Neural knitworks : patched neural implicit representation networks
Czerkawski, Mikolaj and Cardona, Javier and Atkinson, Robert and Michie, Craig and Andonovic, Ivan and Clemente, Carmine and Tachtatzis, Christos (2021) Neural knitworks : patched neural implicit representation networks. Other. arXiv.org, Ithaca, N.Y.. (https://doi.org/10.48550/arXiv.2109.14406)
Preview |
Text.
Filename: Czerkawski_etal_ArXiv_2022_Neural_knitworks_patched_neural_implicit.pdf
Final Published Version License: ![]() Download (20MB)| Preview |
Abstract
Coordinate-based Multilayer Perceptron (MLP) networks, despite being capable of learning neural implicit representations, are not performant for internal image synthesis applications. Convolutional Neural Networks (CNNs) are typically used instead for a variety of internal generative tasks, at the cost of a larger model. We propose Neural Knitwork, an architecture for neural implicit representation learning of natural images that achieves image synthesis by optimizing the distribution of image patches in an adversarial manner and by enforcing consistency between the patch predictions. To the best of our knowledge, this is the first implementation of a coordinate-based MLP tailored for synthesis tasks such as image inpainting, super-resolution, and denoising. We demonstrate the utility of the proposed technique by training on these three tasks. The results show that modeling natural images using patches, rather than pixels, produces results of higher fidelity. The resulting model requires 80% fewer parameters than alternative CNN-based solutions while achieving comparable performance and training time.
ORCID iDs
Czerkawski, Mikolaj





-
-
Item type: Monograph(Other) ID code: 82253 Dates: DateEvent29 September 2021PublishedKeywords: multilayer perceptron (MLP), convolutional neural network (CNN), neural knitwork, Electronic computers. Computer science, Artificial Intelligence Subjects: Science > Mathematics > Electronic computers. Computer science Department: Faculty of Engineering > Electronic and Electrical Engineering
Faculty of Engineering > Chemical and Process Engineering
Strategic Research Themes > Measurement Science and Enabling Technologies
Strategic Research Themes > Ocean, Air and SpaceDepositing user: Pure Administrator Date deposited: 07 Sep 2022 15:58 Last modified: 18 Nov 2023 03:07 URI: https://strathprints.strath.ac.uk/id/eprint/82253