Zero-Shot Aerial Object Detection with Visual Description Regularization

TL;DR

This paper introduces DescReg, a zero-shot aerial object detection method that leverages visual description regularization to improve detection of unseen classes without extensive annotations.

Contribution

It proposes a novel similarity-aware triplet loss to incorporate prior visual similarities from descriptions, addressing the semantic-visual correlation gap in zero-shot detection.

Findings

DescReg outperforms state-of-the-art ZSD methods on DIOR, xView, and DOTA datasets.

It achieves 4.5 mAP improvement on unseen classes in DIOR.

The method generalizes well across different detection architectures.

Abstract

Existing object detection models are mainly trained on large-scale labeled datasets. However, annotating data for novel aerial object classes is expensive since it is time-consuming and may require expert knowledge. Thus, it is desirable to study label-efficient object detection methods on aerial images. In this work, we propose a zero-shot method for aerial object detection named visual Description Regularization, or DescReg. Concretely, we identify the weak semantic-visual correlation of the aerial objects and aim to address the challenge with prior descriptions of their visual appearance. Instead of directly encoding the descriptions into class embedding space which suffers from the representation gap problem, we propose to infuse the prior inter-class visual similarity conveyed in the descriptions into the embedding learning. The infusion process is accomplished with a newly designed similarity-aware triplet loss which incorporates structured regularization on the representation space. We conduct extensive experiments with three challenging aerial object detection datasets, including DIOR, xView, and DOTA. The results demonstrate that DescReg significantly outperforms the state-of-the-art ZSD methods with complex projection designs and generative frameworks, e.g., DescReg outperforms best reported ZSD method on DIOR by 4.5 mAP on unseen classes and 8.1 in HM. We further show the generalizability of DescReg by integrating it into generative ZSD methods as well as varying the detection architecture.