Adapting Vehicle Detector to Target Domain by Adversarial Prediction Alignment

TL;DR

This paper introduces a novel domain adaptation method for vehicle detection in satellite images that aligns prediction outputs via adversarial training, significantly improving detection accuracy.

Contribution

It presents a new approach that aligns prediction outputs, such as locations and class confidences, in addition to feature alignment, for better domain adaptation in object detection.

Findings

Achieved over 5% improvement in AP score.

Effectively aligned prediction outputs using adversarial training.

Enhanced vehicle detection accuracy in satellite images.

Abstract

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.