Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 6
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Using Diffusion Models for Data Augmentation on Limited Rodent OCT Datasets

  • García-Torres, Fernando|||0000-0002-5337-8774
  • del Amor, Rocío
  • Morales, Sandra|||0000-0003-0763-1545
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
  • Barroso, Álvaro
  • Kemper, Björn
  • Schnekenburger, Jürgen
[EN] Optical coherence tomography (OCT) has transformed ophthalmology by enabling accurate diagnosis of retinal diseases. OCT in rodents is used in research studies of ophthalmic diseases and treatment evaluation. However, the scarcity of data limits the effectiveness of deep learning models for this purpose. This study explores the use of synthetic data generation to improve deep learning models for OCT image analysis from rodent retinas. To address this, the research leverages advanced class-conditional generative models to create high-quality synthetic images of ex-vivo rodent retinas embedded in two different mediums (solid resin and liquid media) with a reduced initial database. The conditional Denoising Diffusion Probabilistic Model (DDPM) proved to be the best model, as it produces more realistic images that closely mimic the characteristics of real rodent retinal scans. By augmenting the training datasets with these synthetic images, the performance of deep learning models in segmenting retinal layers improved by 7%. This approach reduces reliance on scarce biological samples and animal testing, promoting ethical research practices while improving the accuracy and efficiency of AI-driven medical diagnostics. The integration of synthetic data generation represents a significant advance in medical imaging, offering a scalable solution to data scarcity, generating reliable samples in two mediums and promoting the development of more robust and accurate diagnostic tools., This work has received funding from Horizon Europe, the European Union s Framework Programme for Research and Innovation, under Grant Agreement No. 101070062 (SEQUOIA). Generalitat Valenciana partially funded this work through project CIPROM/2022/20. The work of Sandra Morales has been cofunded by the Universitat Polit`ecnica de Val`encia, Spain through the program PAID10-20. The work of Rocio del Amor has been supported by the Spanish Ministry of Universities under an FPU Grant (FPU20/05263).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Unsupervised Defect Detection for Infrastructure Inspection

  • Pérez-García de la Puente, Natalia Lourdes|||0009-0009-9704-9102
  • del Amor, Rocío
  • García-Torres, Fernando|||0000-0002-5337-8774
  • Colomer, Adrián|||0000-0002-7616-6029
  • Naranjo Ornedo, Valeriana|||0000-0002-0181-3412
[EN] Artificial Intelligence (AI) provides a fundamental aid in building operations, allowing infrastructure inspection and compliance with safety standards. In the collaborative tasks involved, detecting areas of interest, such as surface defects, is crucial. A drawback of supervised AI-based approaches is that they require manual annotation, which entails additional costs. This paper presents a novel unsupervised anomaly detection approach for locating defects based on generative models that learn the distribution of defect-free images. Using attention maps to validate in a subset, we propose a formulation that does not require accessing labelled images, enabling task automation, maintenance optimisation and cost reduction., This work has received funding from Horizon Europe, the European Union¿s Framework Programme for Research and Innovation, under Grant Agreement No. 101058054 (TURBO) and No. 101057404 (ZDZW). The work of Rocio del Amor has been supported by the Spanish Ministry of Universities (FPU20/05263).
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

How Discriminative Are Your Qrels? How To Study the Statistical Significance of Document Adjudication Methods

  • Otero, David
  • Parapar, Javier
  • Ferro, Nicola
[Abstract]: Creating test collections for offline retrieval evaluation requires human effort to judge documents' relevance. This expensive activity motivated much work in developing methods for constructing benchmarks with fewer assessment costs. In this respect, adjudication methods actively decide both which documents and the order in which experts review them, in order to better exploit the assessment budget or to lower it. Researchers evaluate the quality of those methods by measuring the correlation between the known gold ranking of systems under the full collection and the observed ranking of systems under the lower-cost one. This traditional analysis ignores whether and how the low-cost judgements impact on the statistically significant differences among systems with respect to the full collection. We fill this void by proposing a novel methodology to evaluate how the low-cost adjudication methods preserve the pairwise significant differences between systems as the full collection. In other terms, while traditional approaches look for stability in answering the question "is system A better than system B?", our proposed approach looks for stability in answering the question "is system A significantly better than system B?", which is the ultimate questions researchers need to answer to guarantee the generalisability of their results. Among other results, we found that the best methods in terms of ranking of systems correlation do not always match those preserving statistical significance., This work has received support from: (i) project PLEC2021-007662 (MCIN/AEI/10.13039/501100011033, Ministerio de Ciencia e Inno-vación, Agencia Estatal de Investigación, Plan de Recuperación, Transformación y Resiliencia, Unión Europea-Next GenerationEU); (ii) Programa de Ayudas para la Formación de Profesorado Universitario, grant number FPU20/02659 (Ministerio de Universidades); (iii) project PID2022-137061OB-C21 (Proyectos de Generación de Conocimiento, MCIN); (iv) project ED431-B 2022/33 (Xunta de Galicia/ERDF); (v) CAMEO, PRIN 2022 n. 2022ZLL7MW., Xunta de Galicia; ED431-B 2022/33
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Contribución de congreso (ConferenceOutput)

