Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Tile and Slide : A New Framework for Scaling NeRF from Local to Global 3D Earth Observation

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that large satellite 3D scenes can be reconstructed on a single GPU in linear time with constant memory and no quality loss, using a tile-and-slide NeRF pipeline.

desk verdict A coherent tiling framework for single-GPU satellite NeRF, but the linear-time/no-compromise claims are asserted rather than demonstrated. read the letter →

arxiv 2507.01631 v2 pith:CBNXAN5N submitted 2025-07-02 cs.CV cs.AIcs.GRcs.LG

Camille Billouard , Dawa Derksen , Alexandre Constantin , Bruno Vallet This is my paper · ORCID
classification cs.CVcs.AIcs.GRcs.LG
keywords neuralradiancefieldssatelliteimagery3Dreconstructionlarge-scaletilingsingleGPUtrainingout-of-coreslidingwindowsegmentedraysampling
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Snake-NeRF is a scaling framework that lets a single GPU train a radiance field over satellite scenes far larger than GPU memory, by cutting the region of interest into non-overlapping 3D tiles and training a small set of NeRFs at a time. The paper's central claim is that this tiled pipeline runs in time linear in the number of tiles (hence in surface area when image count and resolution are held fixed) with memory bounded by a fixed budget, while matching the novel-view-synthesis and depth quality of an untiled reference NeRF. If true, large-area 3D Earth observation no longer requires multi-GPU clusters or stitching overlapping reconstructions. The two mechanisms credited for the result are a 2x2 sliding snake window that keeps all intersecting NeRFs in memory per ray, and a segmented ray sampler that prevents density errors from being hallucinated at tile boundaries.

What carries the argument

The load-bearing object is the 2x2 train-and-slide window over a regular UTM-aligned grid of non-overlapping 3D tiles. A snake-shaped traversal moves the window so that only two adjacent NeRFs are loaded or unloaded per step; each NeRF is trained once, twice, or four times depending on whether it is a corner, edge, or central tile, and the optimizer is never reset, so previously seen rays act as a continual-learning recall that prevents catastrophic forgetting. Two supporting mechanisms do the precision work: the segmented ray sampler, which computes near/far bounds per tile and samples each ray segment separately so density is learned right up to the shared boundary, and a global color network shared across tiles (with per-tile multi-resolution hash features and density networks) that regularizes appearance.

What would settle it

Run Snake-NeRF on regions of area A and 2A using the same set of images and resolution, and record total training time and peak GPU memory; if time grows faster than linearly in area, or memory rises with area, the central scaling claim fails. As a second check, render a depth map across a tile boundary and look for the thin wall of hallucinated matter that the segmented sampler is designed to remove.

Watch

Extended reading notes

Core claim

The paper discovers that tiling artifacts and memory limits can be overcome together by training non-overlapping NeRFs in a 2x2 sliding window that traverses the scene in a snake pattern. Each window loads exactly four NeRFs plus the image crops covering them, so every training ray has all the NeRFs it intersects in memory and no gradient is propagated onto unloaded models. Rays are sampled separately within each bounding-box segment rather than along the whole ray, which removes the thin wall of hallucinated matter that otherwise appears at tile edges. With this construction, the authors report PSNR/SSIM and depth MAE close to the single-model reference on four test scenes, and argue the time and memory costs satisfy their three scalability conditions of $\mathcal{O}(N_{\mathrm{nerfs}})$ time, $\mathcal{O}(1)$ memory, and quality similar to the unscaled algorithm.

Load-bearing premise

The paper's time and memory guarantees assume the number of satellite images and the spatial resolution stay fixed while the mapped area grows, so the cost of covering the enlarged area with new imagery is not part of the claimed linear scaling.

