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

Referring image segmentation aims to segment an object mentioned in naturallanguage from an image. A main challenge is language-related localization,which means locating the object with the relevant language. Previous approachesmainly focus on the fusion of vision and language features without fullyaddressing language-related localization. In previous approaches, fusedvision-language features are directly fed into a decoder and pass through aconvolution with a fixed kernel to obtain the result, which follows a similarpattern as traditional image segmentation. This approach does not explicitlyalign language and vision features in the segmentation stage, resulting in asuboptimal language-related localization. Different from previous methods, wepropose Explicitly Align the Vision and Language for Referring ImageSegmentation (EAVL). Instead of using a fixed convolution kernel, we propose anAligner which explicitly aligns the vision and language features in thesegmentation stage. Specifically, a series of unfixed convolution kernels aregenerated based on the input l, and then are use to explicitly align the visionand language features. To achieve this, We generate multiple queries thatrepresent different emphases of the language expression. These queries aretransformed into a series of query-based convolution kernels. Then, we utilizethese kernels to do convolutions in the segmentation stage and obtain a seriesof segmentation masks. The final result is obtained through the aggregation ofall masks. Our method can not only fuse vision and language featureseffectively but also exploit their potential in the segmentation stage. Andmost importantly, we explicitly align language features of different emphaseswith the image features to achieve language-related localization. Our methodsurpasses previous state-of-the-art methods on RefCOCO, RefCOCO+, and G-Ref bylarge margins.

EAVL: Explicitly Align Vision and Language for Referring Image Segmentation