Unsupervised Domain Adaptation from Synthetic to Real Images for Anchorless Object Detection

TL;DR

This paper explores unsupervised domain adaptation techniques to improve the performance of anchorless object detectors, specifically CenterNet, when trained on synthetic images and applied to real images, achieving significant mAP improvements.

Contribution

It adapts entropy minimization and maximum squares loss methods from segmentation to anchorless object detection, enhancing synthetic-to-real domain transfer.

Findings

mAP increased from 61% to 69% with UDA methods

Anchorless detectors like CenterNet are effective for domain adaptation

Proposed methods outperform direct transfer baseline

Abstract

Synthetic images are one of the most promising solutions to avoid high costs associated with generating annotated datasets to train supervised convolutional neural networks (CNN). However, to allow networks to generalize knowledge from synthetic to real images, domain adaptation methods are necessary. This paper implements unsupervised domain adaptation (UDA) methods on an anchorless object detector. Given their good performance, anchorless detectors are increasingly attracting attention in the field of object detection. While their results are comparable to the well-established anchor-based methods, anchorless detectors are considerably faster. In our work, we use CenterNet, one of the most recent anchorless architectures, for a domain adaptation problem involving synthetic images. Taking advantage of the architecture of anchorless detectors, we propose to adjust two UDA methods, viz., entropy minimization and maximum squares loss, originally developed for segmentation, to object detection. Our results show that the proposed UDA methods can increase the mAPfrom61 %to69 %with respect to direct transfer on the considered anchorless detector. The code is available: https://github.com/scheckmedia/centernet-uda.