Editorial extensions

If this is right

  • A 10 km by 10 km scene at 30 cm resolution, estimated to need roughly 600 GB of NeRF weights and 1.4 TB of ray data, becomes trainable on a single GPU because only four tiles and their image crops are resident at once.
  • Scaling time grows linearly with the number of tiles, so doubling the mapped area roughly doubles training time when the image set and resolution are unchanged.
  • Non-overlapping 3D tiles with overlapping image crops remove the need to blend or stitch overlapping reconstructions, eliminating blur and halo artifacts from weighted averages.
  • The framework is architecture-agnostic: the authors state it can wrap satellite NeRF variants such as shadow, transient-object, and seasonal models without changing their loss functions.
  • An open-source implementation with on-the-fly ray computation from RPCs avoids storing all ray origins and directions, cutting memory further.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The linear-time result fixes the number of images $N_{\mathrm{im}}$ and spatial resolution SR while only the area $A_{\mathrm{ROI}}$ grows; if new imagery is added as the area grows, total input pixels grow and the per-tile trade-off shifts, so the claim should be read as area scaling with a fixed image collection rather than scaling the full input pixel count.
  • The snake-window plus segmented-sampler recipe is a general continual-learning pattern: any tiled neural field that must respect domain boundaries could adopt per-segment sampling and a sliding 2x2 recall window, independent of NeRF-specific rendering.
  • A direct testable extension would be to run the released code on a growing-area sequence with fixed imagery and record wall-clock time and peak memory, something the paper could not do because no public large-area multi-view satellite benchmark exists.
  • The quality-equivalence claim is demonstrated on four small urban scenes against a single-GPU reference; extrapolating it to global-scale terrain with strong relief or water would require running the same comparison where the reference can still be trained.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The manuscript introduces Snake-NeRF, an out-of-core framework for training multiple non-overlapping NeRF tiles on a single GPU for satellite 3D reconstruction. The method partitions the ROI into a regular grid of 3D tiles, crops input images via RPC projection so each tile sees all necessary pixels, trains four tiles at a time in a 2x2 sliding window that follows a snake path, and uses a segmented ray sampler to handle rays crossing tile boundaries. The authors claim that this achieves linear time, constant memory, and quality equivalent to an untiled reference NeRF, and support the claim with comparisons on four small scenes from the 2019 Data Fusion Contest, plus an open-source implementation that they state will be released.

Significance. The core idea—non-overlapping 3D tiling combined with an out-of-core sliding-window schedule and a segmented sampler—is a sensible and potentially practical contribution to large-scale satellite NeRF reconstruction. If the scalability claims were verified, the work would fill a clear gap, since prior large-scale NeRF methods either require multiple GPUs (NeRF-XL) or do not address single-GPU memory constraints. The paper also honestly identifies catastrophic forgetting as a central challenge and provides a straightforward recall mechanism for local tile parameters. However, the current evidence is insufficient to support the abstract's 'no compromise in quality' and 'linear time complexity' claims; the experiments are small-scale, single-run, and lack runtime and memory measurements.

major comments (4)
  1. [Section 5.4, Table 1] The claim that Snake-NeRF matches reference quality is based on four small scenes, with a single training run per configuration and no error bars. Relative PSNR values range from 21.80 to 26.14, and the JAX 214 grid3x3 result (21.80) is notably weaker than the others; without repeated seeds and larger grids, the 'without compromise in quality' claim is not established.
  2. [Section 5.2, Eq. (time)] The linear-time argument time = tit Nnerfs Nit = tit HW nit is an identity once iterations per NeRF are fixed; the actual claim depends on the unproven postulate that the optimal HW grows proportionally with A_ROI. No runtime measurements are reported, and the complexity analysis fixes N_im while A_ROI grows, so it excludes the common scenario where new imagery is added as the area increases; in that scenario the method is not linear in total input pixels.
  3. [Section 4.4 and Section 3] The shared color network is trained throughout the entire traversal, but after a tile's last visit its rays are never replayed to that network. The paper's own catastrophic-forgetting experiment (Section 3) shows that continuous training on new regions degrades earlier regions; in a large grid this could silently degrade early-departed tiles even though their local geometry is frozen. Table 1 cannot detect this effect because 3x3 and 4x4 grids have few post-departure updates; the authors should test larger grids and report per-tile quality as a function of traversal position, or add a replay mechanism for departed tiles.
  4. [Section 5.2, memory condition] The constant-memory condition is asserted as 'theoretically verified' because only four NeRFs and ray subsets are loaded, but no peak-memory measurements or ablations on tile size are provided; since constant memory is one of the three stated scalability conditions and appears in the abstract, it should be empirically demonstrated with memory traces for increasing grid sizes.
minor comments (5)
  1. [Section 4.2] The assumption that minimum and maximum Z values are known in advance and equal for all tiles should be stated in the limitations, as it may not hold for arbitrary large ROIs.
  2. [Section 5.3 and Table 1] The definition of 'relative' PSNR, SSIM, and MAE in Table 1 is not clearly explained; specify whether the reference algorithm's output is used as ground truth and how the metrics are aggregated over pixels or images.
  3. [Section 5.4] The text says the NVS results 'closely resemble the unscaled version' and then reports that Snake-NeRF outperforms the unscaled version by a large margin; the wording should be corrected to avoid confusion.
  4. [Abstract and Section 7] The abstract's claim that 'large satellite images can effectively be processed' goes beyond the experimental scope, which is limited to small areas; the limitations section acknowledges this, but the abstract should be aligned with the evidence presented.
  5. [Figure 10 and Section 5.4] The claim that differences are concentrated in ambiguous regions (water, shadows, transient objects) is made visually; a quantitative per-pixel uncertainty or error map would make this claim more convincing.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the core scaling claims are supported by an explicit algorithmic construction and by comparisons against an untiled reference, not by self-citation or fitted predictions.

