Pith. sign in

REVIEW 4 major objections 5 minor 49 references

Bias for Action: Video Implicit Neural Representations with Bias Modulation

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Video motion lives in the bias term, claims ActINR

desk verdict Solid, well-motivated video INR paper with a genuine bias-modulation idea, let down mainly by missing code/error bars and an intro that overclaims large-motion handling beyond its block-local design. read the letter →

arxiv 2501.09277 v2 pith:QMLP2YJZ submitted 2025-01-16 cs.CV

Alper Kayabasi , Anil Kumar Vadathya , Guha Balakrishnan , Vishwanath Saragadam This is my paper · ORCID
classification cs.CV
keywords implicitneuralrepresentationvideobiasmodulationtemporalinterpolationspace-timesuper-resolutiondenoisinginpaintingwaveletactivation
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

The paper proposes ActINR, a video implicit neural representation that keeps one set of network weights for all frames and encodes each frame's motion only in the network's bias vectors. It argues that an INR with compactly supported activations behaves like a dictionary of localized basis functions, where weights set the shapes and biases set the locations, so translating bases through biases is enough to represent moving scene content. A second INR, conditioned on continuous time, outputs the biases, making the representation smooth in time and queryable at arbitrary timestamps. The authors claim this design outperforms prior video INRs on temporal interpolation, space-time super-resolution, denoising, and inpainting, with reported gains of several dB.

What carries the argument

The load-bearing mechanism is bias modulation in a shared-weight INR. Each patch-wise frame INR computes $y^{(l)}_i = \sigma(W^{(l)} y^{(l-1)}_i + b^{(l)}_i)$, with weights $W^{(l)}$ shared across all frames and patches, while a bias INR $\psi$ maps $(\gamma(t), z_i)$ to the per-layer bias vectors $b^{(l)}_i$, where $\gamma(t)$ is a random Fourier feature encoding of time and $z_i$ is a patch-specific latent code. The activation $\sigma$ is WIRE, a wavelet activation with compact spatial support, which is what makes a basis-location interpretation approximately valid; the paper's ablation shows that non-compact activations like SIREN spread motion artifacts into static regions. The representation is trained block-wise, with each 96x96 spatial patch over a short frame group handled by its own small INR, following a divide-and-conquer strategy that keeps the motion local and the parameter count manageable.

What would settle it

Train ActINR on a synthetic video of a bright dot translating at constant speed across a 96x96 patch boundary, holding out intermediate frames for interpolation; if the interpolated dot trajectory breaks or ghosts at the seam, or if PSNR drops sharply as the crossing fraction increases, the block-confinement and bias-location assumptions would be violated.

Watch

Extended reading notes

Core claim

ActINR's central claim is that the motion in a video can be carried entirely by bias modulation: with a compactly supported activation such as WIRE, an INR is a learned dictionary whose basis shapes are fixed by the weights and whose positions are set by the biases, so shifting a basis function is equivalent to shifting a moving object's location. The paper's design shares all frame-INR weights across frames and assigns each frame a unique bias vector produced by a separate bias-INR hypernetwork that takes a continuous time index, random Fourier features, and a patch latent code as input. Training the frame INR and the bias INR together with a pixel-wise MSE loss yields a continuous video representation that supports arbitrary temporal queries and spatial queries. The authors report that this representation outperforms existing video INRs on 2x and 10x temporal interpolation, on combined spatial and temporal super-resolution (up to 8x spatial with 2x temporal), on extreme denoising, and on video inpainting, with improvements often exceeding 6 dB.

Load-bearing premise

The paper's approach assumes that a video's motion can be represented by translating compactly supported basis functions through bias changes alone, with the network weights fixed; the authors state the practical version of this assumption as motion remaining confined within a designated block size, so objects that cross patch boundaries degrade the representation.

Editorial extensions

If this is right

  • Temporal interpolation becomes a query of the bias INR at unseen time indices, producing 10x slow-motion video with reported PSNR gains of up to 5 dB over baselines.
  • Arbitrary spatial queries remain available because the frame INR maps continuous coordinates, so the same model performs 4x spatial upsampling combined with 2x temporal upsampling, with reported gains up to 8.5 dB.
  • The smoothness enforced by the bias INR acts as an implicit temporal regularizer, which the paper shows yields strong denoising (3-5 dB over a baseline) and improved inpainting (about 1.6 dB average) without explicit noise or motion models.
  • The representation remains competitive on standard video regression and is faster at decoding than some image-wise baselines, while the authors note it trails specialized methods on compression.
  • Because biases are the only per-frame parameters, encoding cost is dominated by the shared frame INR, and the authors report per-video encoding and decoding times comparable to prior video INRs.

