Metepec 52149, Mexico
1 shared publication
Eréndira Rendón
1 shared publication
Everardo E. Granda-Gutiérrez
1 shared publication

Latest Publications

Sort by

Open Access
Website
Google Scholar
- 1 November 2026
- Published by Dirección de Bienes Públicos del Conocimiento, Universidad Autonoma del Estado de Mexico in Ideas en Ciencias de la Ingeniería
Optimización Factorial de Redes Neuronales Convolucionales para la Detección de Pulgones (Aphidinae) en Cultivos de RosaL
by José Luis Hernández, Eréndira Rendón, Roberto Alejo, Federico Del Razo, Everardo Efren Granda, Daniel Cervantes Ambriz
La implementación de redes neuronales convolucionales en la agricultura de precisión enfrenta el desafío de la configuración empírica de sus hiperparámetros, lo que deriva en un uso ineficiente de recursos computacionales y en tiempos de entrenamiento prolongados. Para abordar esta limitación, el presente estudio propone una metodología basada en un diseño experimental factorial de tres factores para optimizar el rendimiento de las arquitecturas VGG16 y EfficientNetB0 en la clasificación de pulgones (Aphidinae) en cultivos de rosal. Para ello, se empleó un conjunto de datos conformado por 1438 imágenes, particionado en 1222 instancias para la fase de entrenamiento y 216 para el conjunto de prueba. El entrenamiento de las arquitecturas se ejecutó mediante niveles fijos de épocas, prescindiendo deliberadamente de mecanismos de detención temprana con el propósito de cuantificar de manera aislada el impacto de la convergencia dentro del diseño factorial. Se evaluaron el tamaño de lote, el número de épocas y la función de activación, analizando tanto sus efectos principales como sus interacciones sobre la exactitud del modelo. Los resultados del análisis de varianza mostraron que la función de activación es el factor con mayor influencia estadística en ambas arquitecturas. Como hallazgo principal, EfficientNetB0 superó a VGG16, alcanzando una Accuracy del 94.42% bajo la configuración óptima de lote de 32, 30 épocas y activación softmax. Este enfoque demuestra que la aplicación de diseños factoriales elimina la incertidumbre del ajuste por prueba y error, lo que favorece modelos más robustos, reproducibles y escalables para el monitoreo de plagas. En esta investigación se han evaluado dos arquitecturas convolucionales representativas; no obstante, esta metodología es extensible a otras topologías con parámetros análogos y a diversas aplicaciones.
Website
Google Scholar
- 1 November 2026
- Published by Dirección de Bienes Públicos del Conocimiento, Universidad Autonoma del Estado de Mexico in Ideas en Ciencias de la Ingeniería
Open Access
Optimización Factorial de Redes Neuronales Convolucionales para la Detección de Pulgones (Aphidinae) en Cultivos de RosaL
by José Luis Hernández, Eréndira Rendón, Roberto Alejo, Federico Del Razo, Everardo Efren Granda, Daniel Cervantes Ambriz
La implementación de redes neuronales convolucionales en la agricultura de precisión enfrenta el desafío de la configuración empírica de sus hiperparámetros, lo que deriva en un uso ineficiente de recursos computacionales y en tiempos de entrenamiento prolongados. Para abordar esta limitación, el presente estudio propone una metodología basada en un diseño experimental factorial de tres factores para optimizar el rendimiento de las arquitecturas VGG16 y EfficientNetB0 en la clasificación de pulgones (Aphidinae) en cultivos de rosal. Para ello, se empleó un conjunto de datos conformado por 1438 imágenes, particionado en 1222 instancias para la fase de entrenamiento y 216 para el conjunto de prueba. El entrenamiento de las arquitecturas se ejecutó mediante niveles fijos de épocas, prescindiendo deliberadamente de mecanismos de detención temprana con el propósito de cuantificar de manera aislada el impacto de la convergencia dentro del diseño factorial. Se evaluaron el tamaño de lote, el número de épocas y la función de activación, analizando tanto sus efectos principales como sus interacciones sobre la exactitud del modelo. Los resultados del análisis de varianza mostraron que la función de activación es el factor con mayor influencia estadística en ambas arquitecturas. Como hallazgo principal, EfficientNetB0 superó a VGG16, alcanzando una Accuracy del 94.42% bajo la configuración óptima de lote de 32, 30 épocas y activación softmax. Este enfoque demuestra que la aplicación de diseños factoriales elimina la incertidumbre del ajuste por prueba y error, lo que favorece modelos más robustos, reproducibles y escalables para el monitoreo de plagas. En esta investigación se han evaluado dos arquitecturas convolucionales representativas; no obstante, esta metodología es extensible a otras topologías con parámetros análogos y a diversas aplicaciones.