full rationale

The paper's time-complexity statement in Sec. 5.2 is an analytic identity of the proposed loop structure: with a fixed number of iterations per tile, time = t_it * H * W * n_it, and equal-sized tiling makes H*W proportional to area. This is an explicit construction, labeled a postulate, not an empirical prediction. Memory O(1) follows from loading only four NeRFs and ray subsets. The 'no compromise in quality' claim is tested against an untiled reference on four DFC2019 scenes (Table 1, Figs. 9-10), so it is not manufactured by definition. Self-citations (SAT-NGP [6], scalability definitions [12,25]) are to published, externally usable methods and are not used to justify the correctness of the tiling mechanism. The acknowledged limitations--small-area datasets and the absence of large-area public benchmarks--are empirical scope limitations, not circular reasoning. The potential drift of the shared color network after a tile's last visit is a plausible scaling risk raised by the paper's own continual-learning discussion, but it is a limitation, not a circular step.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central scaling claims rest on domain assumptions about imaging geometry and dataset composition, plus an unstated set of training hyperparameters. No new physical or representational entities are proposed; the contributions are procedural (tiling, window, sampler).

free parameters (3)
  • Tiling grid dimensions (H, W) = 3x3 and 4x4 in experiments
    The number of tiles per row and column is chosen by the authors (Section 5.2, Table 1). It directly sets N_nerfs and therefore the claimed linear-time count.
  • Iterations per NeRF (n_it) = not reported
    Section 5.2 fixes n_it per tile to make time linear in the number of tiles, but the paper never states the value, so the time claim cannot be independently checked.
  • Image crop overlap margin = not reported
    Section 4.3 crops images to cover tile projections with overlap; the margin width is unspecified and trades memory against edge quality.
assumptions (6)
  • standard math RPC camera model correctly projects 3D points to image coordinates.
    Used in Section 4.3 to build image crops from 3D tile corners; the method inherits the standard photogrammetric model for WorldView-3 data.
  • domain assumption In overhead imagery, each ray intersects at most 3 tiles.
    Section 4.4: 'given the low incidence angle of the satellite, each ray can intersect at most 3 tiles.' The 2x2 window and shared-ray exclusion policy rely on this bound.
  • domain assumption Minimum and maximum Z of the scene are known in advance and are equal across all tiles.
    Section 4.2 states this simplification; tile bounding boxes and image crops are defined from these altitude bounds.
  • domain assumption Number of input images N_im stays constant while the mapped area grows.
    Section 1 fixes N_im and SR while scaling A_ROI; this is needed for the claimed linear time in area and is not how real large-area datasets are typically acquired.
  • ad hoc to paper Optimal number of tiles H times W grows proportionally with area.
    Section 5.2 postulates this proportionality, which converts the fixed-iteration-per-tile argument into the linear-time claim.
  • domain assumption Revisiting previously seen rays during later window positions prevents catastrophic forgetting.
    Section 4.4 relies on a recall strategy from continual learning; if this fails, edge quality and stability would degrade. The paper does not provide a dedicated ablation for this mechanism.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tile and Slide : A New Framework for Scaling NeRF from Local to Global 3D Earth Observation." pith.science (2026). https://pith.science/paper/CBNXAN5N

@misc{pith2026250701631,
  author       = {Pith},
  title        = {Pith review of: Tile and Slide : A New Framework for Scaling NeRF from Local to Global 3D Earth Observation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CBNXAN5N}},
  note         = {Machine review of arXiv:2507.01631}
}
abstract

Neural Radiance Fields (NeRF) have recently emerged as a paradigm for 3D reconstruction from multiview satellite imagery. However, state-of-the-art NeRF methods are typically constrained to small scenes due to the memory footprint during training, which we study in this paper. Previous work on large-scale NeRFs palliate this by dividing the scene into NeRFs. This paper introduces Snake-NeRF, a framework that scales to large scenes. Our out-of-core method eliminates the need to load all images and networks simultaneously, and operates on a single device. We achieve this by dividing the region of interest into NeRFs that 3D tile without overlap. Importantly, we crop the images with overlap to ensure each NeRFs is trained with all the necessary pixels. We introduce a novel $2\times 2$ 3D tile progression strategy and segmented sampler, which together prevent 3D reconstruction errors along the tile edges. Our experiments conclude that large satellite images can effectively be processed with linear time complexity, on a single GPU, and without compromise in quality.

Figures

Figures reproduced from arXiv: 2507.01631 by the authors.

