Fraunhofer-Publica :: Explore section

Search Research Outputs

Recent Additions

SDFR: Synthetic Data for Face Recognition Competition

( 2024)
Otroshi Shahreza, Hatef
;
Ecabert, Christophe
;
George, Anjith
;
Unnervik, Alexander
;
Marcel, Sébastien
;
Di Domenico, Nicolò
;
Borghi, Guido
;
Maltoni, Davide
;
Boutros, Fadi
;
Vogel, Julia
;
Damer, Naser
;
Sánchez-Pérez, Ángela
;
Mas-Candela, Enrique
;
Calvo-Zaragoza, Jorge
;
Biesseck, Bernardo
;
Vidal, Pedro
;
Granada, Roger
;
Menotti, David
;
DeAndres-Tame, Ivan
;
Maurizio La Cava, Simone
;
Concas, Sara
;
Melzi, Pietro
;
Tolosana, Ruben
;
Vera-Rodriguez, Ruben
;
Perelli, Gianpaolo
;
Orrù, Giulia
;
Luca Marcialis, Gian
;
Fierrez, Julian

Large-scale face recognition datasets are collected by crawling the Internet and without individuals' consent, raising legal, ethical, and privacy concerns. With the recent advances in generative models, recently several works proposed generating synthetic face recognition datasets to mitigate concerns in web-crawled face recognition datasets. This paper presents the summary of the Synthetic Data for Face Recognition (SDFR) Competition held in conjunction with the 18th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2024) and established to investigate the use of synthetic data for training face recognition models. The SDFR competition was split into two tasks, allowing participants to train face recognition systems using new synthetic datasets and/or existing ones. In the first task, the face recognition backbone was fixed and the dataset size was limited, while the second task provided almost complete freedom on the model backbone, the dataset, and the training pipeline. The submitted models were trained on existing and also new synthetic datasets and used clever methods to improve training with synthetic data. The submissions were evaluated and ranked on a diverse set of seven benchmarking datasets. The paper gives an overview of the submitted face recognition models and reports achieved performance compared to baseline models trained on real and synthetic datasets. Furthermore, the evaluation of submissions is extended to bias assessment across different demography groups. Lastly, an outlook on the current state of the research in training face recognition models using synthetic data is presented, and existing problems as well as potential future directions are also discussed.
Artificial Intelligence Driven Trend Forecasting: Integrating BERT Topic Modelling and Generative Artificial Intelligence for Semantic Insights

( 2024)
Kumar, Deepak
;
Weissenberger-Eibl, Marion

In the fast-paced realm of technological evolution, accurately forecasting emerging trends is critical for both academic inquiry and industry application. Traditional trend analysis methodologies, while valuable, struggle to efficiently process and interpret the vast datasets of today's information age. This paper introduces a novel approach that synergizes Generative AI and Bidirectional Encoder Representations from Transformers (BERT) for semantic insights and trend forecasting, leveraging the power of Retrieval-Augmented Generation (RAG) and the analytical prowess of BERT topic modeling. By automating the analysis of extensive datasets from publications and patents, the presented methodology not only expedites the discovery of emergent trends but also enhances the precision of these findings by generating a short summary for found emergent trends. For validation, three technologies - reinforcement learning, quantum machine learning, and Cryptocurrencies - were analysed prior to their first appearance in the Gartner Hype Cycle. Research highlights the integration of advanced AI techniques in trend forecasting, providing a scalable and accurate tool for strategic planning and innovation management. Results demonstrated a significant correlation between model's predictions and the technologies' appearances in the Hype Cycle, underscoring the potential of this methodology in anticipating technological shifts across various sectors
An LNP-CRISPR gene editing drug demonstrates efficacy and safety in hereditary angioedema patients following in vivo administration

( 2024)
Pardi, Norbert
;
Ivics, Zoltan
Spatio-Temporal Transferability of Drone-Based Models to Predict Forage Supply in Drier Rangelands

( 2024)
Amputu, Vistorina
;
Männer, Florian
;
Tielbörger, Katja
;
Knox, Nichola

Unmanned aerial systems offer a cost-effective and reproducible method for monitoring natural resources in expansive areas. But the transferability of developed models, which are often based on single snapshots, is rarely tested. This is particularly relevant in rangelands where forage resources are inherently patchy in space and time, which may limit model transfer. Here, we investigated the accuracy of drone-based models in estimating key proxies of forage provision across two land tenure systems and between two periods of the growing season in semi-arid rangelands. We tested case-specific models and a landscape model, with the expectation that the landscape model performs better than the case-specific models as it captures the highest variability expected in the rangeland system. The landscape model did achieve the lowest error when predicting herbaceous biomass and predicted land cover with better or similar accuracy to the case-specific models. This reinforces the importance of incorporating the widest variation of conditions in predictive models. This study contributes to understanding model transferability in drier rangeland systems characterized by spatial and temporal heterogeneity. By advancing the integration of drone technology for accurate monitoring of such dynamic ecosystems, this research contributes to sustainable rangeland management practices.
NeRF-FF: A Plug-in Method to Mitigate Defocus Blur for Runtime Optimized Neural Radiance Fields

