A DeNoising FPN With Transformer R-CNN for Tiny Object Detection

TL;DR

This paper introduces DNTR, a novel framework combining denoising feature pyramid networks and transformer-based R-CNN to significantly improve tiny object detection accuracy in remote sensing images.

Contribution

The paper proposes a new plug-in denoising FPN and a transformer-based R-CNN, enhancing tiny object detection by reducing noise and focusing on small object features.

Findings

Outperforms baselines by at least 17.4% in APvt on AI-TOD dataset.

Achieves 9.6% higher AP on VisDrone dataset.

Demonstrates effectiveness of denoising and transformer modules in tiny object detection.

Abstract

Despite notable advancements in the field of computer vision, the precise detection of tiny objects continues to pose a significant challenge, largely owing to the minuscule pixel representation allocated to these objects in imagery data. This challenge resonates profoundly in the domain of geoscience and remote sensing, where high-fidelity detection of tiny objects can facilitate a myriad of applications ranging from urban planning to environmental monitoring. In this paper, we propose a new framework, namely, DeNoising FPN with Trans R-CNN (DNTR), to improve the performance of tiny object detection. DNTR consists of an easy plug-in design, DeNoising FPN (DN-FPN), and an effective Transformer-based detector, Trans R-CNN. Specifically, feature fusion in the feature pyramid network is important for detecting multiscale objects. However, noisy features may be produced during the fusion process since there is no regularization between the features of different scales. Therefore, we introduce a DN-FPN module that utilizes contrastive learning to suppress noise in each level's features in the top-down path of FPN. Second, based on the two-stage framework, we replace the obsolete R-CNN detector with a novel Trans R-CNN detector to focus on the representation of tiny objects with self-attention. Experimental results manifest that our DNTR outperforms the baselines by at least 17.4% in terms of APvt on the AI-TOD dataset and 9.6% in terms of AP on the VisDrone dataset, respectively. Our code will be available at https://github.com/hoiliu-0801/DNTR.