Figure 1
Figure 1. We propose a new framework, to scale up NeRFs without scaling up hardware requirements by an out-of-core strategy. Top row [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Catastrophic forgetting on the scene JAX 214 from [26]. We train a NeRF by sampling only from rays in one area at a time. Each time we train a new area, the NeRF progressively forgets the other areas. This experiment shows an example on JAX 214 scene where once the image crops of a trained area are unloaded, the NeRF is unable to learn new areas without forgetting the previously learned ones. We propose to overcome … view at source ↗
Figure 3
Figure 3. Individual non-overlapping tiles are defined relative to [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 5
Figure 5. Figure 5: Rays (purple arrows) intersecting only bounding box [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 4
Figure 4. Figure 4: View from above, orange ray intersecting a maxi￾mum of 3 NeRFs. In this scenario, if the shared rays are included as training rays, they will not be able to propagate any gradient onto the neighboring NeRFs that are not loaded in mem￾ory. This means that objects out￾si…
Figure 8
Figure 8. Figure 8: Left: novel view synthesis, Right: depth map. When training with uniform sampling (NeRF-XL, [28]), we observe a thin wall of hallucinated matter along the edges of the tiles. 4.6. Implementation We release an open-source highly parallelized implementa￾tion of our metho…
Figure 7
Figure 7. Figure 7: Ray segmentation before sampling (left) ensuring do￾main continuity, in contrast to a ray that is sampled and then dis￾tributed based on its belonging to the NeRF (right). Ni for near i, Fi for far i and Si for segment i [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 9
Figure 9. Figure 9: The reference algorithm (Top), utilizes a single model to learn the entire scene. Errors are caused by shadows and tran￾sient effects. The unscaled case (Middle) where a single NeRFs is trained at a time, demonstrates significant artefacts along the 3D tile edges. Snak…
Figure 10
Figure 10. Figure 10: Differences between depth maps considering the ref [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SwiftGS: Episodic Priors for Immediate Satellite Surface Recovery

    cs.CV 2026-03 unverdicted novelty 6.0 of 10

    SwiftGS uses episodic meta-training to predict geometry-radiation-decoupled Gaussian primitives and a lightweight SDF for zero-shot 3D satellite surface reconstruction with physics-aware rendering.

Reference graph

Works this paper leans on

65 extracted references · 65 canonical work pages · cited by 1 Pith paper

  1. [1]

    Design and development of a neural surface rendering model for lunar satellite photogrammetry

    Caleb Adams, Ignacio Lopez-Francos, Ariel Deutsch, Ellemieke Van Kints, and Aiden Hammond. Design and development of a neural surface rendering model for lunar satellite photogrammetry. image, 110:50, 2024. 2

    work page 2024

  2. [2]

    Gaussian splatting for efficient satellite image photogram- metry

    Luca Savant Aira, Gabriele Facciolo, and Thibaud Ehret. Gaussian splatting for efficient satellite image photogram- metry. In 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2025), 2025. 2

    work page 2025

  3. [3]

    Robust rational polynomial camera modelling for sar and pushbroom imaging

    Roland Akiki, Roger Mar ´ı, Carlo De Franchis, Jean-Michel Morel, and Gabriele Facciolo. Robust rational polynomial camera modelling for sar and pushbroom imaging. In 2021 IEEE International Geoscience and Remote Sensing Sympo- sium IGARSS, pages 7908–7911, 2021. 5

    work page 2021

  4. [4]

    Sundial: 3d satellite understand- ing through direct ambient and complex lighting decomposi- tion

    Nikhil Behari, Akshat Dave, Kushagra Tiwary, William Yang, and Ramesh Raskar. Sundial: 3d satellite understand- ing through direct ambient and complex lighting decomposi- tion. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, pages 522–532, 2024. 3, 8

    work page 2024

  5. [5]

    The Ames Stereo Pipeline: NASA’s open source software for de- riving and processing terrain data

    Ross A Beyer, Oleg Alexandrov, and Scott McMichael. The Ames Stereo Pipeline: NASA’s open source software for de- riving and processing terrain data. Earth and Space Science , 5(9):537–548, 2018. 4

    work page 2018

  6. [6]

    SAT-NGP: Unleashing neural graphics primitives for fast relightable transient-free 3D reconstruc- tion from satellite imagery

    Camille Billouard, Dawa Derksen, Emmanuelle Sarrazin, and Bruno Vallet. SAT-NGP: Unleashing neural graphics primitives for fast relightable transient-free 3D reconstruc- tion from satellite imagery. In IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium , pages 8749–8753. IEEE, 2024. 2, 3, 4

    work page 2024

  7. [7]

    Clnerf: Continual learning meets nerf

    Zhipeng Cai and Matthias M ¨uller. Clnerf: Continual learning meets nerf. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 23185–23194, 2023. 6

    work page 2023

  8. [8]

    Asteroid-nerf: A deep-learning method for 3d surface recon- struction of asteroids

    Shihan Chen, Bo Wu, Hongliang Li, Zhaojin Li, and Yi Liu. Asteroid-nerf: A deep-learning method for 3d surface recon- struction of asteroids. Astronomy & Astrophysics, 687:A278,

    work page

Show all 65 references
  1. [9]

