Domain Adaptation for Object Detection
Loading...
Date
2023
Authors
Шпіганович, Владислав
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This work covers a topic of domain adaptation for object detection, reviews methods used for semi-supervised domain adaptation and researches method for domain adaptive object detection, which is based on one-stage YOLOv5, which is superior in inference time. During experiments, we evaluate possible good hyperparameter changing strategies and apply knowledge distillation based model compressing technique. The results show validity of discussed method and confirm, that knowledge transferring techniques may help in domain adaptation.
Description
Keywords
рseudo-label based self-training, YOLO model, PascalVOC, курсова робота