Reading between the lines

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

  • Beyond the paper, the per-frame bias sequence itself could serve as a compact motion descriptor, potentially supporting motion editing, motion transfer, or a lightweight optical-flow substitute by tracking basis centers over time.
  • The block-confinement failure suggests a testable extension: hierarchical or adaptive patch assignment, or allowing bases to be re-assigned across blocks, could extend the method to large global motions without the reported patch-boundary artifacts.
  • The bias-INR-as-regulator idea is a general trick: any INR-based video or dynamic-scene model could impose temporal smoothness by modulating biases through a time-conditioned hypernetwork, rather than by introducing explicit flow or residual weights.
  • One could test the basis-location hypothesis directly by extracting the bias time series from a video with known rigid motion, e.g., a translating dot, and checking whether the inferred basis centers track the true trajectory linearly.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 paper proposes ActINR, a video implicit neural representation in which the weights of a frame INR are shared across frames while per-frame and per-layer bias vectors are predicted by a separate bias-INR conditioned on a continuous time index. Videos are processed in independent 96x96x10 blocks with WIRE activations, and the bias-INR is shared across patches through a per-patch latent code. The manuscript reports quantitative gains over prior video INRs on 2x interpolation, inpainting, denoising, and space-time super-resolution on UVG and DAVIS, plus a qualitative 10x slow-motion result and ablations on activation choice and on the bias-INR versus linear interpolation of biases. The paper also states an explicit limitation that motion must remain within the block size and includes a supplementary failure case showing artifacts when objects cross patch boundaries.

Significance. ActINR is a conceptually clean and potentially impactful reformulation of video INRs: instead of injecting time through an extra coordinate, residual weights, or optical flow, it represents motion as bias modulation of compact basis functions and predicts the biases with a small hypernetwork. The held-out evaluation is legitimate: frames used for interpolation are unseen during training, and the oracle/linear-interpolation ablation in Table 4 directly supports the choice of a learned continuous bias predictor. If the reported gains survive repeated runs and code release, ActINR would be a strong baseline for continuous video representation and for inverse problems such as denoising and inpainting. The main caveats are the acknowledged block-locality assumption, the absence of error bars and code, and the qualitative-only treatment of the advertised 10x slow-motion and part of the space-time super-resolution experiments.

