The Experts below are selected from a list of 4575 Experts worldwide ranked by ideXlab platform
Aggelos K. Katsaggelos - One of the best experts on this subject based on the ideXlab platform.
-
A Single Video Super-Resolution GAN for Multiple Downsampling Operators based on Pseudo-Inverse Image Formation Models
arXiv: Computer Vision and Pattern Recognition, 2019Co-Authors: Santiago Lopez-tapia, Alice Lucas, Rafael Molina, Aggelos K. KatsaggelosAbstract:The popularity of high and ultra-high definition displays has led to the need for methods to improve the quality of videos already obtained at much lower resolutions. Current Video Super-Resolution methods are not robust to mismatch between training and testing degradation models since they are trained against a single degradation model (usually bicubic downsampling). This causes their performance to deteriorate in real-life applications. At the same time, the use of only the Mean Squared Error during learning causes the resulting Images to be too smooth. In this work we propose a new Convolutional Neural Network for video super resolution which is robust to multiple degradation models. During training, which is performed on a large dataset of scenes with slow and fast motions, it uses the pseudo-Inverse Image formation model as part of the network architecture in conjunction with perceptual losses, in addition to a smoothness constraint that eliminates the artifacts originating from these perceptual losses. The experimental validation shows that our approach outperforms current state-of-the-art methods and is robust to multiple degradations.
-
Multiple-Degradation Video Super-Resolution with Direct Inversion of the Low-Resolution Formation Model
2019 27th European Signal Processing Conference (EUSIPCO), 2019Co-Authors: Santiago Lopez-tapia, Alice Lucas, Rafael Molina, Aggelos K. KatsaggelosAbstract:With the increase of popularity of high and ultra high definition displays, the need to improve the quality of content already obtained at much lower resolutions has grown. Since current video super-resolution methods are trained with a single degradation model (usually bicubic downsampling), they are not robust to mismatch between training and testing degradation models, in which case their performance deteriorates. In this work we propose a new Convolutional Neural Network for video super resolution which is robust to multiple degradation models and uses the pseudo-Inverse Image formation model as part of the network architecture during training. The experimental validation shows that our approach outperforms current state of the art methods.
Amnon Neeman - One of the best experts on this subject based on the ideXlab platform.
-
relation between two twisted Inverse Image pseudofunctors in duality theory
Compositio Mathematica, 2015Co-Authors: Srikanth B Iyengar, Joseph Lipman, Amnon NeemanAbstract:Grothendieck duality theory assigns to essentially-finite-type maps f of noetherian schemes a pseudofunctor f^\times right-adjoint to Rf_*, and a pseudofunctor f^! agreeing with f^\times when f is proper, but equal to the usual Inverse Image f^* when f is etale. We define and study a canonical map from the first pseudofunctor to the second. This map behaves well with respect to flat base change, and is taken to an isomorphism by "compactly supported" versions of standard derived functors. Concrete realizations are described, for instance for maps of affine schemes. Applications include proofs of reduction theorems for Hochschild homology and cohomology, and of a remarkable formula for the fundamental class of a flat map of affine schemes.
-
Relation between two twisted Inverse Image pseudofunctors in duality theory
Compositio Mathematica, 2014Co-Authors: Srikanth B Iyengar, Joseph Lipman, Amnon NeemanAbstract:Grothendieck duality theory assigns to essentially finite-type maps $f$ of noetherian schemes a pseudofunctor $f^{\times }$ right-adjoint to $\mathsf{R}f_{\ast }$, and a pseudofunctor $f^{!}$ agreeing with $f^{\times }$ when $f$ is proper, but equal to the usual Inverse Image $f^{\ast }$ when $f$ is étale. We define and study a canonical map from the first pseudofunctor to the second. This map behaves well with respect to flat base change, and is taken to an isomorphism by ‘compactly supported’ versions of standard derived functors. Concrete realizations are described, for instance for maps of affine schemes. Applications include proofs of reduction theorems for Hochschild homology and cohomology, and of a remarkable formula for the fundamental class of a flat map of affine schemes.
-
quasi perfect scheme maps and boundedness of the twisted Inverse Image functor
Illinois Journal of Mathematics, 2007Co-Authors: Joseph Lipman, Amnon NeemanAbstract:For a map f : X → Y of quasi-compact quasi-separated schemes, we discuss quasi-perfection, i.e., the right adjoint f of Rf∗ respects small direct sums. This is equivalent to the existence of a functorial isomorphism f×OY ⊗ L Lf(− ) −→ ∼ f(−); to quasi-properness (preservation by Rf∗ of pseudo-coherence, or just properness in the noetherian case) plus boundedness of Lf (finite tor-dimensionality), or of the functor f; and to some other conditions. We use a globalization, previously known only for divisorial schemes, of the local definition of pseudo-coherence of complexes, as well as a refinement of the known fact that the derived category of complexes with quasi-coherent homology is generated by a single perfect complex.
