DroneKey++: A Size Prior-free Method and New Benchmark for Drone 3D Pose Estimation from Sequential Images

TL;DR

DroneKey++ introduces a size prior-free method for drone 3D pose estimation from sequential images, along with a large synthetic benchmark, enabling improved generalization and real-time performance without relying on prior drone size information.

Contribution

It proposes a novel prior-free framework for drone 3D pose estimation and introduces a large-scale synthetic dataset for robust evaluation.

Findings

Achieves low MAE and MedAE in rotation and translation estimation.

Demonstrates strong generalization across multiple drone models.

Operates in real-time with high inference speeds.

Abstract

Accurate 3D pose estimation of drones is essential for security and surveillance systems. However, existing methods often rely on prior drone information such as physical sizes or 3D meshes. At the same time, current datasets are small-scale, limited to single models, and collected under constrained environments, which makes reliable validation of generalization difficult. We present DroneKey++, a prior-free framework that jointly performs keypoint detection, drone classification, and 3D pose estimation. The framework employs a keypoint encoder for simultaneous keypoint detection and classification, and a pose decoder that estimates 3D pose using ray-based geometric reasoning and class embeddings. To address dataset limitations, we construct 6DroneSyn, a large-scale synthetic benchmark with over 50K images covering 7 drone models and 88 outdoor backgrounds, generated using 360-degree panoramic synthesis. Experiments show that DroneKey++ achieves MAE 17.34 deg and MedAE 17.1 deg for rotation, MAE 0.135 m and MedAE 0.242 m for translation, with inference speeds of 19.25 FPS (CPU) and 414.07 FPS (GPU), demonstrating both strong generalization across drone models and suitability for real-time applications. The dataset is publicly available.