major comments (4)
  1. [Section 5 (Limitations), Section 3, Eq. (2), and supplementary Fig. 13] The limitation statement in Section 5 ('Our approach assumes that motion remains confined within a designated block size') directly contradicts the Introduction's claim that ActINR 'can model small and large motions alike.' Because the method fits independent 96x96x10 blocks and Eq. (2)'s per-patch biases cannot transfer an object's basis functions across patch boundaries, the central interpolation and space-time super-resolution claims are conditional on all motion between training frames staying within roughly 96 pixels. The paper's own supplementary failure case (Fig. 13) demonstrates artifacts when an object crosses a patch boundary, and the overlapping-window remedy (Fig. 14) is not used in any main quantitative comparison. Please add a quantitative characterization of per-sequence motion magnitudes relative to the patch and GOP sizes used in Tables 1 and 3, and either restrict the large-motion claims accordingly or incorporate the overlapping-window strategy into the main evaluation.
  2. [Tables 1-3 and 5] All quantitative comparisons report a single run with no variance estimates, and no code is released. This is load-bearing for the SOTA claims, particularly in Table 1 on UVG: ActINR's average PSNR gain over H-NeRV Boost is only 0.4 dB, and ActINR is worse on Honey (37.6 vs 38.8) and Ready (25.9 vs 26.6). With a single run, these differences may be within run-to-run noise. Please release code and report mean and standard deviation over at least three seeds for the main comparisons, or otherwise provide per-sequence repeated-run numbers.
  3. [Section 4.1, Figure 5, and Figure 8] The headline capabilities of '10x video slow motion' and space-time super-resolution lack quantitative tables. Figure 5 is qualitative only, and Figure 8 reports PSNR numbers in the caption for a single scene (Bosphorus) without stating how the 10x or 8x targets are generated, how baselines are trained for those factors, or how the GOP is defined when every tenth frame is skipped. Since the abstract and introduction advertise these capabilities, they need formal evaluation with tables, including the exact temporal sampling protocol and baseline configurations.
  4. [Section 4.1 (Denoising) and Implementation Details] The denoising setup is internally inconsistent: the text says input PSNR is between 16 and 18 dB, the Implementation Details say the noise settings produce approximately 20 dB, and the Figure 6 captions state 18.6 dB and 16.6 dB in different places. Moreover, Table 3 compares only a single baseline. Please correct the noise-level reporting, state the exact photon-count and readout-noise parameters used for each sequence, and justify the single-baseline comparison for a task that is presented as one of ActINR's main contributions.
minor comments (5)
  1. [Section 3, Eq. (3)] The first line of Eq. (3) reads y_t^(0) = [x, y]^T, but the bias-INR is supposed to be conditioned on time, not spatial coordinates; this appears to be a typo and should read y_t^(0) = t (or the RFF of t).
  2. [Section 3, Eq. (3)] The latent code z_l in Eq. (3) is written with the subscript l, which clashes with the layer index l used in Eqs. (1) and (2); please rename it, for example z_i for patch i, to avoid confusion.
  3. [Throughout] The baseline name is inconsistent across the paper: H-NeRV, HNeRV, H-NeRV Boost, and HBoostNeRV all refer to the same method; please unify the notation.
  4. [References] References [18] and [19] are identical (both are 'Motion-adjustable neural implicit video representation'); the Phase-INR citation needs a distinct entry.
  5. [Dataset names] The dataset is referred to as 'DA VIS' throughout; the standard name is DAVIS, and the spacing should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: held-out evaluation and independent component usage keep the derivation self-contained.

full rationale

ActINR's claimed derivation chain is not circular. The core design—shared weights across frames with per-frame biases produced by a bias-INR—is an architectural hypothesis, not a tautology: the video INR and bias-INR are optimized only on training frames, and interpolation, super-resolution, denoising, and inpainting metrics are computed on held-out frames (e.g., Table 1 skips every other frame; Section 4.1). The 'biases control location' statement is a direct reading of Eq. (2) for fixed weights with compact activations, but the paper does not define 'motion' as 'change in bias,' so the empirical claim that bias modulation represents motion is independently tested by held-out reconstruction and by the interpolation and denoising experiments. The cited dictionary-perspective work ([35]) and WIRE activation ([37]) overlap with the authors, but they are used as a conceptual framing and as a component, respectively; neither is used to forbid alternatives or to justify the central prediction by fiat. Hyperparameters such as WIRE scale and patch size are task-specific choices, which is standard model selection, not fitted-input-called-prediction. The stated block-locality limitation (Section 5) is an acknowledged scope condition and a correctness concern, not a circular step. Therefore no circular step meets the evidence bar, and the appropriate verdict is no significant circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 2 invented entities

The central claim rests on the empirical assumption that bias modulation alone can capture video motion; this is what the paper calls the bias-motion interplay. The free parameters are task-level hyperparameters that are hand-tuned, while the architectural components (bias-INR and per-patch latents) are introduced by the paper. No new physical entities are postulated.