    Hallucinated neural radiance fields in the wild

    Xingyu Chen, Qi Zhang, Xiaoyu Li, Yue Chen, Ying Feng, Xuan Wang, and Jue Wang. Hallucinated neural radiance fields in the wild. In 2022 IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), pages 12933– 12942, 2022. 2

    2022

  2. [10]

    Scalar-nerf: scalable large- scale neural radiance fields for scene reconstruction

    Yu Chen and Gim Hee Lee. Scalar-nerf: scalable large- scale neural radiance fields for scene reconstruction. arXiv preprint arXiv:2311.16657, 2023. 4

    2023 arXiv

  3. [11]

    Large scale segmentation algorithm for object based image analysis suitable for hpc architectures in hybrid distributed-shared memory context

    R ´emi Cresson, Julien Michel, Arnaud Mallen, and Pierre Lassalle. Large scale segmentation algorithm for object based image analysis suitable for hpc architectures in hybrid distributed-shared memory context. In GEOBIA 2018-From pixels to ecosystems and global sustainability ...

    2018

  4. [12]

    Scaling up slic superpixels using a tile-based approach

    Dawa Derksen, Jordi Inglada, and Julien Michel. Scaling up slic superpixels using a tile-based approach. IEEE transac- tions on Geoscience and Remote Sensing , 57(5):3073–3085,

  5. [13]

    Shadow neural radiance fields for multi-view satellite photogrammetry

    Dawa Derksen and Dario Izzo. Shadow neural radiance fields for multi-view satellite photogrammetry. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 1152–1161, 2021. 2, 3, 4

    2021

  6. [14]

    Radar fields: An extension of radiance fields to sar

    Thibaud Ehret, Roger Mar ´ı, Dawa Derksen, Nicolas Gas- nier, and Gabriele Facciolo. Radar fields: An extension of radiance fields to sar. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 564–574, 2024. 2

    2024

  7. [15]

    Automatic 3D reconstruction from multi-date satel- lite images

    Gabriele Facciolo, Carlo de Franchis, and Enric Meinhardt- Llopis. Automatic 3D reconstruction from multi-date satel- lite images. In 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 1542– 1551, 2017. 4

    2017

  8. [16]

    Incorporating season and solar specificity into renderings made by a NeRF archi- tecture using satellite images

    Michael Gableman and Avinash Kak. Incorporating season and solar specificity into renderings made by a NeRF archi- tecture using satellite images. IEEE Transactions on Pattern Analysis and Machine Intelligence , 46(6):4348–4365, 2024. 2, 8

    2024

  9. [17]

    NeRF: Neural radiance field in 3D vision, a comprehensive review

    Kyle Gao, Yina Gao, Hongjie He, Denning Lu, Linlin Xu, and Jonathan Li. NeRF: Neural radiance field in 3D vision, a comprehensive review. arXiv e-prints, pages arXiv–2210,

  10. [18]

    Bayes’ rays: Uncertainty quan- tification for neural radiance fields

    Lily Goli, Cody Reading, Silvia Sell ´an, Alec Jacobson, and Andrea Tagliasacchi. Bayes’ rays: Uncertainty quan- tification for neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20061–20070, 2024. 8

    2024

  11. [19]

    Calisa-nerf: Neural radiance field with pinhole camera images lidar point clouds and satel- lite imagery for urban scene representation

    Juyeop Han, Guilherme Cavalheiro, Josef Biberstein, Elham Alkabawi, Shahad Alqhatni, Fadwa Alaskar, Eman Bin Khu- nayn, and Sertac Karaman. Calisa-nerf: Neural radiance field with pinhole camera images lidar point clouds and satel- lite imagery for urban scene representation. ...

    2025

  12. [20]

    Spacecraft state esti- mation using neural radiance fields

    Aneesh M Heintz and Mason Peck. Spacecraft state esti- mation using neural radiance fields. Journal of Guidance, Control, and Dynamics, 46(8):1596–1609, 2023. 2

    2023

  13. [21]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42(4):139(1– 14), 2023. 2, 8

    2023

  14. [22]

    Overcoming catastrophic forgetting in neu- ral networks

    James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska- Barwinska, et al. Overcoming catastrophic forgetting in neu- ral networks. Proceedings of the national academy of sc...

    2017

  15. [23]

    NeROIC: Neural rendering of objects from online image collections

    Zhengfei Kuang, Kyle Olszewski, Menglei Chai, Zeng Huang, Panos Achlioptas, and Sergey Tulyakov. NeROIC: Neural rendering of objects from online image collections. ACM Transactions on Graphics (TOG), 41(4):1–12, 2022. 2

    2022

  16. [24]

    Geochat: Grounded large vision-language model for remote sensing

    Kartik Kuckreja, Muhammad Sohail Danish, Muzammal Naseer, Abhijit Das, Salman Khan, and Fahad Shahbaz Khan. Geochat: Grounded large vision-language model for remote sensing. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) , pages 2...

