Drone-NeRF: Efficient NeRF Based 3D Scene Reconstruction for Large-Scale Drone Survey

TL;DR

Drone-NeRF introduces an efficient method for large-scale 3D scene reconstruction from drone imagery by dividing scenes into sub-blocks, parallel training, and merging, improving accuracy and speed in neural rendering.

Contribution

The paper presents a novel framework that enables scalable and accurate 3D scene reconstruction from drone images using scene partitioning, parallel NeRF training, and hash-coded fusion for large-scale environments.

Findings

Effective scene division and parallel training improve reconstruction speed.

Hash-coded fusion accelerates density representation.

Framework handles occlusion and reduces noise in large-scale scenes.

Abstract

Neural rendering has garnered substantial attention owing to its capacity for creating realistic 3D scenes. However, its applicability to extensive scenes remains challenging, with limitations in effectiveness. In this work, we propose the Drone-NeRF framework to enhance the efficient reconstruction of unbounded large-scale scenes suited for drone oblique photography using Neural Radiance Fields (NeRF). Our approach involves dividing the scene into uniform sub-blocks based on camera position and depth visibility. Sub-scenes are trained in parallel using NeRF, then merged for a complete scene. We refine the model by optimizing camera poses and guiding NeRF with a uniform sampler. Integrating chosen samples enhances accuracy. A hash-coded fusion MLP accelerates density representation, yielding RGB and Depth outputs. Our framework accounts for sub-scene constraints, reduces parallel-training noise, handles shadow occlusion, and merges sub-regions for a polished rendering result. This Drone-NeRF framework demonstrates promising capabilities in addressing challenges related to scene complexity, rendering efficiency, and accuracy in drone-obtained imagery.