free parameters (4)
  • WIRE scale and frequency = scale=10, frequency=100 for interpolation; scale=5, frequency=50 for inpainting
    Chosen per task by hand; controls compactness and oscillation of basis functions, which is central to the bias-motion hypothesis.
  • RFF frequency variance (bias-INR) = 5
    Variance of the zero-mean Gaussian from which time-encoding frequencies are sampled; set by hand and not swept.
  • Patch size = 96x96 pixels
    Balances the local-motion assumption against parameter cost; the supplementary ablation shows a moderate patch size is optimal.
  • GOP size (frames per block) = 10
    Number of frames per spatial block; performance degrades as GOP grows, consistent with the local-motion assumption.
assumptions (4)
  • domain assumption Video motion can be modeled as translation of compactly supported basis functions; shared weights suffice if biases move the bases.
    Core hypothesis of the paper, introduced in Section 3 'Bias-motion Interplay'; validated only on a toy example and indirectly by experiments.
  • ad hoc to paper WIRE (wavelet) activation provides compact basis functions whose biases control location.
    Chosen over SIREN and Gauss based on ablation; the paper argues WIRE's oscillatory compact support is best (Section 4.2).
  • domain assumption The bias-INR's smooth temporal output provides a good prior for motion continuity.
    Used to justify interpolation, denoising, and inpainting; the denoising result is attributed to this implicit regularization.
  • standard math Random Fourier Features with fixed sampled frequencies approximate a useful temporal kernel.
    Standard RFF construction relying on Bochner's theorem; not proven in the paper.
invented entities (2)
  • Bias-INR hypernetwork
    purpose: Generates per-layer bias vectors for the frame INR as a function of continuous time and a per-patch latent code; enables temporal interpolation and smoothness.
    An architectural component; its necessity is supported only by the paper's internal ablation (Tab. 4), and it makes no falsifiable prediction outside the paper.
  • Per-patch latent code z
    purpose: Conditions the shared bias-INR on each spatial patch to capture local static content, avoiding separate bias-INRs per patch.
    A learned per-patch parameter; no external evidence beyond the paper's experiments.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bias for Action: Video Implicit Neural Representations with Bias Modulation." pith.science (2026). https://pith.science/paper/QMLP2YJZ

@misc{pith2026250109277,
  author       = {Pith},
  title        = {Pith review of: Bias for Action: Video Implicit Neural Representations with Bias Modulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QMLP2YJZ}},
  note         = {Machine review of arXiv:2501.09277}
}
abstract

We propose a new continuous video modeling framework based on implicit neural representations (INRs) called ActINR. At the core of our approach is the observation that INRs can be considered as a learnable dictionary, with the shapes of the basis functions governed by the weights of the INR, and their locations governed by the biases. Given compact non-linear activation functions, we hypothesize that an INR's biases are suitable to capture motion across images, and facilitate compact representations for video sequences. Using these observations, we design ActINR to share INR weights across frames of a video sequence, while using unique biases for each frame. We further model the biases as the output of a separate INR conditioned on time index to promote smoothness. By training the video INR and this bias INR together, we demonstrate unique capabilities, including $10\times$ video slow motion, 4x spatial super resolution along with 2x slow motion, denoising, and video inpainting. ActINR performs remarkably well across numerous video processing tasks (often achieving more than 6dB improvement), setting a new standard for continuous modeling of videos.

Figures

Figures reproduced from arXiv: 2501.09277 by the authors.