    2024

  17. [25]

    A scalable tile-based framework for region-merging segmentation

    Pierre Lassalle, Jordi Inglada, Julien Michel, Manuel Gri- zonnet, and Julien Malik. A scalable tile-based framework for region-merging segmentation. IEEE Transactions on Geoscience and Remote Sensing , 53(10):5473–5485, 2015. 2

    2015

  18. [26]

    2019 data fusion contest [technical committees]

    Bertrand Le Saux, Naoto Yokoya, Ronny Hansch, Myron Brown, and Greg Hager. 2019 data fusion contest [technical committees]. IEEE Geoscience and Remote Sensing Maga- zine, 7(1):103–105, 2019. 2, 4, 6, 8

    2019

  19. [27]

    Urban semantic 3D reconstruc- tion from multiview satellite imagery

    Matthew J Leotta, Chengjiang Long, Bastien Jacquet, Matthieu Zins, Dan Lipsa, Jie Shan, Bo Xu, Zhixin Li, Xu Zhang, Shih-Fu Chang, et al. Urban semantic 3D reconstruc- tion from multiview satellite imagery. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition...

    2019

  20. [28]

    Nerf-xl: Scaling nerfs with multiple gpus

    Ruilong Li, Sanja Fidler, Angjoo Kanazawa, and Francis Williams. Nerf-xl: Scaling nerfs with multiple gpus. In European Conference on Computer Vision , pages 92–107. Springer, 2024. 2, 3, 4, 6

    2024

  21. [29]

    Nerfacc: Efficient sampling accelerates nerfs

    Ruilong Li, Hang Gao, Matthew Tancik, and Angjoo Kanazawa. Nerfacc: Efficient sampling accelerates nerfs. In Proceedings of the IEEE/CVF international conference on computer vision, pages 18537–18546, 2023. 6

    2023

  22. [30]

    Evaluation of the urban heat island effect based on 3D modeling and planning indicators for urban planning proposals

    Yanwen Luo and Jiang He and. Evaluation of the urban heat island effect based on 3D modeling and planning indicators for urban planning proposals. Journal of Asian Architecture and Building Engineering, 0(0):1–23, 2024. 2

    2024

  23. [31]

    Sat- NeRF: Learning multi-view satellite photogrammetry with transient objects and shadow modeling using RPC cameras

    Roger Mar ´ı, Gabriele Facciolo, and Thibaud Ehret. Sat- NeRF: Learning multi-view satellite photogrammetry with transient objects and shadow modeling using RPC cameras. In 2022 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition Workshops (CVPRW) , pages 1310–1320,

    2022

  24. [32]

    Multi- date earth observation NeRF: The detail is in the shadows

    Roger Mar ´ı, Gabriele Facciolo, and Thibaud Ehret. Multi- date earth observation NeRF: The detail is in the shadows. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2034–2044, 2023. 2, 8

    2023

  25. [33]

    NeRF in the wild: Neural radiance fields for uncon- strained photo collections

    Ricardo Martin-Brualla, Noha Radwan, Mehdi SM Sajjadi, Jonathan T Barron, Alexey Dosovitskiy, and Daniel Duck- worth. NeRF in the wild: Neural radiance fields for uncon- strained photo collections. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), ...

    2021

  26. [34]

    Vision-based neural scene representations for space- craft

    Anne Mergy, Gurvan Lecuyer, Dawa Derksen, and Dario Izzo. Vision-based neural scene representations for space- craft. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition , pages 2002–2011,

    2002

  27. [35]

    Switch-nerf: Learning scene de- composition with mixture of experts for large-scale neural radiance fields

    Zhenxing Mi and Dan Xu. Switch-nerf: Learning scene de- composition with mixture of experts for large-scale neural radiance fields. In The Eleventh International Conference on Learning Representations, 2022. 3

    2022

  28. [36]

    A new satellite imagery stereo pipeline designed for scal- ability, robustness and performance

    Julien Michel, Emmanuelle Sarrazin, David Youssefi, Myr- iam Cournet, Fabrice Buffe, Jean-Marc Delvit, Aur ´elie Em- ilien, Julien Bosman, Olivier Melet, and C ´eline L’Helguen. A new satellite imagery stereo pipeline designed for scal- ability, robustness and performance. ISP...

