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

Urban water-surface robust perception serves as the foundation forintelligent monitoring of aquatic environments and the autonomous navigationand operation of unmanned vessels, especially in the context of waterwaysafety. It is worth noting that current multi-sensor fusion and multi-tasklearning models consume substantial power and heavily rely on high-power GPUsfor inference. This contributes to increased carbon emissions, a concern thatruns counter to the prevailing emphasis on environmental preservation and thepursuit of sustainable, low-carbon urban environments. In light of theseconcerns, this paper concentrates on low-power, lightweight, multi-taskpanoptic perception through the fusion of visual and 4D radar data, which isseen as a promising low-cost perception method. We propose a framework namedAchelous++ that facilitates the development and comprehensive evaluation ofmulti-task water-surface panoptic perception models. Achelous++ cansimultaneously execute five perception tasks with high speed and low powerconsumption, including object detection, object semantic segmentation,drivable-area segmentation, waterline segmentation, and radar point cloudsemantic segmentation. Furthermore, to meet the demand for developers tocustomize models for real-time inference on low-performance devices, a novelmulti-modal pruning strategy known as Heterogeneous-Aware SynFlow (HA-SynFlow)is proposed. Besides, Achelous++ also supports random pruning at initializationwith different layer-wise sparsity, such as Uniform and Erdos-Renyi-Kernel(ERK). Overall, our Achelous++ framework achieves state-of-the-art performanceon the WaterScenes benchmark, excelling in both accuracy and power efficiencycompared to other single-task and multi-task models. We release and maintainthe code at https://github.com/GuanRunwei/Achelous.

Achelous++: Power-Oriented Water-Surface Panoptic Perception Framework  on Edge Devices based on Vision-Radar Fusion and Pruning of Heterogeneous  Modalities