Figure 1
Figure 1. ActINR: Continuous Video Representation via Bias Modulation. We propose a new continuous representation of videos called ActINR. At its core is a frame INR that takes con￾tinuous spatial coordinates, and a bias INR that takes continuous time coordinates, and outputs bias values for the frame INR at that instant. This bias modulated framework enables us to repre￾sent videos in a concise manner, enabling several video… view at source ↗
Figure 2
Figure 2. Schematic overview of ActINR framework. ActINR is a novel video implicit representation model that consists of a frame INR with shared weights across all frames, that takes continuous coordinates as input, and outputs the intensity at that location, and a bias-INR, that takes a continuous time index as input, and outputs the biases for the frame INR at that time index. ActINR is therefore able to model the video con… view at source ↗
Figure 3
Figure 3. Toy example motivating the role of biases. Our toy example illustrates a video featuring two Gaussian blobs: the left blob moves toward the stationary one on the right. Column 1 dis￾plays snapshots of the ground truth at two different times, while Columns 2 and 3 show two distinct basis functions in ActINR cor￾responding to those times. These basis functions evolve over time in response to motion in scene, with thei… view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Qualitative results for 2× interpolation. In the first row, all methods except ours fail to capture fine-grained details in the video, such as the tree leaves, whereas our approach clearly delineates these intricate elements. In the second row, all other methods strugg…
Figure 5
Figure 5. Figure 5: Qualitative Results for 10× interpolation. In the first row, our method preserves the integrity of the lava fumes, whereas other methods blend them into the background. In the second row, our approach renders the child’s shoes nearly identical to the ground truth, whil…
Figure 6
Figure 6. Figure 6: Video Denoising. Our model inherently provides implicit regularization, unlike H-NeRV, which directly overfits to video. By constructing its function space with translations of a common building block, it accurately models dynamic evolution. Projecting video into this …
Figure 7
Figure 7. Figure 7: Video inpainting. In the first row, HNeRV Boost ex￾tends a cow’s spot beyond its actual length, and in the second row, it confuses a dog’s forehead with its ear. We attribute these error modes to the locality bias of convolution. ActINR, free from lo￾cality bias, recon…
Figure 9
Figure 9. Figure 9: Effect of activation function. The figure shows a toy scene with a circle moving to the right. The result above shows an interpolated frame. While SIREN results in strong spatial artifacts, even in the static background, Gauss and WIRE, with spatially compact activatio…
Figure 11
Figure 11. Figure 11: Effect of varying patch sizes on performance. The parameter size is fixed at 1.5 million, and the number of frames in GOP is kept constant at 20. The results demonstrate that a moderate-sized window yields optimal performance. Performance with varying number of parame…
Figure 10
Figure 10. Figure 10: Effect of varying GOP size on performance. The plot reveals a downward trend in reconstruction performance as the number of frames in GOP increases. This finding is consistent with our argument that Act-INR is designed to model local motion. Performance with varying p…
Figure 14
Figure 14. Figure 14: Results of training with window blending for the same Video in [PITH_FULL_IMAGE:figures/full_fig_p010_14.png]
Figure 13
Figure 13. Figure 13: The above illustration highlights a failure case of Act [PITH_FULL_IMAGE:figures/full_fig_p010_13.png]
Figure 15
Figure 15. Figure 15: Video of a vibrating tuning fork. Upon applying a [PITH_FULL_IMAGE:figures/full_fig_p011_15.png]
Figure 16
Figure 16. Figure 16: Rate-Distortion curve for Bosphorus video of a vibrating tuning fork, and fit our ActINR to it. We then median filtered the time series of biases across time to smoothen the motion artifacts, as shown in [PITH_FULL_IMAGE:figures/full_fig_p011_16.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 38 canonical work pages

  1. [1]

    Mip-nerf: A multiscale representation for anti-aliasing neu- ral radiance fields

    Jonathan T Barron, Ben Mildenhall, Matthew Tancik, Peter Hedman, Ricardo Martin-Brualla, and Pratul P Srinivasan. Mip-nerf: A multiscale representation for anti-aliasing neu- ral radiance fields. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 5855–5864,

  2. [2]

    Zip-nerf: Anti-aliased grid-based neural radiance fields

    Jonathan T Barron, Ben Mildenhall, Dor Verbin, Pratul P Srinivasan, and Peter Hedman. Zip-nerf: Anti-aliased grid-based neural radiance fields. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 19697–19705, 2023. 2

    2023

  3. [3]

    Nerv: Neural representations for videos

    Hao Chen, Bo He, Hanyu Wang, Yixuan Ren, Ser Nam Lim, and Abhinav Shrivastava. Nerv: Neural representations for videos. Advances in Neural Information Processing Systems, 34:21557–21568, 2021. 1, 3

    work page 2021

  4. [4]

    Hnerv: A hybrid neural representation for videos

    Hao Chen, Matthew Gwilliam, Ser-Nam Lim, and Abhi- nav Shrivastava. Hnerv: A hybrid neural representation for videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 10270– 10279, 2023. 1, 3, 11

    work page 2023

  5. [5]

    Towards scalable neural representation for di- verse videos

    Bo He, Xitong Yang, Hanyu Wang, Zuxuan Wu, Hao Chen, Shuaiyi Huang, Yixuan Ren, Ser-Nam Lim, and Abhinav Shrivastava. Towards scalable neural representation for di- verse videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 6132– 6142, 2023. 1

    work page 2023

  6. [6]

    Learn- ing a neural 3d texture space from 2d exemplars

    Philipp Henzler, Niloy J Mitra, and Tobias Ritschel. Learn- ing a neural 3d texture space from 2d exemplars. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8356–8364, 2020. 1

    work page 2020

  7. [7]

    Physics-informed machine learning

    George Em Karniadakis, Ioannis G Kevrekidis, Lu Lu, Paris Perdikaris, Sifan Wang, and Liu Yang. Physics-informed machine learning. Nature Reviews Physics , 3(6):422–440,

    work page

  8. [8]

    Incode: Implicit neural condition- ing with prior knowledge embeddings

    Amirhossein Kazerouni, Reza Azad, Alireza Hosseini, Dorit Merhof, and Ulas Bagci. Incode: Implicit neural condition- ing with prior knowledge embeddings. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 1298–1307, 2024. 2

    work page 2024

Show all 49 references
  1. [9]

    C3: High-performance and low-complexity neural compression from a single image or video

    Hyunjik Kim, Matthias Bauer, Lucas Theis, Jonathan Richard Schwarz, and Emilien Dupont. C3: High-performance and low-complexity neural compression from a single image or video. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9347–935...

