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

Amidst the swift advancements in photography and sensor technologies,high-definition cameras have become commonplace in the deployment of UnmannedAerial Vehicles (UAVs) for diverse operational purposes. Within the domain ofUAV imagery analysis, the segmentation of ultra-high resolution images emergesas a substantial and intricate challenge, especially when grappling with theconstraints imposed by GPU memory-restricted computational devices. This paperdelves into the intricate problem of achieving efficient and effectivesegmentation of ultra-high resolution UAV imagery, while operating understringent GPU memory limitation. The strategy of existing approaches is todownscale the images to achieve computationally efficient segmentation.However, this strategy tends to overlook smaller, thinner, and curvilinearregions. To address this problem, we propose a GPU memory-efficient andeffective framework for local inference without accessing the context beyondlocal patches. In particular, we introduce a novel spatial-guidedhigh-resolution query module, which predicts pixel-wise segmentation resultswith high quality only by querying nearest latent embeddings with the guidanceof high-resolution information. Additionally, we present an efficientmemory-based interaction scheme to correct potential semantic bias of theunderlying high-resolution information by associating cross-image contextualsemantics. For evaluation of our approach, we perform comprehensive experimentsover public benchmarks and achieve superior performance under both conditionsof small and large GPU memory usage limitations. We will release the model andcodes in the future.

Memory-Constrained Semantic Segmentation for Ultra-High Resolution UAV Imagery