-
quasi perfect scheme maps and boundedness of the twisted Inverse Image functor
arXiv: Algebraic Geometry, 2006Co-Authors: Joseph Lipman, Amnon NeemanAbstract:For a map f: X -> Y of quasi-compact quasi-separated schemes, we discuss quasi-perfection, that is, the right adjoint f^\times of the derived functor Rf_* respects small direct sums. This is equivalent to the existence of a functorial isomorphism f^\times O_{Y} \otimes^L Lf^*(-) \to f^\times (-); to quasi-properness (preservation by Rf_* of pseudo-coherence, or just properness in the noetherian case) plus boundedness of Lf^* (finite tor-dimensionality), or of the functor f^\times; and to some other conditions. We use a globalization, previously known only for divisorial schemes, of the local definition of pseudo-coherence of complexes, as well as a refinement of the known fact that the derived category of complexes with quasi-coherent homology is generated by a single perfect complex.
Santiago Lopez-tapia - One of the best experts on this subject based on the ideXlab platform.
-
A Single Video Super-Resolution GAN for Multiple Downsampling Operators based on Pseudo-Inverse Image Formation Models
arXiv: Computer Vision and Pattern Recognition, 2019Co-Authors: Santiago Lopez-tapia, Alice Lucas, Rafael Molina, Aggelos K. KatsaggelosAbstract:The popularity of high and ultra-high definition displays has led to the need for methods to improve the quality of videos already obtained at much lower resolutions. Current Video Super-Resolution methods are not robust to mismatch between training and testing degradation models since they are trained against a single degradation model (usually bicubic downsampling). This causes their performance to deteriorate in real-life applications. At the same time, the use of only the Mean Squared Error during learning causes the resulting Images to be too smooth. In this work we propose a new Convolutional Neural Network for video super resolution which is robust to multiple degradation models. During training, which is performed on a large dataset of scenes with slow and fast motions, it uses the pseudo-Inverse Image formation model as part of the network architecture in conjunction with perceptual losses, in addition to a smoothness constraint that eliminates the artifacts originating from these perceptual losses. The experimental validation shows that our approach outperforms current state-of-the-art methods and is robust to multiple degradations.
-
Multiple-Degradation Video Super-Resolution with Direct Inversion of the Low-Resolution Formation Model
2019 27th European Signal Processing Conference (EUSIPCO), 2019Co-Authors: Santiago Lopez-tapia, Alice Lucas, Rafael Molina, Aggelos K. KatsaggelosAbstract:With the increase of popularity of high and ultra high definition displays, the need to improve the quality of content already obtained at much lower resolutions has grown. Since current video super-resolution methods are trained with a single degradation model (usually bicubic downsampling), they are not robust to mismatch between training and testing degradation models, in which case their performance deteriorates. In this work we propose a new Convolutional Neural Network for video super resolution which is robust to multiple degradation models and uses the pseudo-Inverse Image formation model as part of the network architecture during training. The experimental validation shows that our approach outperforms current state of the art methods.
Joseph Lipman - One of the best experts on this subject based on the ideXlab platform.
-
relation between two twisted Inverse Image pseudofunctors in duality theory
Compositio Mathematica, 2015Co-Authors: Srikanth B Iyengar, Joseph Lipman, Amnon NeemanAbstract:Grothendieck duality theory assigns to essentially-finite-type maps f of noetherian schemes a pseudofunctor f^\times right-adjoint to Rf_*, and a pseudofunctor f^! agreeing with f^\times when f is proper, but equal to the usual Inverse Image f^* when f is etale. We define and study a canonical map from the first pseudofunctor to the second. This map behaves well with respect to flat base change, and is taken to an isomorphism by "compactly supported" versions of standard derived functors. Concrete realizations are described, for instance for maps of affine schemes. Applications include proofs of reduction theorems for Hochschild homology and cohomology, and of a remarkable formula for the fundamental class of a flat map of affine schemes.
