# Fast and Generalizable NeRF Architecture Selection for Satellite Scene Reconstruction

arXiv: https://arxiv.org/abs/2603.18306 · HTML: https://mlsanity.com/paper/2603.18306/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 29 of its 41 references that are in the map.
- First posted on arXiv 18 Mar 2026.

## Builds on (most cited first, 29 of 29)

- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [A Unified Approach to Interpreting Model Predictions](https://mlsanity.com/paper/1705.07874/) (2017)
- [Instant Neural Graphics Primitives with a Multiresolution Hash Encoding](https://mlsanity.com/paper/2201.05989/) (2022)
- [pixelNeRF: Neural Radiance Fields from One or Few Images](https://mlsanity.com/paper/2012.02190/) (2020)
- [NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collections](https://mlsanity.com/paper/2008.02268/) (2020)
- [EfficientNetV2: Smaller Models and Faster Training](https://mlsanity.com/paper/2104.00298/) (2021)
- [MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereo](https://mlsanity.com/paper/2103.15595/) (2021)
- [ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders](https://mlsanity.com/paper/2301.00808/) (2023)
- [Image Segmentation Using Deep Learning: A Survey](https://mlsanity.com/paper/2001.05566/) (2020)
- [FastNeRF: High-Fidelity Neural Rendering at 200FPS](https://mlsanity.com/paper/2103.10380/) (2021)
- [VoxFormer: Sparse Voxel Transformer for Camera-based 3D Semantic Scene Completion](https://mlsanity.com/paper/2302.12251/) (2023)
- [Urban Radiance Fields](https://mlsanity.com/paper/2111.14643/) (2021)
- [A Comprehensive Survey of Neural Architecture Search: Challenges and Solutions](https://mlsanity.com/paper/2006.02903/) (2020)
- [GRF: Learning a General Radiance Field for 3D Representation and Rendering](https://mlsanity.com/paper/2010.04595/) (2020)
- [Tree-Structured Parzen Estimator: Understanding Its Algorithm Components and Their Roles for Better Empirical Performance](https://mlsanity.com/paper/2304.11127/) (2023)
- [Neural Architecture Search: Insights from 1000 Papers](https://mlsanity.com/paper/2301.08727/) (2023)
- [Is Attention All That NeRF Needs?](https://mlsanity.com/paper/2207.13298/) (2022)
- [NerfAcc: Efficient Sampling Accelerates NeRFs](https://mlsanity.com/paper/2305.04966/) (2023)
- [Sat-NeRF: Learning Multi-View Satellite Photogrammetry With Transient Objects and Shadow Modeling Using RPC Cameras](https://mlsanity.com/paper/2203.08896/) (2022)
- [Shadow Neural Radiance Fields for Multi-view Satellite Photogrammetry](https://mlsanity.com/paper/2104.09877/) (2021)
- [WaveNeRF: Wavelet-based Generalizable Neural Radiance Fields](https://mlsanity.com/paper/2308.04826/) (2023)
- [SparseSat-NeRF: Dense Depth Supervised Neural Radiance Fields for Sparse Satellite Images](https://mlsanity.com/paper/2309.00277/) (2023)
- [SAT-NGP : Unleashing Neural Graphics Primitives for Fast Relightable Transient-Free 3D reconstruction from Satellite Imagery](https://mlsanity.com/paper/2403.18711/) (2024)
- [AutoNeRF: Training Implicit Scene Representations with Autonomous Agents](https://mlsanity.com/paper/2304.11241/) (2023)
- [Predicting Directionality in Causal Relations in Text](https://mlsanity.com/paper/2103.13606/) (2021)
- [Geometry-aware Reconstruction and Fusion-refined Rendering for Generalizable Neural Radiance Fields](https://mlsanity.com/paper/2404.17528/) (2024)
- [GO-N3RDet: Geometry Optimized NeRF-enhanced 3D Object Detector](https://mlsanity.com/paper/2503.15211/) (2025)
- [Self-Evolving Neural Radiance Fields](https://mlsanity.com/paper/2312.01003/) (2023)
- [NAS-NeRF: Generative Neural Architecture Search for Neural Radiance Fields](https://mlsanity.com/paper/2309.14293/) (2023)

## Abstract (arXiv)

Neural Radiance Fields (NeRF) have emerged as a powerful approach for photorealistic 3D reconstruction from multi-view images. However, deploying NeRF for satellite imagery remains challenging. Each scene requires individual training, and optimizing architectures via Neural Architecture Search (NAS) demands hours to days of GPU time. While existing approaches focus on architectural improvements, our SHAP analysis reveals that multi-view consistency, rather than model architecture, determines reconstruction quality. Based on this insight, we develop PreSCAN, a predictive framework that estimates NeRF quality prior to training using lightweight geometric and photometric descriptors. PreSCAN selects suitable architectures in < 30 seconds with < 1 dB prediction error, achieving 1000× speedup over NAS. We further demonstrate PreSCAN's deployment utility on edge platforms (Jetson Orin), where combining its predictions with offline cost profiling reduces inference power by 26% and latency by 43% with minimal quality loss. Experiments on DFC2019 datasets confirm that PreSCAN generalizes across diverse satellite scenes without retraining.