English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Conventional CNNs for texture synthesis consist of a sequence of(de)-convolution and up/down-sampling layers, where each layer operates locallyand lacks the ability to capture the long-term structural dependency requiredby texture synthesis. Thus, they often simply enlarge the input texture, ratherthan perform reasonable synthesis. As a compromise, many recent methodssacrifice generalizability by training and testing on the same single (or fixedset of) texture image(s), resulting in huge re-training time costs for unseenimages. In this work, based on the discovery that the assembling/stitchingoperation in traditional texture synthesis is analogous to a transposedconvolution operation, we propose a novel way of using transposed convolutionoperation. Specifically, we directly treat the whole encoded feature map of theinput texture as transposed convolution filters and the features'self-similarity map, which captures the auto-correlation information, as inputto the transposed convolution. Such a design allows our framework, oncetrained, to be generalizable to perform synthesis of unseen textures with asingle forward pass in nearly real-time. Our method achieves state-of-the-arttexture synthesis quality based on various metrics. While self-similarity helpspreserve the input textures' regular structural patterns, our framework canalso take random noise maps for irregular input textures instead ofself-similarity maps as transposed convolution inputs. It allows to get morediverse results as well as generate arbitrarily large texture outputs bydirectly sampling large noise maps in a single pass as well.

Transposer: Universal Texture Synthesis Using Feature Maps as Transposed Convolution Filter