Descrizione del progetto
Aiutare i computer a vedere meglio le cose
La creazione di reti neurali convoluzionali (CNN, una classe di algoritmi di apprendimento profondo) ha rivoluzionato la visione artificiale consentendo ai computer di «vedere» le cose e reagire ad esse. Tuttavia, le CNN non hanno risolto tutti i problemi. Ad esempio, per la formazione sono ancora necessarie grandi quantità di dati etichettati e ciò non è possibile in tutte le potenziali aree di applicazione. Inoltre, la maggior parte delle reti profonde nella visione artificiale sono deboli in termini di esplicabilità. Il progetto RED, finanziato dall’UE, lavorerà per migliorare la robustezza e l’esplicabilità delle reti profonde nella visione artificiale. Esplorerà progetti di reti strutturate, metodi probabilistici e modelli generativi/discriminativi ibridi. Farà inoltre avanzare la ricerca su come valutare la robustezza e gli aspetti relativi all’esplicabilità attraverso set di dati e metriche dedicati, tenendo in considerazione le sfide dell’analisi della scena 3D.
Obiettivo
"Deep learning approaches, mostly in the form of convolutional neural networks (CNNs), have taken the field of computer vision by storm. While the progress in recent years has been astounding, it would be dangerous to believe that important problems in computer vision are close to being solved. Many canonical deep networks for vision tasks ranging from image understanding to 3D reconstruction or motion estimation perform incredibly well ""on dataset"", i.e.~in the very setting in which they have been trained. The generalization to novel, related scenarios is still lacking, however. Moreover, large amounts of labeled data are required for training, which are not available in all potential application areas. In addition, the majority of deep networks in computer vision show deficiencies in terms of explainability. That is, the role of network components is often opaque and most deep networks in vision do not output reliable quantifications of the uncertainty of the prediction, limiting the comprehension by users. In this project, we aim to significantly advance deep networks in computer vision toward improved robustness and explainability. To that end, we will investigate structured network architectures, probabilistic methods, and hybrid generative/discriminative models, all with the goal of increasing robustness and gaining explainability. This is accompanied by research on how to assess robustness and aspects of explainability via appropriate datasets and metrics. While we aim to develop a toolbox that is as independent of specific tasks as possible, the work program is grounded in concrete vision problems to monitor progress. We specifically consider the challenges of 3D scene analysis from images and video, including tasks such as panoptic segmentation, 3D reconstruction, and motion estimation. We expect the project to have significant impact in applications of computer vision where robustness is key, data is limited, and user trust is paramount."
Campo scientifico
Programma(i)
Argomento(i)
Meccanismo di finanziamento
ERC-COG - Consolidator GrantIstituzione ospitante
64289 Darmstadt
Germania