Cover Image

Conditional invertible neural networks for diverse image-to-image translation

Show other versions (1)

We introduce a new architecture called a conditional invertible neural network (cINN), and use it to address the task of diverse image-to-image translation for natural images. This is not easily possible with existing INN models due to some fundamental limitations. The cINN combines the purely gener...

Full description

Saved in:
Bibliographic Details
Main Authors: Ardizzone, Lynton (Author) , Kruse, Jakob (Author) , Lüth, Carsten (Author) , Bracher, Niels (Author) , Rother, Carsten (Author) , Köthe, Ullrich (Author)
Format: Article (Journal) Chapter/Article
Language:English
Published: 5 May 2021
In: Arxiv
Year: 2021, Pages: 1-15
DOI:10.48550/arXiv.2105.02104
Online Access:Verlag, lizenzpflichtig, Volltext: https://doi.org/10.48550/arXiv.2105.02104
Verlag, lizenzpflichtig, Volltext: http://arxiv.org/abs/2105.02104
Get full text
Author Notes:Lynton Ardizzone, Jakob Kruse, Carsten Lüth, Niels Bracher, Carsten Rother, Ullrich Köthe

MARC

LEADER 00000caa a2200000 c 4500
001 1817338129
003 DE-627
005 20230118161740.0
007 cr uuu---uuuuu
008 220923s2021 xx |||||o 00| ||eng c
024 7 |a 10.48550/arXiv.2105.02104  |2 doi 
035 |a (DE-627)1817338129 
035 |a (DE-599)KXP1817338129 
035 |a (OCoLC)1361714249 
040 |a DE-627  |b ger  |c DE-627  |e rda 
041 |a eng 
084 |a 29  |2 sdnb 
100 1 |a Ardizzone, Lynton  |d 1994-  |e VerfasserIn  |0 (DE-588)1194988512  |0 (DE-627)1677182296  |4 aut 
245 1 0 |a Conditional invertible neural networks for diverse image-to-image translation  |c Lynton Ardizzone, Jakob Kruse, Carsten Lüth, Niels Bracher, Carsten Rother, Ullrich Köthe 
264 1 |c 5 May 2021 
300 |a 15 
336 |a Text  |b txt  |2 rdacontent 
337 |a Computermedien  |b c  |2 rdamedia 
338 |a Online-Ressource  |b cr  |2 rdacarrier 
500 |a Gesehen am 28.09.2022 
520 |a We introduce a new architecture called a conditional invertible neural network (cINN), and use it to address the task of diverse image-to-image translation for natural images. This is not easily possible with existing INN models due to some fundamental limitations. The cINN combines the purely generative INN model with an unconstrained feed-forward network, which efficiently preprocesses the conditioning image into maximally informative features. All parameters of a cINN are jointly optimized with a stable, maximum likelihood-based training procedure. Even though INN-based models have received far less attention in the literature than GANs, they have been shown to have some remarkable properties absent in GANs, e.g. apparent immunity to mode collapse. We find that our cINNs leverage these properties for image-to-image translation, demonstrated on day to night translation and image colorization. Furthermore, we take advantage of our bidirectional cINN architecture to explore and manipulate emergent properties of the latent space, such as changing the image style in an intuitive way. 
650 4 |a 68T01 
650 4 |a Computer Science - Artificial Intelligence 
650 4 |a Computer Science - Computer Vision and Pattern Recognition 
700 1 |a Kruse, Jakob  |d 1990-  |e VerfasserIn  |0 (DE-588)1194989497  |0 (DE-627)1677183136  |4 aut 
700 1 |a Lüth, Carsten  |e VerfasserIn  |0 (DE-588)1262948428  |0 (DE-627)1810862000  |4 aut 
700 1 |a Bracher, Niels  |e VerfasserIn  |0 (DE-588)1269038583  |0 (DE-627)1817695398  |4 aut 
700 1 |a Rother, Carsten  |e VerfasserIn  |0 (DE-588)1181464692  |0 (DE-627)1662676883  |4 aut 
700 1 |a Köthe, Ullrich  |e VerfasserIn  |0 (DE-588)123963435  |0 (DE-627)594480884  |0 (DE-576)304484520  |4 aut 
773 0 8 |i Enthalten in  |t Arxiv  |d Ithaca, NY : Cornell University, 1991  |g (2021), Artikel-ID 2105.02104, Seite 1-15  |h Online-Ressource  |w (DE-627)509006531  |w (DE-600)2225896-6  |w (DE-576)28130436X  |7 nnas  |a Conditional invertible neural networks for diverse image-to-image translation 
