Pith. sign in

REVIEW 1 major objections 4 minor 51 references

Learning Switchable Priors for Neural Image Compression

T0 review · 1 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a finite set of trainable priors, addressed by a single index predicted by the entropy model, decouples entropy-coding complexity from the probabilistic model and, with a learned skip mode, yields a lightweight codec…

desk verdict A genuinely useful NIC paper with a real new mechanism, but the skip-mode training objective in Eq. (9) may not match what the decoder actually reconstructs, and the authors need to answer that before the headline skip-mode gains can be trusted. read the letter →

arxiv 2504.16586 v1 pith:2F4PHJUK submitted 2025-04-23 cs.MM

Haotian Zhang , Yuqi Li , Li Li , Dong Liu This is my paper · ORCID
classification cs.MM
keywords switchablepriorsneuralimagecompressionentropycodinggeneralizedGaussianmodelskipmodeFastNICrate-distortionoptimizationCDFtables
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

Neural image compression faces a dilemma: probability models with more parameters fit latents better but make entropy coding slower and more memory-hungry. This paper claims the dilemma can be dissolved by learning a finite set of trainable priors and having the entropy model predict only an index into that set, with the priors converted once into shared CDF tables. Because coding then uses pre-stored tables and a simple rounding operator, its complexity is decoupled from the form of the probability model, so richer models such as the generalized Gaussian become nearly free at test time. The paper also adds a skip mode that omits near-zero-rate latents and builds a lightweight codec, FastNIC, reporting 4.10% BD-rate savings over BPG on Kodak with encoding below 12 KMACs/pixel and decoding below 10 KMACs/pixel, and roughly 30-50% faster entropy coding when the method is inserted into strong baselines. A sympathetic reader would care because the method attacks the practical bottleneck that keeps advanced probability models out of deployed neural codecs.

What carries the argument

The load-bearing object is the trainable prior set: a finite collection of parameterized distributions \(\{p(\cdot|\theta_m)\}_{m=1}^M\) whose parameters are jointly optimized with the codec and then frozen into shared CDF tables. The entropy model's only job is to predict a continuous index \(i\), which is turned into a hard prior index by clipping and rounding; during training, rate is a softmax-weighted average over neighboring priors, and a Top-2 variant of this weighting accelerates training. The same prior set is reused for the hyperlatents, removing per-channel CDF tables. For skip mode, an auxiliary binary mask is produced from the same entropy module using a continuous relaxation of discrete sampling, so latents with negligible rate are omitted from entropy coding and the reconstruction is built from \(\hat{y}\cdot\tilde{b}\). A two-dimensional variant of the prior set lets models with separate mean and scale parameters, such as a Gaussian mixture, use the same decoupling.

What would settle it

Instrument or reimplement the codec at one quality point on Kodak, record exactly what reconstruction value a skipped latent receives, then modify the decoder to fill skipped positions with the predicted mean \(\mu\) instead and recompute BD-rate versus BPG; if the skip-mode gain disappears or reverses, Eq. (9) does not match inference.

Watch

Extended reading notes

Core claim

The paper's central claim is that the entropy model can stop estimating continuous distribution parameters and instead choose from a learned, discrete set of priors, and that this discrete choice is enough to keep rate-distortion performance while cutting entropy-coding complexity. Training uses soft assignment, with a temperature-controlled weight \(\pi_m \propto \exp(-|i-m|/\tau)\) for each prior, and inference rounds the predicted continuous index \(i\) to a table index. The same machinery is extended to a two-dimensional prior set when the probability model has both mean and scale parameters, and a skip mode is added by predicting a binary mask through a temperature-softened categorical reparameterization. With the generalized Gaussian model as the richest prior, the method uses 40 CDF tables where the LUT-based implementation needs 12,800, and reports slightly better rate-distortion performance; FastNIC+Switch gives a 4.10% BD-rate saving over BPG on Kodak, and implanting the priors into ELIC, TCM, and FM-intra reduces entropy coding time by roughly 30-50% with small compression gains.

Load-bearing premise

The load-bearing premise is that a latent omitted from the bitstream is reconstructed as zero during actual decoding, matching the training loss in Eq. (9); the paper never states whether the deployed decoder instead fills skipped positions with the predicted mean \(\mu\), and if it does, the reported skip-mode rate savings may not reproduce.

Editorial extensions

If this is right

  • Richer probability models such as the generalized Gaussian can be adopted in a codec at almost no entropy-coding cost, since the network predicts an index instead of shape and scale parameters.
  • Storing a few dozen shared CDF tables instead of thousands reduces memory traffic during arithmetic coding, which is where most of the measured time saving comes from.
  • Skip mode converts the many near-zero-rate latents in natural images into direct complexity savings; the paper reports skip ratios around 68-72% and larger gains when the prior set is small.
  • A practical lightweight codec can beat BPG on Kodak with encode complexity below 12 KMACs/pixel and decode below 10 KMACs/pixel, and roughly 100 ms per 4K image on the tested hardware.
  • Because the prior set is inherited rather than learned per architecture, existing zero-center-quantization codecs such as ELIC, TCM, and FM-intra can absorb the method with no change to their transforms.

