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

Existing techniques for text detection can be broadly classified into twoprimary groups: segmentation-based and regression-based methods. Segmentationmodels offer enhanced robustness to font variations but require intricatepost-processing, leading to high computational overhead. Regression-basedmethods undertake instance-aware prediction but face limitations in robustnessand data efficiency due to their reliance on high-level representations. In ouracademic pursuit, we propose SRFormer, a unified DETR-based model withamalgamated Segmentation and Regression, aiming at the synergistic harnessingof the inherent robustness in segmentation representations, along with thestraightforward post-processing of instance-level regression. Our empiricalanalysis indicates that favorable segmentation predictions can be obtained atthe initial decoder layers. In light of this, we constrain the incorporation ofsegmentation branches to the first few decoder layers and employ progressiveregression refinement in subsequent layers, achieving performance gains whileminimizing computational load from the mask.Furthermore, we propose aMask-informed Query Enhancement module. We take the segmentation result as anatural soft-ROI to pool and extract robust pixel representations, which arethen employed to enhance and diversify instance queries. Extensiveexperimentation across multiple benchmarks has yielded compelling findings,highlighting our method's exceptional robustness, superior training and dataefficiency, as well as its state-of-the-art performance. Our code is availableat https://github.com/retsuh-bqw/SRFormer-Text-Det.

SRFormer: Text Detection Transformer with Incorporated Segmentation and Regression