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

Since the introduction of Vision Transformers, the landscape of many computervision tasks (e.g., semantic segmentation), which has been overwhelminglydominated by CNNs, recently has significantly revolutionized. However, thecomputational cost and memory requirement render these methods unsuitable onthe mobile device, especially for the high-resolution per-pixel semanticsegmentation task. In this paper, we introduce a new method squeeze-enhancedAxial TransFormer (SeaFormer) for mobile semantic segmentation. Specifically,we design a generic attention block characterized by the formulation of squeezeAxial and detail enhancement. It can be further used to create a family ofbackbone architectures with superior cost-effectiveness. Coupled with a lightsegmentation head, we achieve the best trade-off between segmentation accuracyand latency on the ARM-based mobile devices on the ADE20K and Cityscapesdatasets. Critically, we beat both the mobile-friendly rivals andTransformer-based counterparts with better performance and lower latencywithout bells and whistles. Beyond semantic segmentation, we further apply theproposed SeaFormer architecture to image classification problem, demonstratingthe potentials of serving as a versatile mobile-friendly backbone.

SeaFormer: Squeeze-enhanced Axial Transformer for Mobile Semantic Segmentation