-
Relation between two twisted Inverse Image pseudofunctors in duality theory
Compositio Mathematica, 2014Co-Authors: Srikanth B Iyengar, Joseph Lipman, Amnon NeemanAbstract:Grothendieck duality theory assigns to essentially finite-type maps $f$ of noetherian schemes a pseudofunctor $f^{\times }$ right-adjoint to $\mathsf{R}f_{\ast }$, and a pseudofunctor $f^{!}$ agreeing with $f^{\times }$ when $f$ is proper, but equal to the usual Inverse Image $f^{\ast }$ when $f$ is étale. We define and study a canonical map from the first pseudofunctor to the second. This map behaves well with respect to flat base change, and is taken to an isomorphism by ‘compactly supported’ versions of standard derived functors. Concrete realizations are described, for instance for maps of affine schemes. Applications include proofs of reduction theorems for Hochschild homology and cohomology, and of a remarkable formula for the fundamental class of a flat map of affine schemes.
-
quasi perfect scheme maps and boundedness of the twisted Inverse Image functor
Illinois Journal of Mathematics, 2007Co-Authors: Joseph Lipman, Amnon NeemanAbstract:For a map f : X → Y of quasi-compact quasi-separated schemes, we discuss quasi-perfection, i.e., the right adjoint f of Rf∗ respects small direct sums. This is equivalent to the existence of a functorial isomorphism f×OY ⊗ L Lf(− ) −→ ∼ f(−); to quasi-properness (preservation by Rf∗ of pseudo-coherence, or just properness in the noetherian case) plus boundedness of Lf (finite tor-dimensionality), or of the functor f; and to some other conditions. We use a globalization, previously known only for divisorial schemes, of the local definition of pseudo-coherence of complexes, as well as a refinement of the known fact that the derived category of complexes with quasi-coherent homology is generated by a single perfect complex.
-
quasi perfect scheme maps and boundedness of the twisted Inverse Image functor
arXiv: Algebraic Geometry, 2006Co-Authors: Joseph Lipman, Amnon NeemanAbstract:For a map f: X -> Y of quasi-compact quasi-separated schemes, we discuss quasi-perfection, that is, the right adjoint f^\times of the derived functor Rf_* respects small direct sums. This is equivalent to the existence of a functorial isomorphism f^\times O_{Y} \otimes^L Lf^*(-) \to f^\times (-); to quasi-properness (preservation by Rf_* of pseudo-coherence, or just properness in the noetherian case) plus boundedness of Lf^* (finite tor-dimensionality), or of the functor f^\times; and to some other conditions. We use a globalization, previously known only for divisorial schemes, of the local definition of pseudo-coherence of complexes, as well as a refinement of the known fact that the derived category of complexes with quasi-coherent homology is generated by a single perfect complex.
Alice Lucas - One of the best experts on this subject based on the ideXlab platform.
-
A Single Video Super-Resolution GAN for Multiple Downsampling Operators based on Pseudo-Inverse Image Formation Models
arXiv: Computer Vision and Pattern Recognition, 2019Co-Authors: Santiago Lopez-tapia, Alice Lucas, Rafael Molina, Aggelos K. KatsaggelosAbstract:The popularity of high and ultra-high definition displays has led to the need for methods to improve the quality of videos already obtained at much lower resolutions. Current Video Super-Resolution methods are not robust to mismatch between training and testing degradation models since they are trained against a single degradation model (usually bicubic downsampling). This causes their performance to deteriorate in real-life applications. At the same time, the use of only the Mean Squared Error during learning causes the resulting Images to be too smooth. In this work we propose a new Convolutional Neural Network for video super resolution which is robust to multiple degradation models. During training, which is performed on a large dataset of scenes with slow and fast motions, it uses the pseudo-Inverse Image formation model as part of the network architecture in conjunction with perceptual losses, in addition to a smoothness constraint that eliminates the artifacts originating from these perceptual losses. The experimental validation shows that our approach outperforms current state-of-the-art methods and is robust to multiple degradations.
-
Multiple-Degradation Video Super-Resolution with Direct Inversion of the Low-Resolution Formation Model
2019 27th European Signal Processing Conference (EUSIPCO), 2019Co-Authors: Santiago Lopez-tapia, Alice Lucas, Rafael Molina, Aggelos K. KatsaggelosAbstract:With the increase of popularity of high and ultra high definition displays, the need to improve the quality of content already obtained at much lower resolutions has grown. Since current video super-resolution methods are trained with a single degradation model (usually bicubic downsampling), they are not robust to mismatch between training and testing degradation models, in which case their performance deteriorates. In this work we propose a new Convolutional Neural Network for video super resolution which is robust to multiple degradation models and uses the pseudo-Inverse Image formation model as part of the network architecture during training. The experimental validation shows that our approach outperforms current state of the art methods.