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

Multi-task learning has emerged as a powerful paradigm to solve a range oftasks simultaneously with good efficiency in both computation resources andinference time. However, these algorithms are designed for different tasksmostly not within the scope of autonomous driving, thus making it hard tocompare multi-task methods in autonomous driving. Aiming to enable thecomprehensive evaluation of present multi-task learning methods in autonomousdriving, we extensively investigate the performance of popular multi-taskmethods on the large-scale driving dataset, which covers four common perceptiontasks, i.e., object detection, semantic segmentation, drivable areasegmentation, and lane detection. We provide an in-depth analysis of currentmulti-task learning methods under different common settings and find out thatthe existing methods make progress but there is still a large performance gapcompared with single-task baselines. To alleviate this dilemma in autonomousdriving, we present an effective multi-task framework, VE-Prompt, whichintroduces visual exemplars via task-specific prompting to guide the modeltoward learning high-quality task-specific representations. Specifically, wegenerate visual exemplars based on bounding boxes and color-based markers,which provide accurate visual appearances of target categories and furthermitigate the performance gap. Furthermore, we bridge transformer-based encodersand convolutional layers for efficient and accurate unified perception inautonomous driving. Comprehensive experimental results on the diverseself-driving dataset BDD100K show that the VE-Prompt improves the multi-taskbaseline and further surpasses single-task models.

Visual Exemplar Driven Task-Prompting for Unified Perception in Autonomous Driving