AdAPT: Advertisement detector adaptation under newspaper domain shift with null-based pseudo-labelingShow others and affiliations
2026 (English)In: Machine Learning with Applications, E-ISSN 2666-8270, Vol. 23, article id 100806Article in journal (Refereed) Published
Abstract [en]
Detecting advertisements in digitized newspapers is a key step in large-scale media analytics and digital archiving. However, variations in layout, typography, and advertisement design across publishers and time periods cause significant domain shifts that reduce the generalization ability of supervised detectors. This paper presents AdAPT, a confidence-guided pseudo-labeling pipeline for unsupervised domain adaptation in advertisement detection. The proposed method leverages both advertisement-free (Null) and advertisement-containing pages from unlabeled target domains to generate reliable pseudo-labels. By retraining a YOLO-based detector using labeled source data combined with filtered pseudo-labeled target samples, AdAPT achieves robust adaptation without requiring manual annotation. Experiments conducted on two unseen newspapers (Adresseavisen and iTromsø) demonstrate that Null-based pseudo-labeling provides the most stable and accurate adaptation, yielding up to 38% error reduction compared to the baseline. The results highlight AdAPT as a simple, scalable, and annotation-efficient solution for maintaining high-performance advertisement detection across diverse newspaper collections.
Place, publisher, year, edition, pages
Elsevier, 2026. Vol. 23, article id 100806
Keywords [en]
Cross-domain advertisement detection, Deep learning, Object detection, Domain adaptation, Pseudo labeling
National Category
Computer Sciences Computer graphics and computer vision
Research subject
Virtual Production Development (VPD)
Identifiers
URN: urn:nbn:se:his:diva-26067DOI: 10.1016/j.mlwa.2025.100806ISI: 001635811600001OAI: oai:DiVA.org:his-26067DiVA, id: diva2:2021634
Funder
Knowledge FoundationMid Sweden University
Note
CC BY 4.0
Corresponding author: Faeze Zakaryapour Sayyad
Received 16 October 2025, Revised 21 November 2025, Accepted 30 November 2025, Available online 1 December 2025, Version of Record 3 December 2025.
Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Faeze Zakaryapour Sayyad reports financial support was provided by Mid Sweden University. Faeze Zakaryapour Sayyad reports a relationship with Knowledge Foundation (kks.se) within the Industrial graduate school Smart Industry Sweden, Media Research company that includes: employment and funding grants. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgment
This work was supported in part by The Knowledge Foundation(kks.se) within the Industrial graduate school Smart Industry Sweden and Media Research AB. The authors would also like to thank Oscar Berg for his valuable comments on this work.
2025-12-152025-12-152025-12-30Bibliographically approved