Enhanced Detection of Tiny Objects in Aerial Images

TL;DR

This work targets tiny object detection in aerial imagery, where one-stage detectors like YOLOv8 struggle with small targets. It introduces MoonNet, a hybrid backbone that integrates SE and CBAM attention modules and employs an alternative gating function, alongside four practical enhancement strategies: input-resolution optimization, targeted data augmentation, and careful attention deployment. Through extensive experiments on Modified DOTAv2.0 and VisDrone, the authors demonstrate that 928×928 input and calibrated augmentations substantially boost performance, and that MoonNet—especially when integrated with YOLC—achieves state-of-the-art results for cluster-aware tiny-object detection, with modest computational overhead. The findings underscore the value of hybrid attention, gating choices, and resolution strategies for robust tiny-object detection in aerial imagery, while also highlighting the need for tuning when combining multiple attention modules across different frameworks. The work provides publicly available code to reproduce the MoonNet-based enhancements.

Abstract

While one-stage detectors like YOLOv8 offer fast training speed, they often under-perform on detecting small objects as a trade-off. This becomes even more critical when detecting tiny objects in aerial imagery due to low-resolution targets and cluttered backgrounds. To address this, we introduce four enhancement strategies-input image resolution adjustment, data augmentation, attention mechanisms, and an alternative gating function for attention modules-that can be easily implemented on YOLOv8. We demonstrate that image size enlargement and the proper use of augmentation can lead to enhancement. Additionally, we designed a Mixture of Orthogonal Neural-modules Network (MoonNet) pipeline which consists of multiple attention-module-augmented CNNs. Two well-known attention modules, Squeeze-and-Excitation (SE) Block and Convolutional Block Attention Module (CBAM), were integrated into the backbone of YOLOv8 to form the MoonNet design, and the MoonNet backbone obtained improved detection accuracy compared to the original YOLOv8 backbone and single-type attention-module-augmented backbones. MoonNet further proved its adaptability and potential by achieving state-of-the-art performance on a tiny-object benchmark when integrated with the YOLC model. Our code is available at: https://github.com/Kihyun11/MoonNet

Figures (3)

• Figure 1: Object categories of the Modified-DOTA
• Figure 2: Backbone designs
• Figure 3: Visual comparison of detection performance on tiny objects across two different aerial scenes.