773 1 8 |g year:2021  |g elocationid:2105.02104  |g pages:1-15  |g extent:15  |a Conditional invertible neural networks for diverse image-to-image translation 
856 4 0 |u https://doi.org/10.48550/arXiv.2105.02104  |x Verlag  |x Resolving-System  |z lizenzpflichtig  |3 Volltext 
856 4 0 |u http://arxiv.org/abs/2105.02104  |x Verlag  |z lizenzpflichtig  |3 Volltext 
951 |a AR 
992 |a 20220923 
993 |a Article 
994 |a 2021 
998 |g 123963435  |a Köthe, Ullrich  |m 123963435:Köthe, Ullrich  |d 700000  |d 708070  |d 700000  |d 728500  |e 700000PK123963435  |e 708070PK123963435  |e 700000PK123963435  |e 728500PK123963435  |k 0/700000/  |k 1/700000/708070/  |k 0/700000/  |k 1/700000/728500/  |p 6  |y j 
998 |g 1181464692  |a Rother, Carsten  |m 1181464692:Rother, Carsten  |d 700000  |d 708070  |d 700000  |d 728500  |e 700000PR1181464692  |e 708070PR1181464692  |e 700000PR1181464692  |e 728500PR1181464692  |k 0/700000/  |k 1/700000/708070/  |k 0/700000/  |k 1/700000/728500/  |p 5 
998 |g 1262948428  |a Lüth, Carsten  |m 1262948428:Lüth, Carsten  |d 110000  |e 110000PL1262948428  |k 0/110000/  |p 3 
998 |g 1194989497  |a Kruse, Jakob  |m 1194989497:Kruse, Jakob  |d 700000  |d 708000  |e 700000PK1194989497  |e 708000PK1194989497  |k 0/700000/  |k 1/700000/708000/  |p 2 
998 |g 1194988512  |a Ardizzone, Lynton  |m 1194988512:Ardizzone, Lynton  |d 110000  |d 700000  |d 728500  |e 110000PA1194988512  |e 700000PA1194988512  |e 728500PA1194988512  |k 0/110000/  |k 0/700000/  |k 1/700000/728500/  |p 1  |x j 
999 |a KXP-PPN1817338129  |e 4191093185 
BIB |a Y 
JSO |a {"title":[{"title_sort":"Conditional invertible neural networks for diverse image-to-image translation","title":"Conditional invertible neural networks for diverse image-to-image translation"}],"person":[{"role":"aut","roleDisplay":"VerfasserIn","display":"Ardizzone, Lynton","given":"Lynton","family":"Ardizzone"},{"roleDisplay":"VerfasserIn","display":"Kruse, Jakob","given":"Jakob","family":"Kruse","role":"aut"},{"display":"Lüth, Carsten","roleDisplay":"VerfasserIn","given":"Carsten","family":"Lüth","role":"aut"},{"family":"Bracher","display":"Bracher, Niels","roleDisplay":"VerfasserIn","given":"Niels","role":"aut"},{"role":"aut","family":"Rother","given":"Carsten","roleDisplay":"VerfasserIn","display":"Rother, Carsten"},{"role":"aut","family":"Köthe","roleDisplay":"VerfasserIn","display":"Köthe, Ullrich","given":"Ullrich"}],"type":{"bibl":"chapter","media":"Online-Ressource"},"name":{"displayForm":["Lynton Ardizzone, Jakob Kruse, Carsten Lüth, Niels Bracher, Carsten Rother, Ullrich Köthe"]},"recId":"1817338129","id":{"doi":["10.48550/arXiv.2105.02104"],"eki":["1817338129"]},"language":["eng"],"physDesc":[{"extent":"15 S."}],"relHost":[{"origin":[{"publisherPlace":"Ithaca, NY ; [Erscheinungsort nicht ermittelbar]","dateIssuedDisp":"1991-","publisher":"Cornell University ; Arxiv.org","dateIssuedKey":"1991"}],"physDesc":[{"extent":"Online-Ressource"}],"note":["Gesehen am 28.05.2024"],"language":["eng"],"disp":"Conditional invertible neural networks for diverse image-to-image translationArxiv","id":{"eki":["509006531"],"zdb":["2225896-6"]},"recId":"509006531","pubHistory":["1991 -"],"type":{"bibl":"edited-book","media":"Online-Ressource"},"part":{"text":"(2021), Artikel-ID 2105.02104, Seite 1-15","extent":"15","pages":"1-15","year":"2021"},"title":[{"title_sort":"Arxiv","title":"Arxiv"}],"titleAlt":[{"title":"Arxiv.org"},{"title":"Arxiv.org e-print archive"},{"title":"Arxiv e-print archive"},{"title":"De.arxiv.org"}]}],"note":["Gesehen am 28.09.2022"],"origin":[{"dateIssuedDisp":"5 May 2021","dateIssuedKey":"2021"}]} 
SRT |a ARDIZZONELCONDITIONA5202 

Similar Items