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

Distracted driving is one of the major reasons for vehicle accidents.Therefore, detecting distracted driving behaviors is of paramount importance toreduce the millions of deaths and injuries occurring worldwide. Distracted oranomalous driving behaviors are deviations from 'normal' driving that need tobe identified correctly to alert the driver. However, these driving behaviorsdo not comprise one specific type of driving style and their distribution canbe different during the training and test phases of a classifier. We formulatethis problem as a supervised contrastive learning approach to learn a visualrepresentation to detect normal, and seen and unseen anomalous drivingbehaviors. We made a change to the standard contrastive loss function to adjustthe similarity of negative pairs to aid the optimization. Normally, in a (self)supervised contrastive framework, the projection head layers are omitted duringthe test phase as the encoding layers are considered to contain general visualrepresentative information. However, we assert that for a video-basedsupervised contrastive learning task, including a projection head can bebeneficial. We showed our results on a driver anomaly detection dataset thatcontains 783 minutes of video recordings of normal and anomalous drivingbehaviors of 31 drivers from the various top and front cameras (both depth andinfrared). Out of 9 video modalities combinations, our proposed contrastiveapproach improved the ROC AUC on 6 in comparison to the baseline models (from4.23% to 8.91% for different modalities). We performed statistical tests thatshowed evidence that our proposed method performs better than the baselinecontrastive learning setup. Finally, the results showed that the fusion ofdepth and infrared modalities from the top and front views achieved the bestAUC ROC of 0.9738 and AUC PR of 0.9772.

Supervised Contrastive Learning for Detecting Anomalous Driving Behaviours from Multimodal Videos