    2024

  2. [10]

    Learning continuous representation of audio for arbi- trary scale super resolution

    Jaechang Kim, Yunjoo Lee, Seunghoon Hong, and Jungseul Ok. Learning continuous representation of audio for arbi- trary scale super resolution. In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 3703–3707. IEEE, 2022. 2 11

    2022

  3. [11]

    Controllable style transfer via test-time training of implicit neural representation

    Sunwoo Kim, Youngjo Min, Younghun Jung, and Seungry- ong Kim. Controllable style transfer via test-time training of implicit neural representation. Pattern Recognition, 146: 109988, 2024. 2

    2024

  4. [12]

    Hinerv: Video compression with hierarchical encoding-based neural representation

    Ho Man Kwan, Ge Gao, Fan Zhang, Andrew Gower, and David Bull. Hinerv: Video compression with hierarchical encoding-based neural representation. Advances in Neural Information Processing Systems, 36, 2024. 1

    2024

  5. [13]

    Siamese siren: Audio compression with implicit neural representa- tions

    Luca A Lanzend ¨orfer and Roger Wattenhofer. Siamese siren: Audio compression with implicit neural representa- tions. arXiv preprint arXiv:2306.12957, 2023. 2

    2023 arXiv

  6. [14]

    Ffnerv: Flow-guided frame-wise neural representa- tions for videos

    Joo Chan Lee, Daniel Rho, Jong Hwan Ko, and Eunbyung Park. Ffnerv: Flow-guided frame-wise neural representa- tions for videos. In Proceedings of the 31st ACM Interna- tional Conference on Multimedia , pages 7859–7870, 2023. 1, 2, 3

    2023

  7. [15]

    Superpixel-informed implicit neural representation for multi-dimensional data

    Jiayi Li, Xile Zhao, Jianli Wang, Chao Wang, and Min Wang. Superpixel-informed implicit neural representation for multi-dimensional data. In European Conference on Computer Vision, pages 258–276. Springer, 2025. 3

    2025

  8. [16]

    Learning deep implicit functions for 3d shapes with dynamic code clouds

    Tianyang Li, Xin Wen, Yu-Shen Liu, Hua Su, and Zhizhong Han. Learning deep implicit functions for 3d shapes with dynamic code clouds. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 12840–12850, 2022. 2

    2022

  9. [17]

    E-nerv: Expedite neural video rep- resentation with disentangled spatial-temporal context

    Zizhang Li, Mengmeng Wang, Huaijin Pi, Kechun Xu, Jian- biao Mei, and Yong Liu. E-nerv: Expedite neural video rep- resentation with disentangled spatial-temporal context. In European Conference on Computer Vision , pages 267–284. Springer, 2022. 1

    2022

  10. [18]

    Motion-adjustable neural implicit video representation

    Long Mai and Feng Liu. Motion-adjustable neural implicit video representation. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 10738–10747, 2022. 2

    2022

  11. [19]

    Motion-adjustable neural implicit video representation

    Long Mai and Feng Liu. Motion-adjustable neural implicit video representation. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 10738–10747, 2022. 3, 6

    2022

  12. [20]

    Nirvana: Neural implicit representations of videos with adaptive networks and au- toregressive patch-wise modeling

    Shishira R Maiya, Sharath Girish, Max Ehrlich, Hanyu Wang, Kwot Sin Lee, Patrick Poirson, Pengxiang Wu, Chen Wang, and Abhinav Shrivastava. Nirvana: Neural implicit representations of videos with adaptive networks and au- toregressive patch-wise modeling. In Proceedings of the...

    2023

  13. [21]

    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 Proceedings of the IEEE/CVF conference on computer vision and pattern recogn...

