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

Traffic light recognition, as a critical component of the perception moduleof self-driving vehicles, plays a vital role in the intelligent transportationsystems. The prevalent deep learning based traffic light recognition methodsheavily hinge on the large quantity and rich diversity of training data.However, it is quite challenging to collect data in various rare scenarios suchas flashing, blackout or extreme weather, thus resulting in the imbalanceddistribution of training data and consequently the degraded performance inrecognizing rare classes. In this paper, we seek to improve traffic lightrecognition by leveraging data synthesis. Inspired by the generativeadversarial networks (GANs), we propose a novel traffic light generationapproach TL-GAN to synthesize the data of rare classes to improve traffic lightrecognition for autonomous driving. TL-GAN disentangles traffic light sequencegeneration into image synthesis and sequence assembling. In the image synthesisstage, our approach enables conditional generation to allow full control of thecolor of the generated traffic light images. In the sequence assembling stage,we design the style mixing and adaptive template to synthesize realistic anddiverse traffic light sequences. Extensive experiments show that the proposedTL-GAN renders remarkable improvement over the baseline without using thegenerated data, leading to the state-of-the-art performance in comparison withthe competing algorithms that are used for general image synthesis and dataimbalance tackling.

TL-GAN: Improving Traffic Light Recognition via Data Synthesis for Autonomous Driving