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/

Zendy Plus

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

Zendy Tools

Zendy Open

Understanding actions within surgical workflows is essential for evaluatingpost-operative outcomes. However, capturing long sequences of actions performedin surgical settings poses challenges, as individual surgeons have their uniqueapproaches shaped by their expertise, leading to significant variability. Totackle this complex problem, we focused on segmentation with preciseboundaries, a demanding task due to the inherent variability in actiondurations and the subtle transitions often observed in untrimmed videos. Thesetransitions, marked by ambiguous starting and ending points, complicate thesegmentation process. Traditional models, such as MS-TCN, which depend on largereceptive fields, frequently face challenges of over-segmentation (resulting infragmented segments) or under-segmentation (merging distinct actions). Both ofthese issues negatively impact the quality of segmentation. To overcome thesechallenges, we present the Multi-Stage Boundary-Aware Transformer Network(MSBATN) with hierarchical sliding window attention, designed to enhance actionsegmentation. Our proposed approach incorporates a novel unified loss functionthat treats action classification and boundary detection as distinct yetinterdependent tasks. Unlike traditional binary boundary detection methods, ourboundary voting mechanism accurately identifies start and end points byleveraging contextual information. Extensive experiments using threechallenging surgical datasets demonstrate the superior performance of theproposed method, achieving state-of-the-art results in F1 scores at thresholdsof 25% and 50%, while also delivering comparable performance in other metrics.

Multi-Stage Boundary-Aware Transformer Network for Action Segmentation in Untrimmed Surgical Videos