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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Multi-task learning is commonly used in autonomous driving for solvingvarious visual perception tasks. It offers significant benefits in terms ofboth performance and computational complexity. Current work on multi-tasklearning networks focus on processing a single input image and there is noknown implementation of multi-task learning handling a sequence of images. Inthis work, we propose a multi-stream multi-task network to take advantage ofusing feature representations from preceding frames in a video sequence forjoint learning of segmentation, depth, and motion. The weights of the currentand previous encoder are shared so that features computed in the previous framecan be leveraged without additional computation. In addition, we propose to usethe geometric mean of task losses as a better alternative to the weightedaverage of task losses. The proposed loss function facilitates better handlingof the difference in convergence rates of different tasks. Experimental resultson KITTI, Cityscapes and SYNTHIA datasets demonstrate that the proposedstrategies outperform various existing multi-task learning solutions.

MultiNet++: Multi-Stream Feature Aggregation and Geometric Loss Strategy for Multi-Task Learning