Open Access
Website
Google Scholar
- 3 July 2026
- Published by MDPI AG in Mathematics
Empirical Review of Traditional and Recent Balancing Techniques for Image-Based Physical Violence Classification Across Diverse Imbalance Scenarios and Multiple Datasets
by Daniel Cervantes Ambriz, Daniel Villanueva Vásquez, Federico del Razo López, Everardo Efrén Granda Gutiérrez, Vicente García Jiménez, Roberto Alejo Eleuterio
The automated classification of physical violence in images faces a critical methodological obstacle: class imbalance, which undermines the discriminative capacity of Deep Learning (DL) models by leading them to overfit the majority class. Although numerous balancing strategies have been proposed in the literature, which range from traditional resampling to advanced generative models, a systematic and controlled empirical evaluation of their behavior in the specific domain of physical violence detection remains absent. In this study, we present a systematic empirical review that evaluates traditional balancing techniques, generative models, and data augmentation methods under controlled imbalance conditions. The methodology is applied to five datasets (UBI-FIGHTS, RLVSD, RWF-2000, RLVS, and AIRTLAB), across six imbalance scenarios defined by Balance Percentage levels of 1%, 5%, 10%, 25%, 50%, and 75%. To isolate the effect of the balancing strategies from architectural confounders, a single stable backbone (ResNet-18) is employed as a controlled classifier across all experimental conditions. The experimental results demonstrate the absence of a universally optimal balancing technique: in highly imbalanced scenarios, traditional oversampling methods such as ROS achieved the best average performance, while generative approaches such as DCGAN and DeepSMOTE became increasingly competitive as class balance is improved. These findings confirm that the effectiveness of balancing techniques depends on both the degree of asymmetry and the dataset’s intrinsic characteristics. Thus, it is clear that there is a need for context-aware strategy selection rather than one-size-fits-all solutions. Beyond the empirical findings, this work provides a structured synthesis of the theoretical foundations and state-of-the-art methods for imbalanced violence detection.
Website
Google Scholar
- 3 July 2026
- Published by MDPI AG in Mathematics
Open Access
Empirical Review of Traditional and Recent Balancing Techniques for Image-Based Physical Violence Classification Across Diverse Imbalance Scenarios and Multiple Datasets
by Daniel Cervantes Ambriz, Daniel Villanueva Vásquez, Federico del Razo López, Everardo Efrén Granda Gutiérrez, Vicente García Jiménez, Roberto Alejo Eleuterio
The automated classification of physical violence in images faces a critical methodological obstacle: class imbalance, which undermines the discriminative capacity of Deep Learning (DL) models by leading them to overfit the majority class. Although numerous balancing strategies have been proposed in the literature, which range from traditional resampling to advanced generative models, a systematic and controlled empirical evaluation of their behavior in the specific domain of physical violence detection remains absent. In this study, we present a systematic empirical review that evaluates traditional balancing techniques, generative models, and data augmentation methods under controlled imbalance conditions. The methodology is applied to five datasets (UBI-FIGHTS, RLVSD, RWF-2000, RLVS, and AIRTLAB), across six imbalance scenarios defined by Balance Percentage levels of 1%, 5%, 10%, 25%, 50%, and 75%. To isolate the effect of the balancing strategies from architectural confounders, a single stable backbone (ResNet-18) is employed as a controlled classifier across all experimental conditions. The experimental results demonstrate the absence of a universally optimal balancing technique: in highly imbalanced scenarios, traditional oversampling methods such as ROS achieved the best average performance, while generative approaches such as DCGAN and DeepSMOTE became increasingly competitive as class balance is improved. These findings confirm that the effectiveness of balancing techniques depends on both the degree of asymmetry and the dataset’s intrinsic characteristics. Thus, it is clear that there is a need for context-aware strategy selection rather than one-size-fits-all solutions. Beyond the empirical findings, this work provides a structured synthesis of the theoretical foundations and state-of-the-art methods for imbalanced violence detection.
