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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Object detection neural network models need to perform reliably in highlydynamic and safety-critical environments like automated driving or robotics.Therefore, it is paramount to verify the robustness of the detection underunexpected hardware faults like soft errors that can impact a systemsperception module. Standard metrics based on average precision produce modelvulnerability estimates at the object level rather than at an image level. Aswe show in this paper, this does not provide an intuitive or representativeindicator of the safety-related impact of silent data corruption caused by bitflips in the underlying memory but can lead to an over- or underestimation oftypical fault-induced hazards. With an eye towards safety-related real-timeapplications, we propose a new metric IVMOD (Image-wise Vulnerability Metricfor Object Detection) to quantify vulnerability based on an incorrectimage-wise object detection due to false positive (FPs) or false negative (FNs)objects, combined with a severity analysis. The evaluation of severalrepresentative object detection models shows that even a single bit flip canlead to a severe silent data corruption event with potentially critical safetyimplications, with e.g., up to (much greater than) 100 FPs generated, or up toapprox. 90% of true positives (TPs) are lost in an image. Furthermore, with asingle stuck-at-1 fault, an entire sequence of images can be affected, causingtemporally persistent ghost detections that can be mistaken for actual objects(covering up to approx. 83% of the image). Furthermore, actual objects in thescene are continuously missed (up to approx. 64% of TPs are lost). Our workestablishes a detailed understanding of the safety-related vulnerability ofsuch critical workloads against hardware faults.

Hardware faults that matter: Understanding and Estimating the safety impact of hardware faults on object detection DNNs