    2021

  14. [22]

    Mod- ulated periodic activations for generalizable local functional representations

    Ishit Mehta, Micha ¨el Gharbi, Connelly Barnes, Eli Shecht- man, Ravi Ramamoorthi, and Manmohan Chandraker. Mod- ulated periodic activations for generalizable local functional representations. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision, pages ...

  15. [23]

    Uvg dataset: 50/120fps 4k sequences for video codec analysis and development

    Alexandre Mercat, Marko Viitanen, and Jarno Vanne. Uvg dataset: 50/120fps 4k sequences for video codec analysis and development. In Proceedings of the 11th ACM Multimedia Systems Conference, pages 297–302, 2020. 4

    2020

  16. [25]

    Resfields: Residual neural fields for spatiotem- poral signals

    Marko Mihajlovic, Sergey Prokudin, Marc Pollefeys, and Siyu Tang. Resfields: Residual neural fields for spatiotem- poral signals. arXiv preprint arXiv:2309.03160, 2023. 2

    2023 arXiv

  17. [26]

    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. Communications of the ACM, 65(1):99–106, 2021. 1, 2

    2021

  18. [27]

    Texture fields: Learning tex- ture representations in function space

    Michael Oechsle, Lars Mescheder, Michael Niemeyer, Thilo Strauss, and Andreas Geiger. Texture fields: Learning tex- ture representations in function space. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 4531–4540, 2019. 1

    2019

  19. [28]

    Deepsdf: Learning con- tinuous signed distance functions for shape representation

    Jeong Joon Park, Peter Florence, Julian Straub, Richard Newcombe, and Steven Lovegrove. Deepsdf: Learning con- tinuous signed distance functions for shape representation. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 165–174, 2019. 2

    2019

  20. [29]

    A benchmark dataset and evaluation methodology for video object segmentation

    Federico Perazzi, Jordi Pont-Tuset, Brian McWilliams, Luc Van Gool, Markus Gross, and Alexander Sorkine-Hornung. A benchmark dataset and evaluation methodology for video object segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition , pag...

  21. [30]

    On the spectral bias of neural networks

    Nasim Rahaman, Aristide Baratin, Devansh Arpit, Felix Draxler, Min Lin, Fred Hamprecht, Yoshua Bengio, and Aaron Courville. On the spectral bias of neural networks. In International conference on machine learning , pages 5301–

  22. [31]

    Beyond periodicity: Towards a unifying framework for activations in coordinate- mlps

    Sameera Ramasinghe and Simon Lucey. Beyond periodicity: Towards a unifying framework for activations in coordinate- mlps. In European Conference on Computer Vision , pages 142–158. Springer, 2022. 2

    2022

  23. [32]

    Derf: Decom- posed radiance fields

    Daniel Rebain, Wei Jiang, Soroosh Yazdani, Ke Li, Kwang Moo Yi, and Andrea Tagliasacchi. Derf: Decom- posed radiance fields. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 14153–14161, 2021. 3

    2021

  24. [33]

    Dynamic ct reconstruction from limited views with implicit neural representations and parametric motion fields

    Albert W Reed, Hyojin Kim, Rushil Anirudh, K Aditya Mo- han, Kyle Champley, Jingu Kang, and Suren Jayasuriya. Dynamic ct reconstruction from limited views with implicit neural representations and parametric motion fields. In Proceedings of the IEEE/CVF International Conference...

    2021

  25. [34]

    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. 2, 3

    2021

  26. [35]

    Implicit neural represen- 12 tations and the algebra of complex wavelets

    T Mitchell Roddenberry, Vishwanath Saragadam, Maarten V de Hoop, and Richard G Baraniuk. Implicit neural represen- 12 tations and the algebra of complex wavelets. arXiv preprint arXiv:2310.00545, 2023. 2, 3

    2023 arXiv

  27. [36]

    Combining frame and gop embeddings for neural video representation

    Jens Eirik Saethre, Roberto Azevedo, and Christopher Schroers. Combining frame and gop embeddings for neural video representation. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pages 9253–9263, 2024. 1

    2024

  28. [37]

    Wire: Wavelet implicit neural representations

    Vishwanath Saragadam, Daniel LeJeune, Jasper Tan, Guha Balakrishnan, Ashok Veeraraghavan, and Richard G Bara- niuk. Wire: Wavelet implicit neural representations. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18507–18516, 2023. 1, 2, 4

