< Back to previous page

Publication

Object detection with semi-supervised adversarial domain adaptation for real-time edge devices

Book Contribution - Book Abstract Conference Contribution

Object detection on real-time edge devices for new applications with no or a limited amount of annotated labels is difficult. Where traditional data-hungry methods fail, transfer learning can provide a solution by transferring knowledge from a source domain to the target application domain. We explore domain adaptation techniques on a one-stage detection architecture, i.e. YOLOv3, which enables use on edge devices. Existing methods in domain adaptation with deep learning for object detection, use two-stage detectors like Faster-RCNN with adversarial adaptation. By using a one-stage detector, the speed increases by a factor of eight. With our proposed method, we reduce by 28% the changes in performance introduced by the gap between the source and target domains.
Book: Artificial Intelligence and Machine Learning : 33rd Benelux Conference on Artificial Intelligence, BNAIC/Benelearn 2021, Esch-sur-Alzette, Luxembourg, November 10–12, 2021: revised selected papers
Pages: 86 - 102
ISBN:978-3-030-93841-3
Publication year:2022
Keywords:P1 Proceeding
Accessibility:Closed