( 2024)
Wirth, Tristan
;
Rak, Arne
;
Buelow, Max von
;
Knauthe, Volker
;
Kuijper, Arjan
;
Fellner, Dieter

Neural radiance fields (NeRFs) have revolutionized novel view synthesis, leading to an unprecedented level of realism in rendered images. However, the reconstruction quality of NeRFs suffers significantly from out-of-focus regions in the input images. We propose NeRF-FF, a plug-in strategy that estimates image masks based on Focus Frustums (FFs), i.e., the visible volume in the scene space that is in-focus. NeRF-FF enables a subsequently trained NeRF model to omit out-of-focus image regions during the training process. Existing methods to mitigate the effects of defocus blurred input images often leverage dynamic ray generation. This makes them incompatible with the static ray assumptions employed by runtime-performance-optimized NeRF variants, such as Instant-NGP, leading to high training times. Our experiments show that NeRF-FF outperforms state-of-the-art approaches regarding training time by two orders of magnitude - reducing it to under 1 min on end-consumer hardware - while maintaining comparable visual quality.

Most viewed

RFID system for the identification of biological samples

( 2010)
Hichri, K.
;
Wick, H.
;
Zimmermann, H.
;
Ihmig, F.

We describe a RFID (Radio Frequency IDentification) technique suitable for use with the high sample packing densities found in biotechnology laboratories and storage systems. An innovative hardware design overcomes the collision problem. The transponders themselves are stored inside screening coils that can be magnetically disabled, allowing the reading of a single device in a close packed array. The technique can also be used to find the position of a particular transponder. Testing has, so far, been at room temperature but the system has been designed to operate down to liquid nitrogen temperatures.
Was wir über Innovationen wissen: Konturen einer Wissenswirtschaft

( 2006)
Grupp, H.
A neural network based on first principles

( 2020)
Baggenstoss, P.M.

In this paper, a Neural network is derived from first principles, assuming only that each layer begins with a linear dimension-reducing transformation. The approach appeals to the principle of Maximum Entropy (MaxEnt) to find the posterior distribution of the input data of each layer, conditioned on the layer output variables. This posterior has a well-defined mean, the conditional mean estimator, that is calculated using a type of neural network with theoretically-derived activation functions similar to sigmoid, softplus, and relu. This implicitly provides a theoretical justification for their use. A theorem that finds the conditional distribution and conditional mean estimator under the MaxEnt prior is proposed, unifying results for special cases. Combining layers results in an auto-encoder with conventional feed-forward analysis network and a type of linear Bayesian belief network in the reconstruction path.
Surface etching of methacrylic microparticles via basic hydrolysis and introduction of functional groups for click chemistry

( 2013)
Speyerer, Christian
;
Borchers, K.
;
Hirth, Thomas
;
Tovar, G.E.M.
;
Weber, A.

Controlled basic hydrolysis of poly(methyl methacrylate-co-ethylene glycol dimethacrylate) P(MMA-co-EGDMA) microparticles with a diameter d(50) = 6 mu m led to high densities of carboxylic groups at the particles' surface of up to 1.288 mu eq g(-1) (equivalent to 1.277 mu mol m(-2)). The microparticles' core has not been altered by this surface activation procedure as seen by fluorescent staining. The kinetics of the hydrolysis reaction was investigated via electrophoretic light scattering and particle charge detection employing polycation titration under shear condition. The activated microparticle's surface was subsequently exploited in carbodiimide-mediated coupling reactions using a variety of molecular reactants, that is, 11-azido-3,6,9-trioxaundecan-1-amine, cysteamine, propargylamine, and fluoresceinamine, thus enabling the introduction of chemically reactive moieties such as azides, thiols, and alkynes. Fluorescent staining of the particles' surface successfully demonstrated the versatile applications of surface functionalized microparticles via copper-catalyzed huisgen cycloaddition. Carrying on this two-step procedure in a controlled manner provides an excellent way for relatively simple but highly effective surface functionalization.
Reinigungstechnik

( 2000)
Elkmann, N.