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

Interactive Image Segmentation (IIS) has emerged as a promising technique fordecreasing annotation time. Substantial progress has been made in pre- andpost-processing for IIS, but the critical issue of interaction ambiguitynotably hindering segmentation quality, has been under-researched. To addressthis, we introduce AdaptiveClick -- a clicks-aware transformer incorporating anadaptive focal loss, which tackles annotation inconsistencies with tools formask- and pixel-level ambiguity resolution. To the best of our knowledge,AdaptiveClick is the first transformer-based, mask-adaptive segmentationframework for IIS. The key ingredient of our method is the Clicks-awareMask-adaptive Transformer Decoder (CAMD), which enhances the interactionbetween clicks and image features. Additionally, AdaptiveClick enablespixel-adaptive differentiation of hard and easy samples in the decision space,independent of their varying distributions. This is primarily achieved byoptimizing a generalized Adaptive Focal Loss (AFL) with a theoreticalguarantee, where two adaptive coefficients control the ratio of gradient valuesfor hard and easy pixels. Our analysis reveals that the commonly used Focal andBCE losses can be considered special cases of the proposed AFL loss. With aplain ViT backbone, extensive experimental results on nine datasets demonstratethe superiority of AdaptiveClick compared to state-of-the-art methods. Codewill be publicly available at https://github.com/lab206/AdaptiveClick.

AdaptiveClick: Clicks-aware Transformer with Adaptive Focal Loss for Interactive Image Segmentation