Open Access
Website
Google Scholar
- 15 September 2025
- Published by MDPI AG in Symmetry
Strategic Sample Selection in Deep Learning: A Case Study on Violence Detection Using Confidence-Based Subsets
by Francisco Primero Primero, Daniel Cervantes Ambriz, Roberto Alejo Eleuterio, Everardo E. Granda Gutiérrez, Jorge Sánchez Jaime, Rosa M. Valdovinos Rosas
Automated violence detection in images presents a technical and scientific challenge that demands specialized methods to enhance classification systems. This study introduces an approach for automatically identifying relevant samples to improve the performance of neural network models, specifically DenseNet121, with a focus on violence classification in images. The proposed methodology begins with an initial training phase using a balanced dataset (DS1, 6000 images). Based on the model’s output scores ( $o u t_{N}$ ), three confidence levels are defined: Safe ( $o u t_{N} \geq 0.9 + σ$ or $o u t_{N} \leq 0.1 - σ$ ), Border ( $0.5 - σ \leq o u t_{N} \leq 0.5 + σ$ ), and Average ( $0.4 - σ \leq o u t_{N} \leq 0.6 + σ$ ). These levels correspond to scenarios with low, moderate, and high prediction error probabilities, respectively, where $σ$ is an adjustable threshold. The Border subset exhibits symmetry around the decision boundary ( $o u t_{N} = 0.5$ ), capturing maximally uncertain samples, while the Safe regions reflect functional asymmetries in high-confidence predictions. Subsequently, these thresholds are applied to a second dataset (DS2, 5600 images) to extract specialized subsets for retraining ( $D S_{S a f e}$ , $D S_{B o r d e r}$ , and $D S_{A v e r a g e}$ ). Finally, the model is evaluated using an independent test set ( $D S_{t e s t}$ , 4400 images), ensuring complete data isolation. The experimental results demonstrate that the confidence-based subsets offer competitive performance despite using significantly fewer samples. The Average subset achieved an F1-Score of 0.89 and a g-mean of 0.93 using only 20% of the data, making it a promising alternative for efficient training. These findings highlight that strategic sample selection based on confidence thresholds enables effective training with reduced data, offering a practical balance between performance and efficiency when symmetric uncertainty modeling is exploited.
Website
Google Scholar
- 15 September 2025
- Published by MDPI AG in Symmetry
Open Access
Strategic Sample Selection in Deep Learning: A Case Study on Violence Detection Using Confidence-Based Subsets
by Francisco Primero Primero, Daniel Cervantes Ambriz, Roberto Alejo Eleuterio, Everardo E. Granda Gutiérrez, Jorge Sánchez Jaime, Rosa M. Valdovinos Rosas
Automated violence detection in images presents a technical and scientific challenge that demands specialized methods to enhance classification systems. This study introduces an approach for automatically identifying relevant samples to improve the performance of neural network models, specifically DenseNet121, with a focus on violence classification in images. The proposed methodology begins with an initial training phase using a balanced dataset (DS1, 6000 images). Based on the model’s output scores ( $o u t_{N}$ ), three confidence levels are defined: Safe ( $o u t_{N} \geq 0.9 + σ$ or $o u t_{N} \leq 0.1 - σ$ ), Border ( $0.5 - σ \leq o u t_{N} \leq 0.5 + σ$ ), and Average ( $0.4 - σ \leq o u t_{N} \leq 0.6 + σ$ ). These levels correspond to scenarios with low, moderate, and high prediction error probabilities, respectively, where $σ$ is an adjustable threshold. The Border subset exhibits symmetry around the decision boundary ( $o u t_{N} = 0.5$ ), capturing maximally uncertain samples, while the Safe regions reflect functional asymmetries in high-confidence predictions. Subsequently, these thresholds are applied to a second dataset (DS2, 5600 images) to extract specialized subsets for retraining ( $D S_{S a f e}$ , $D S_{B o r d e r}$ , and $D S_{A v e r a g e}$ ). Finally, the model is evaluated using an independent test set ( $D S_{t e s t}$ , 4400 images), ensuring complete data isolation. The experimental results demonstrate that the confidence-based subsets offer competitive performance despite using significantly fewer samples. The Average subset achieved an F1-Score of 0.89 and a g-mean of 0.93 using only 20% of the data, making it a promising alternative for efficient training. These findings highlight that strategic sample selection based on confidence thresholds enables effective training with reduced data, offering a practical balance between performance and efficiency when symmetric uncertainty modeling is exploited.
