Effect of Label Noise on Robustness of Deep Neural Network Object Detectors

Bishwo Adhikari, Jukka Peltomäki, Saeed Bakhshi Germi, Esa Rahtu, Heikki Huttunen

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

7 Downloads (Pure)


Label noise is a primary point of interest for safety concerns in previous works as it affects the robustness of a machine learning system by a considerable amount. This paper studies the sensitivity of object detection loss functions to label noise in bounding box detection tasks. Although label noise has been widely studied in the classification context, less attention is paid to its effect on object detection. We characterize different types of label noise and concentrate on the most common type of annotation error, which is missing labels. We simulate missing labels by deliberately removing bounding boxes at training time and study its effect on different deep learning object detection architectures and their loss functions. Our primary focus is on comparing two particular loss functions: cross-entropy loss and focal loss. We also experiment on the effect of different focal loss hyperparameter values with varying amounts of noise in the datasets and discover that even up to 50% missing labels can be tolerated with an appropriate selection of hyperparameters. The results suggest that focal loss is more sensitive to label noise, but increasing the gamma value can boost its robustness.
Original languageEnglish
Title of host publicationComputer Safety, Reliability, and Security. SAFECOMP 2021 Workshops
Subtitle of host publicationDECSoS, MAPSOD, DepDevOps, USDAI, and WAISE, York, UK, September 7, 2021, Proceedings
EditorsIbrahim Habli, Mark Sujan, Simos Gerasimou, Erwin Schoitsch, Friedemann Bitsch
Number of pages12
ISBN (Electronic)978-3-030-83906-2
ISBN (Print)978-3-030-83905-5
Publication statusPublished - 2021
Publication typeA4 Article in conference proceedings
EventInternational Conference on Computer Safety, Reliability, and Security - York, United Kingdom
Duration: 7 Sept 202110 Sept 2021

Publication series

NameLecture Notes in Computer Science
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


ConferenceInternational Conference on Computer Safety, Reliability, and Security
Country/TerritoryUnited Kingdom

Publication forum classification

  • Publication forum level 1


Dive into the research topics of 'Effect of Label Noise on Robustness of Deep Neural Network Object Detectors'. Together they form a unique fingerprint.

Cite this