    2023

  29. [38]

    Implicit neural representa- tions with periodic activation functions

    Vincent Sitzmann, Julien Martel, Alexander Bergman, David Lindell, and Gordon Wetzstein. Implicit neural representa- tions with periodic activation functions. Advances in neural information processing systems, 33:7462–7473, 2020. 2

    2020

  30. [39]

    Nerv: Neural reflectance and visibility fields for relighting and view synthesis

    Pratul P Srinivasan, Boyang Deng, Xiuming Zhang, Matthew Tancik, Ben Mildenhall, and Jonathan T Barron. Nerv: Neural reflectance and visibility fields for relighting and view synthesis. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition , pa...

    2021

  31. [40]

    Implicit neural representations for image compression

    Yannick Str ¨umpler, Janis Postels, Ren Yang, Luc Van Gool, and Federico Tombari. Implicit neural representations for image compression. In European Conference on Computer Vision, pages 74–91. Springer, 2022. 2

    2022

  32. [41]

    Overview of the high efficiency video coding (hevc) standard

    Gary J Sullivan, Jens-Rainer Ohm, Woo-Jin Han, and Thomas Wiegand. Overview of the high efficiency video coding (hevc) standard. IEEE Transactions on circuits and systems for video technology, 22(12):1649–1668, 2012. 3

    2012

  33. [42]

    Coil: Coordinate-based inter- nal learning for imaging inverse problems

    Yu Sun, Jiaming Liu, Mingyang Xie, Brendt Wohlberg, and Ulugbek S Kamilov. Coil: Coordinate-based inter- nal learning for imaging inverse problems. arXiv preprint arXiv:2102.05181, 2021. 1

    2021 arXiv

  34. [43]

    Mul- tiscale structural similarity for image quality assessment

    Zhou Wang, Eero P Simoncelli, and Alan C Bovik. Mul- tiscale structural similarity for image quality assessment. In The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003, pages 1398–1402. Ieee, 2003. 4

    2003

  35. [44]

    Overview of the h

    Thomas Wiegand, Gary J Sullivan, Gisle Bjontegaard, and Ajay Luthra. Overview of the h. 264/avc video coding stan- dard. IEEE Transactions on circuits and systems for video technology, 13(7):560–576, 2003. 3

    2003

  36. [45]

    Implicit neural representations for deformable image registration

    Jelmer M Wolterink, Jesse C Zwienenberg, and Christoph Brune. Implicit neural representations for deformable image registration. In International Conference on Medical Imag- ing with Deep Learning , pages 1349–1359. PMLR, 2022. 2

    2022

  37. [46]

    Revisiting implicit neu- ral representations in low-level vision

    Wentian Xu and Jianbo Jiao. Revisiting implicit neu- ral representations in low-level vision. arXiv preprint arXiv:2304.10250, 2023. 2

    2023 arXiv

  38. [47]

    Ds-nerv: Implicit neural video representation with decomposed static and dynamic codes

    Hao Yan, Zhihui Ke, Xiaobo Zhou, Tie Qiu, Xidong Shi, and Dadong Jiang. Ds-nerv: Implicit neural video representation with decomposed static and dynamic codes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 23019–23029, 2024. 1, 4, 6

    2024

  39. [48]

    A structured dictionary perspective on implicit neural representations

    Gizem Y ¨uce, Guillermo Ortiz-Jim ´enez, Beril Besbinar, and Pascal Frossard. A structured dictionary perspective on implicit neural representations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 19228–19238, 2022. 2, 3

    2022

  40. [49]

    Boosting neu- ral representations for videos with a conditional decoder

    Xinjie Zhang, Ren Yang, Dailan He, Xingtong Ge, Tongda Xu, Yan Wang, Hongwei Qin, and Jun Zhang. Boosting neu- ral representations for videos with a conditional decoder. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 2556–2566, 20...

    2024

  41. [50]

    Dnerv: Model- ing inherent dynamics via difference neural representation for videos

    Qi Zhao, M Salman Asif, and Zhan Ma. Dnerv: Model- ing inherent dynamics via difference neural representation for videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 2031– 2040, 2023. 1, 2, 3 13

    2023

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.