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 accident prediction in driving videos aims to provide an earlywarning of the accident occurrence, and supports the decision making of safedriving systems. Previous works usually concentrate on the spatial-temporalcorrelation of object-level context, while they do not fit the inherentlong-tailed data distribution well and are vulnerable to severe environmentalchange. In this work, we propose a Cognitive Accident Prediction (CAP) methodthat explicitly leverages human-inspired cognition of text description on thevisual observation and the driver attention to facilitate model training. Inparticular, the text description provides a dense semantic description guidancefor the primary context of the traffic scene, while the driver attentionprovides a traction to focus on the critical region closely correlating withsafe driving. CAP is formulated by an attentive text-to-vision shift fusionmodule, an attentive scene context transfer module, and the driver attentionguided accident prediction module. We leverage the attention mechanism in thesemodules to explore the core semantic cues for accident prediction. In order totrain CAP, we extend an existing self-collected DADA-2000 dataset (withannotated driver attention for each frame) with further factual textdescriptions for the visual observations before the accidents. Besides, weconstruct a new large-scale benchmark consisting of 11,727 in-the-wild accidentvideos with over 2.19 million frames (named as CAP-DATA) together with labeledfact-effect-reason-introspection description and temporal accident frame label.Based on extensive experiments, the superiority of CAP is validated comparedwith state-of-the-art approaches. The code, CAP-DATA, and all results will bereleased in \url{https://github.com/JWFanggit/LOTVS-CAP}.

Cognitive Accident Prediction in Driving Scenes: A Multimodality Benchmark