While recent advancement of domain adaptation techniques is significant, most of methods only align a feature extractor and do not adapt a classifier to target domain, which would be a cause of performance degradation. We propose novel domain adaptation technique for object detection that aligns prediction output space. In addition to feature alignment, we aligned predictions of locations and class confidences of our vehicle detector for satellite images by adversarial training. The proposed method significantly improved AP score by over 5%, which shows effectivity of our method for object detection tasks in satellite images.
@article{arxiv.2107.02411,
title = {Adapting Vehicle Detector to Target Domain by Adversarial Prediction Alignment},
author = {Yohei Koga and Hiroyuki Miyazaki and Ryosuke Shibasaki},
journal= {arXiv preprint arXiv:2107.02411},
year = {2021}
}
Comments
The accepted version of the article in IEEE International Geoscience and Remote Sensing Symposium (IGARSS) 2021. Copyright 2021 IEEE. Code available: https://github.com/monotaro3/vd_pred_align