Efficient Oriented Object Detection with Enhanced Small Object Recognition in Aerial Images

TL;DR

This paper introduces a lightweight, efficient oriented object detection model based on YOLOv8, optimized for small object detection in aerial images, combining novel modules for enhanced feature extraction and scale handling.

Contribution

The paper presents a novel enhancement to YOLOv8 with modules like wavelet transform-based C2f and ASFP, reducing parameters while maintaining accuracy for small and multi-scale object detection in aerial imagery.

Findings

Achieves competitive mAP on DOTAv1.0 dataset

Reduces model parameters compared to DecoupleNet

Maintains high detection accuracy with improved efficiency

Abstract

Achieving a balance between computational efficiency and detection accuracy in the realm of rotated bounding box object detection within aerial imagery is a significant challenge. While prior research has aimed at creating lightweight models that enhance computational performance and feature extraction, there remains a gap in the performance of these networks when it comes to the detection of small and multi-scale objects in remote sensing (RS) imagery. To address these challenges, we present a novel enhancement to the YOLOv8 model, tailored for oriented object detection tasks and optimized for environments with limited computational resources. Our model features a wavelet transform-based C2f module for capturing associative features and an Adaptive Scale Feature Pyramid (ASFP) module that leverages P2 layer details. Additionally, the incorporation of GhostDynamicConv significantly contributes to the model's lightweight nature, ensuring high efficiency in aerial imagery analysis. Featuring a parameter count of 21.6M, our approach provides a more efficient architectural design than DecoupleNet, which has 23.3M parameters, all while maintaining detection accuracy. On the DOTAv1.0 dataset, our model demonstrates a mean Average Precision (mAP) that is competitive with leading methods such as DecoupleNet. The model's efficiency, combined with its reduced parameter count, makes it a strong candidate for aerial object detection, particularly in resource-constrained environments.