    2020

  29. [37]

    NeRF: Representing scenes as neural radiance fields for view syn- thesis

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. NeRF: Representing scenes as neural radiance fields for view syn- thesis. In Computer Vision – ECCV 2020 , pages 405–421,

    2020

  30. [38]

    Instant neural graphics primitives with a multires- olution hash encoding

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a multires- olution hash encoding. ACM Trans. Graph., 41(4), 2022. 3, 4, 6

    2022

  31. [39]

    Pseudo pansharpening nerf for satellite image collec- tions

    Emilie Pic, Thibaud Ehret, Gabriele Facciolo, and Roger Mar´ı. Pseudo pansharpening nerf for satellite image collec- tions. In IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium , pages 2650–2655. IEEE,

    2024

  32. [40]

    Instant continual learning of neural ra- diance fields

    Ryan Po, Zhengyang Dong, Alexander W Bergman, and Gordon Wetzstein. Instant continual learning of neural ra- diance fields. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 3334–3344, 2023. 6

    2023

  33. [41]

    Sat-mesh: Learning neural im- plicit surfaces for multi-view satellite reconstruction.Remote Sensing, 15(17):4297, 2023

    Yingjie Qu and Fei Deng. Sat-mesh: Learning neural im- plicit surfaces for multi-view satellite reconstruction.Remote Sensing, 15(17):4297, 2023. 2

    2023

  34. [42]

    Kilonerf: Speeding up neural radiance fields with thousands of tiny mlps

    Christian Reiser, Songyou Peng, Yiyi Liao, and Andreas Geiger. Kilonerf: Speeding up neural radiance fields with thousands of tiny mlps. In Proceedings of the IEEE/CVF international conference on computer vision , pages 14335– 14345, 2021. 3, 4

    2021

  35. [43]

    MicMac–a free, open-source solution for pho- togrammetry

    Ewelina Rupnik, Mehdi Daakir, and Marc Pierrot- Deseilligny. MicMac–a free, open-source solution for pho- togrammetry. Open Geospatial Data, Software and Stan- dards, 2(14), 2017. 4

    2017

  36. [44]

    c3-nerf: Modeling multiple scenes via conditional-cum-continual neural radiance fields

    Prajwal Singh, Ashish Tiwari, Gautam Vashishtha, and Shanmuganathan Raman. c3-nerf: Modeling multiple scenes via conditional-cum-continual neural radiance fields. arXiv preprint arXiv:2411.19903, 2024. 6

    2024 arXiv

  37. [45]

    Fu- sionrf: High-fidelity satellite neural radiance fields from multispectral and panchromatic acquisitions

    Michael Sprintson, Rama Chellappa, and Cheng Peng. Fu- sionrf: High-fidelity satellite neural radiance fields from multispectral and panchromatic acquisitions. arXiv preprint arXiv:2409.15132, 2024. 2

    2024

  38. [46]

    Block-nerf: Scalable large scene neural view synthesis

    Matthew Tancik, Vincent Casser, Xinchen Yan, Sabeek Prad- han, Ben Mildenhall, Pratul P Srinivasan, Jonathan T Barron, and Henrik Kretzschmar. Block-nerf: Scalable large scene neural view synthesis. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern rec...

    2022

  39. [47]

    Nerfstudio: A modular framework for neural radiance field development

    Matthew Tancik, Ethan Weber, Evonne Ng, Ruilong Li, Brent Yi, Terrance Wang, Alexander Kristoffersen, Jake Austin, Kamyar Salahi, Abhik Ahuja, et al. Nerfstudio: A modular framework for neural radiance field development. In ACM SIGGRAPH 2023 conference proceedings, pages 1– 12...

    2023

  40. [48]

    Silvr: Scalable lidar-visual reconstruction with neural radiance fields for robotic inspection

    Yifu Tao, Yash Bhalgat, Lanke Frank Tarimo Fu, Matias Mattamala, Nived Chebrolu, and Maurice Fallon. Silvr: Scalable lidar-visual reconstruction with neural radiance fields for robotic inspection. In IEEE International Confer- ence on Robotics and Automation (ICRA) , 2024. 3

    2024

  41. [49]

    Ar- tificial intelligence to advance earth observation: A review of models, recent trends, and pathways forward

    Devis Tuia, Konrad Schindler, Beg ¨um Demir, Xiao Xiang Zhu, Mrinalini Kochupillai, Sa ˇso Dˇzeroski, Jan N van Rijn, Holger H Hoos, Fabio Del Frate, Mihai Datcu, et al. Ar- tificial intelligence to advance earth observation: A review of models, recent trends, and pathways for...

    2024

  42. [50]

    Mega-NeRF: Scalable construction of large- scale NeRFs for virtual fly-throughs

    Haithem Turki, Deva Ramanan, and Mahadev Satya- narayanan. Mega-NeRF: Scalable construction of large- scale NeRFs for virtual fly-throughs. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12922–12931, 2022. 2, 3, 4

    2022

  43. [51]

    Suds: Scalable urban dynamic scenes

    Haithem Turki, Jason Y Zhang, Francesco Ferroni, and Deva Ramanan. Suds: Scalable urban dynamic scenes. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12375–12385, 2023. 3

    2023

  44. [52]

    Semantic neural radiance fields for multi-date satellite data

    Valentin Wagner, Sebastian Bullinger, Christoph Boden- steiner, and Michael Arens. Semantic neural radiance fields for multi-date satellite data. In Proceedings of the Win- ter Conference on Applications of Computer Vision , pages 1238–1246, 2025. 2

