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

High-Definition (HD) maps are pivotal to autopilot navigation. Integratingthe capability of lightweight HD map construction at runtime into aself-driving system recently emerges as a promising direction. In this surge,vision-only perception stands out, as a camera rig can still perceive thestereo information, let alone its appealing signature of portability andeconomy. The latest MapTR architecture solves the online HD map constructiontask in an end-to-end fashion but its potential is yet to be explored. In thiswork, we present a full-scale upgrade of MapTR and propose MapNeXt, the nextgeneration of HD map learning architecture, delivering major contributions fromthe model training and scaling perspectives. After shedding light on thetraining dynamics of MapTR and exploiting the supervision from map elementsthoroughly, MapNeXt-Tiny raises the mAP of MapTR-Tiny from 49.0% to 54.8%,without any architectural modifications. Enjoying the fruit of map segmentationpre-training, MapNeXt-Base further lifts the mAP up to 63.9% that has alreadyoutperformed the prior art, a multi-modality MapTR, by 1.4% while being$\sim1.8\times$ faster. Towards pushing the performance frontier to the nextlevel, we draw two conclusions on practical model scaling: increased queryfavors a larger decoder network for adequate digestion; a large backbonesteadily promotes the final accuracy without bells and whistles. Building uponthese two rules of thumb, MapNeXt-Huge achieves state-of-the-art performance onthe challenging nuScenes benchmark. Specifically, we push the maplessvision-only single-model performance to be over 78% for the first time,exceeding the best model from existing methods by 16%.

MapNeXt: Revisiting Training and Scaling Practices for Online Vectorized HD Map Construction