Website
Google Scholar
- 14 July 2025
Empirical study of editing sampling on deep learning hidden layers space to classify imbalanced hyperspectral remote sensing images
by Daniel Cervantes Ambriz, Everardo E. Granda-Gutiérrez, Carlos Mauricio Castorena Lara, Roberto Alejo Eleuterio, Everardo Efrén Granda Gutiérrez, Federico Razo López, Vicente García Jiménez
Website
Google Scholar
- 14 July 2025
Empirical study of editing sampling on deep learning hidden layers space to classify imbalanced hyperspectral remote sensing images
by Daniel Cervantes Ambriz, Everardo E. Granda-Gutiérrez, Carlos Mauricio Castorena Lara, Roberto Alejo Eleuterio, Everardo Efrén Granda Gutiérrez, Federico Razo López, Vicente García Jiménez
Open Access
Website
Google Scholar
- 31 December 2024
- Published by Consejo Nacional de Rectores in Nexo Revista Científica
Detección automática de rostro con mascarilla para evitar contagio por COVID, mediante algoritmo LBP embebido en Rasp-Berry Pi y uso de alerta audible
by Juan Alberto Antonio Velázquez, Leopoldo Gil Antonio, Héctor Caballero Hernández, Daniel Cervantes Ambriz
The detection of the faces of people in self-access systems, is a challenging task since the pandemic of the coronavirus (SARS-CoV2) has changed our way of living, especially because companies and some government entities such as hospitals need to grant permission to access their facilities especially if the person does not bring the mask, and although many facilities no longer allow its use, since the above mentioned entities if they justify its use especially if their facilities there are people with chronic diseases that need and justify the use of the mask. In this work, we mention an access system that consists of software embedded in a Rasp-berry PI device, which contains an application formed by the use of a neural network Multi-task cascade convolutional networks (MTCNN), with which the training of the images of faces that bring and do not bring mouthguards is done, Later, and for a better identification of the faces, an algorithm based on Local Binary Patterns Histogram (LBPH) is used, with which the characteristics of the face are obtained and later it is possible to classify if the person to be identified has or does not have a face mask. In the end, our proposed system has an average accuracy of 93% in the detection of faces with face masks.
Website
Google Scholar
- 31 December 2024
- Published by Consejo Nacional de Rectores in Nexo Revista Científica
Open Access
Detección automática de rostro con mascarilla para evitar contagio por COVID, mediante algoritmo LBP embebido en Rasp-Berry Pi y uso de alerta audible
by Juan Alberto Antonio Velázquez, Leopoldo Gil Antonio, Héctor Caballero Hernández, Daniel Cervantes Ambriz
The detection of the faces of people in self-access systems, is a challenging task since the pandemic of the coronavirus (SARS-CoV2) has changed our way of living, especially because companies and some government entities such as hospitals need to grant permission to access their facilities especially if the person does not bring the mask, and although many facilities no longer allow its use, since the above mentioned entities if they justify its use especially if their facilities there are people with chronic diseases that need and justify the use of the mask. In this work, we mention an access system that consists of software embedded in a Rasp-berry PI device, which contains an application formed by the use of a neural network Multi-task cascade convolutional networks (MTCNN), with which the training of the images of faces that bring and do not bring mouthguards is done, Later, and for a better identification of the faces, an algorithm based on Local Binary Patterns Histogram (LBPH) is used, with which the characteristics of the face are obtained and later it is possible to classify if the person to be identified has or does not have a face mask. In the end, our proposed system has an average accuracy of 93% in the detection of faces with face masks.