Robust fusion for bayesian semantic mapping

  • Morilla-Cabello, David
  • Mur-Labadia, Lorenzo
  • Martinez-Cantin, Ruben
  • Montijano, Eduardo
The integration of semantic information in a map allows robots to understand better their environment and make high-level decisions. In the last few years, neural networks have shown enormous progress in their perception capabilities. However, when fusing multiple observations from a neural network in a semantic map, its inherent overconfidence with unknown data gives too much weight to the outliers and decreases the robustness. To mitigate this issue we propose a novel robust fusion method to combine multiple Bayesian semantic predictions. Our method uses the uncertainty estimation provided by a Bayesian neural network to calibrate the way in which the measurements are fused. This is done by regularizing the observations to mitigate the problem of overconfident outlier predictions and using the epistemic uncertainty to weigh their influence in the fusion, resulting in a different formulation of the probability distributions. We validate our robust fusion strategy by performing experiments on photo-realistic simulated environments and real scenes. In both cases, we use a network trained on different data to expose the model to varying data distributions. The results show that considering the model's uncertainty and regularizing the probability distribution of the observations distribution results in a better semantic segmentation performance and more robustness to outliers, compared with other methods. Video - https://youtu.be/5xVGm7z9c-0
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Contribución de congreso (ConferenceOutput)

LightDepth: Single-view depth self-supervision from illumination decline

  • Rodríguez-Puigvert, Javier
  • Batlle, Víctor M.
  • Martínez Montiel, J.M.
  • Martinez-Cantin, Ruben
  • Fua, Pascal
  • Tardós, Juan D.
  • Civera, Javier
Single-view depth estimation can be remarkably effective if there is enough ground-truth depth data for supervised training. However, there are scenarios, especially in medicine in the case of endoscopies, where such data cannot be obtained. In such cases, multi-view self-supervision and synthetic-to-real transfer serve as alternative approaches, however, with a considerable performance reduction in comparison to supervised case. Instead, we propose a single-view self-supervised method that achieves a performance similar to the supervised case. In some medical devices, such as endoscopes, the camera and light sources are co-located at a small distance from the target surfaces. Thus, we can exploit that, for any given albedo and surface orientation, pixel brightness is inversely proportional to the square of the distance to the surface, providing a strong single-view self-supervisory signal. In our experiments, our self-supervised models deliver accuracies comparable to those of fully supervised ones, while being applicable without depth ground-truth data.
Academica-e. Repositorio Institucional de la Universidad Pública de Navarra
Publicación Contribución de congreso (ConferenceOutput)

Tolerancias de la tecnología CLAF-SIW para filtros de cavidades apiladas en frecuencias milimétricas

  • Biedma Pérez, Andrés
  • Segura-Gómez, Cleofás
  • Santiago Arriazu, David
  • Palomares-Caballero, Ángel
  • Arregui Padilla, Iván
  • Gómez Laso, Miguel Ángel
  • Padilla, Pablo
Este trabajo presenta un estudio exhaustivo de las tolerancias de fabricación observadas en la tecnología utilizada en [1]. El trabajo referenciado presenta un filtro paso banda apilado en cavidades que opera en el rango de frecuencias de ondas milimétricas. La tecnología elegida, una guía de ondas integrada en el sustrato llena de aire y sin contacto (CLAF-SIW), es idónea para este tipo de filtros gracias a su estructura multicapa. Sin embargo, a medida que aumenta la frecuencia, los errores de fabricación se vuelven proporcionalmente más significativos en comparación con el tamaño del dispositivo. Para mitigarlos, se realiza un análisis detallado de las tolerancias simuladas. A continuación, se evalúan las tolerancias observadas en los laminados fabricados., This work presents a comprehensive study of the manufacturing tolerances observed in the technology used in [1]. The referenced work features a cavity-stacked bandpass filter operating in the millimetre-wave frequency range. The chosen technology, a contactless air-filled substrate-integrated waveguide (CLAF-SIW), is well-suited for such filters due to its multilayer structure. However, as frequency increases, manufacturing errors become proportionally more significant compared to device size. In order to mitigate them, a detailed analysis of simulated tolerances is conducted. This is followed by an evaluation of the tolerances observed in the manufactured laminates., Este trabajo ha sido apoyado por la ayudas PID2020-112545RB-C54 y PID2020-112545RB-C53 financiadas por MCIN/AEI/10.13039/501100011033 y Union Europea NextGenerationEU/PRTR. También ha sido apoyado por las ayudas TED2021-129938B-I00, TED2021-131699B-I00 y PDC2023-145862-I00 financiadas por MCIN/AEI/10.13039/501100011033 y por la ayuda FPU20/00256 financiada por el Ministerio de Universidades.