Aerial Image Object Detection With Vision Transformer Detector (ViTDet)

TL;DR

This paper evaluates the effectiveness of the Vision Transformer Detector (ViTDet) for aerial image object detection, demonstrating its superior performance over CNN-based methods across multiple datasets and bounding box types.

Contribution

It provides the first comprehensive study of ViTDet's application to aerial images, showing its consistent outperformance and establishing a baseline for future research.

Findings

ViTDet outperforms CNN counterparts by up to 17% in average precision.

ViTDet achieves competitive results for oriented bounding box detection.

The study offers a baseline for future aerial image detection research.

Abstract

The past few years have seen an increased interest in aerial image object detection due to its critical value to large-scale geo-scientific research like environmental studies, urban planning, and intelligence monitoring. However, the task is very challenging due to the birds-eye view perspective, complex backgrounds, large and various image sizes, different appearances of objects, and the scarcity of well-annotated datasets. Recent advances in computer vision have shown promise tackling the challenge. Specifically, Vision Transformer Detector (ViTDet) was proposed to extract multi-scale features for object detection. The empirical study shows that ViTDet's simple design achieves good performance on natural scene images and can be easily embedded into any detector architecture. To date, ViTDet's potential benefit to challenging aerial image object detection has not been explored. Therefore, in our study, 25 experiments were carried out to evaluate the effectiveness of ViTDet for aerial image object detection on three well-known datasets: Airbus Aircraft, RarePlanes, and Dataset of Object DeTection in Aerial images (DOTA). Our results show that ViTDet can consistently outperform its convolutional neural network counterparts on horizontal bounding box (HBB) object detection by a large margin (up to 17% on average precision) and that it achieves the competitive performance for oriented bounding box (OBB) object detection. Our results also establish a baseline for future research.