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

ML-enabled software systems have been incorporated in many publicdemonstrations for automated driving (AD) systems. Such solutions have alsobeen considered as a crucial approach to aim at SAE Level 5 systems, where thepassengers in such vehicles do not have to interact with the system at allanymore. Already in 2016, Nvidia demonstrated a complete end-to-end approachfor training the complete software stack covering perception, planning anddecision making, and the actual vehicle control. While such approaches show thegreat potential of such ML-enabled systems, there have also been demonstrationswhere already changes to single pixels in a video frame can potentially lead tocompletely different decisions with dangerous consequences. In this paper, astructured analysis has been conducted to explore video degradation effects onthe performance of an ML-enabled pedestrian detector. Firstly, a baseline ofapplying YOLO to 1,026 frames with pedestrian annotations in the KITTI VisionBenchmark Suite has been established. Next, video degradation candidates foreach of these frames were generated using the leading video codecs libx264,libx265, Nvidia HEVC, and AV1: 52 frames for the various compression presetsfor color and gray-scale frames resulting in 104 degradation candidates peroriginal KITTI frame and 426,816 images in total. YOLO was applied to eachimage to compute the intersection-over-union (IoU) metric to compare theperformance with the original baseline. While aggressively lossy compressionsettings result in significant performance drops as expected, it was alsoobserved that some configurations actually result in slightly better IoUresults compared to the baseline. The findings show that carefully chosen lossyvideo configurations preserve a decent performance of particular ML-enabledsystems while allowing for substantial savings when storing or transmittingdata.

A Structured Analysis of the Video Degradation Effects on the Performance of a Machine Learning-enabled Pedestrian Detector