Neural Knitworks: Patched Neural Implicit Representation Networks
This work addresses the need for efficient neural implicit representations in image synthesis for applications like inpainting and denoising, though it appears incremental as it adapts existing patch-based methods to a new architecture.
The paper tackled the problem of using coordinate-based MLPs for image synthesis tasks like inpainting, super-resolution, and denoising by proposing Neural Knitwork, which optimizes image patches adversarially and enforces consistency, resulting in 80% fewer parameters than CNN-based solutions while maintaining comparable performance and training time.
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.