    2025

  45. [53]

    Scarf: Scalable continual learn- ing framework for memory-efficient multiple neural radi- ance fields

    Yuze Wang, Junyi Wang, Chen Wang, Wantong Duan, Yong- tang Bao, and Yue Qi. Scarf: Scalable continual learn- ing framework for memory-efficient multiple neural radi- ance fields. In Computer Graphics F orum, volume 43, page e15255. Wiley Online Library, 2024. 6

    2024

  46. [54]

    Scanerf: Scalable bundle-adjusting neural radiance fields for large-scale scene rendering

    Xiuchao Wu, Jiamin Xu, Xin Zhang, Hujun Bao, Qix- ing Huang, Yujun Shen, James Tompkin, and Weiwei Xu. Scanerf: Scalable bundle-adjusting neural radiance fields for large-scale scene rendering. ACM Trans. Graph., 42(6), Dec

  47. [55]

    Scalable neural in- door scene rendering

    Xiuchao Wu, Jiamin Xu, Zihan Zhu, Hujun Bao, Qixing Huang, James Tompkin, and Weiwei Xu. Scalable neural in- door scene rendering. ACM Trans. Graph., 41(4), July 2022. 3

    2022

  48. [56]

    Grid-guided neural radiance fields for large urban scenes

    Linning Xu, Yuanbo Xiangli, Sida Peng, Xingang Pan, Nanxuan Zhao, Christian Theobalt, Bo Dai, and Dahua Lin. Grid-guided neural radiance fields for large urban scenes. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 8296–8306, 2023...

    2023

  49. [57]

    Scalable and high-quality neural implicit representation for 3D recon- struction

    Leyuan Yang, Bailin Deng, and Juyong Zhang. Scalable and high-quality neural implicit representation for 3D recon- struction. IEEE Transactions on Visualization and Computer Graphics, pages 1–17, 2025. 3

    2025

  50. [58]

    Application research on dig- ital twins of urban earthquake disasters

    Sihan Yu, Qiyun Lei, Chao Liu, Nan Zhang, Shuaishuai Shan, and Xiaoming Zeng and. Application research on dig- ital twins of urban earthquake disasters. Geomatics, Natural Hazards and Risk, 14(1):2278274, 2023. 2

    2023

  51. [59]

    SuperNeRF: High-precision 3D reconstruction for large- scale scenes

    Guangyun Zhang, Chaozhong Xue, and Rongting Zhang. SuperNeRF: High-precision 3D reconstruction for large- scale scenes. IEEE Transactions on Geoscience and Remote Sensing, 2024. 2

    2024

  52. [60]

    NeRS: Neural reflectance surfaces for sparse-view 3D reconstruction in the wild

    Jason Zhang, Gengshan Yang, Shubham Tulsiani, and Deva Ramanan. NeRS: Neural reflectance surfaces for sparse-view 3D reconstruction in the wild. Advances in Neural Informa- tion Processing Systems, 34:29835–29847, 2021. 2

    2021

  53. [61]

    Il-nerf: Incremental learning for neural radiance fields with camera pose alignment

    Letian Zhang, Ming Li, Chen Chen, and Jie Xu. Il-nerf: Incremental learning for neural radiance fields with camera pose alignment. arXiv preprint arXiv:2312.05748, 2023. 6

    2023 arXiv

  54. [62]

    Sparsesat-NeRF: Dense depth supervised neural radiance fields for sparse satellite images

    Lulin Zhang and Ewelina Rupnik. Sparsesat-NeRF: Dense depth supervised neural radiance fields for sparse satellite images. In ISPRS Annals 2023, 2023. 2

    2023

  55. [63]

    Brdf-nerf: Neural radiance fields with optical satellite images and brdf modelling

    Lulin Zhang, Ewelina Rupnik, Tri Dung Nguyen, St ´ephane Jacquemoud, and Yann Klinger. Brdf-nerf: Neural radiance fields with optical satellite images and brdf modelling. arXiv preprint arXiv:2409.12014, 2024. 2

    2024 arXiv

  56. [64]

    Satensorf: Fast satellite tensorial radiance field for multidate satellite imagery of large size

    Tongtong Zhang, Yu Zhou, Yuanxiang Li, and Xian Wei. Satensorf: Fast satellite tensorial radiance field for multidate satellite imagery of large size. IEEE Transactions on Geo- science and Remote Sensing , 62:1–15, 2024. 2

    2024

  57. [65]

    An overview of the applications of Earth observation satellite data: impacts and future trends

    Qiang Zhao, Le Yu, Zhenrong Du, Dailiang Peng, Pengyu Hao, Yongguang Zhang, and Peng Gong. An overview of the applications of Earth observation satellite data: impacts and future trends. Remote Sensing, 14(8):1863, 2022. 1

    2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.