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

The long-standing theory that a colour-naming system evolves under dualpressure of efficient communication and perceptual mechanism is supported bymore and more linguistic studies, including analysing four decades ofdiachronic data from the Nafaanra language. This inspires us to explore whethermachine learning could evolve and discover a similar colour-naming system viaoptimising the communication efficiency represented by high-level recognitionperformance. Here, we propose a novel colour quantisation transformer,CQFormer, that quantises colour space while maintaining the accuracy of machinerecognition on the quantised images. Given an RGB image, Annotation Branch mapsit into an index map before generating the quantised image with a colourpalette; meanwhile the Palette Branch utilises a key-point detection way tofind proper colours in the palette among the whole colour space. By interactingwith colour annotation, CQFormer is able to balance both the machine visionaccuracy and colour perceptual structure such as distinct and stable colourdistribution for discovered colour system. Very interestingly, we even observethe consistent evolution pattern between our artificial colour system and basiccolour terms across human languages. Besides, our colour quantisation methodalso offers an efficient quantisation method that effectively compresses theimage storage while maintaining high performance in high-level recognitiontasks such as classification and detection. Extensive experiments demonstratethe superior performance of our method with extremely low bit-rate colours,showing potential to integrate into quantisation network to quantities fromimage to network activation. The source code is available athttps://github.com/ryeocthiv/CQFormer

Name Your Colour For the Task: Artificially Discover Colour Naming via Colour Quantisation Transformer