Reading between the lines

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

  • Beyond the paper: the same switchable-prior design could become a rate-complexity knob, since the reported trend shows skip gains and time savings grow as the prior set shrinks; a deployment could expose prior-set size as a tunable parameter.
  • Beyond the paper: because the index is predicted from shared hyperlatents, the mechanism should transfer to learned video compression or multi-frame coding, where entropy models are updated per frame; the paper tests still images only.
  • Beyond the paper: the prior set acts as a learned vector quantizer over distributions, which suggests future work could add a rate penalty on the index itself or train the set under an explicit entropy bottleneck, neither of which the paper explores.
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

1 major / 4 minor

Summary. The paper proposes a switchable-priors method for neural image compression. Instead of predicting continuous parameters of a parametric probabilistic model, the entropy model predicts an index into a finite set of trainable prior distributions (CDF tables), which are pre-stored and shared between encoder and decoder. This is extended to multi-dimensional prior sets, a Top-K approximation for training, a skip mode that omits low-information latents, and reuse of the main-latent prior set for hyperlatents. The authors also introduce FastNIC, a lightweight hyperprior-based codec using stride-2 convolutions and FasterNet blocks. Experiments on Kodak, Tecnick, CLIC, and USTC-TD with FastNIC and with existing models (Shallow-2layer, ELIC, TCM, FM-intra) report consistent BD-rate improvements and reduced entropy coding time relative to LUT-based and dynamically computed CDF baselines.

Significance. If the results hold, the method is practically significant: it decouples entropy-coding complexity from the complexity of the assumed probabilistic model, enabling generalized Gaussian or mixture models with only tens of CDF tables, and it yields a competitive lightweight codec (FastNIC) with sub-12/10 KMACs per pixel encode/decode. The paper is commendably thorough: the switchable-priors idea is validated across five architectures, four datasets, four bitrate points, and with ablations on prior-set size, Top-K acceleration, hyperprior reuse, and multi-dimensional priors. The main correctness risk is the skip-mode training/inference mismatch described in the major comments; until that is resolved, the headline skip-mode gains should be treated with caution.

major comments (1)
  1. [Sec. III-D, Eq. (9)] The training objective for the skip mode is L = R(ŷ)·b̃ + R(ẑ) + λD(x, gs(ŷ·b̃)), which computes distortion from gs(ŷ·b̃), i.e., skipped latent positions are set to zero in the latent space. At inference, the codec codes zero-centered residuals ⌊y−μ⌉ and the decoder adds μ to the decoded residual (Eq. (13)). The paper never states what value the decoder substitutes for a skipped latent position. If the decoder reconstructs that position as μ (the natural choice, because the omitted residual symbol is 0), then the mask is trained against the wrong distortion: Eq. (9) assumes a zero reconstruction, which is generally farther from the original latent than μ. The skip decisions would then be suboptimal for the actual decoder, and the reported skip-mode gains (Table I: FastNIC -4.10% vs -3.51% without skip; Table II: 30-50% entropy coding time reductions; Tables III and VII likewise) may not reproduce. Please specify the decoder's handling of skipped positions; if it is μ, revise the training objective to D(x, gs(ŷ·b̃ + μ·(1−b̃))) (or the equivalent for the actual reconstruction rule) and rerun the affected experiments. If it is indeed 0, please explain how a zero latent is obtained from the zero-centered residual coder in Eq. (13).
minor comments (4)
  1. [Sec. V-A2] The statement "All experiments were conducted with the same random seed to ensure consistency" does not establish robustness; please report variance over multiple seeds or provide confidence intervals for the main BD-rate comparisons, especially the headline numbers in Tables I and II.
  2. [Sec. III-D and Fig. 1] The skip-mode encode/decode procedure would benefit from a short pseudocode block, since the reconstruction rule for skipped latents is essential for reproducibility and is not currently stated explicitly anywhere in the paper.
  3. [Sec. III-A, Eq. (4)] The rounding operator in ⌊Clip(i,1,M)⌉ is not defined for ties (e.g., i = 1.5); please specify the tie-breaking rule or use floor/ceil consistently.
  4. [Table I] The footnote markers in the methods column are ambiguous: 'Switch w/o skip 4' and 'Switch4' both carry footnote 4, but the footnote text defines 'Switch' and 'Switch w/o skip'; please clarify the markers so readers know which rows include the skip mode.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: switchable priors are a trained codebook scheme, and all headline numbers are measured on held-out datasets.

full rationale

I walked the derivation chain from Eq. (1) through Eq. (12). The switchable-priors rate in Eq. (2) is a definition of the proposed training objective: a weighted sum over the trainable prior set. Eqs. (3)-(4) define the soft assignment used in training and the rounding rule used at inference. These are constructions of the method, not predictions fitted to test data. The BD-rate and coding-time results in Tables I-VII are empirical measurements on Kodak, Tecnick, CLIC, and USTC-TD; they are not outputs of a derivation that re-imports its own inputs. Table VIII's 'estimated' row is the model's own cross-entropy, used only as an internal diagnostic of how much the finite prior set costs, not as evidence that the method predicts its own results. The paper builds on the authors' prior GGM work [22], but it explicitly states that the advantage of GGM is not a contribution of this study and evaluates switchable priors with GM, GMM, and GGM alike; the citation is not load-bearing. No uniqueness theorem from the authors' prior work is invoked to force the design. The self-citations to [33] and [51] concern implementation details (kernel sizes, quantization surrogates) and are not load-bearing for the central claim. The only notable concern, in Sec. III-D Eq. (9), is that the skip-mode training objective computes distortion from gs(y_hat * b_tilde) while the paper does not state whether a skipped latent is decoded as zero or as the predicted mean mu; if the decoder uses mu, there is a train/test distortion mismatch. That is a correctness or reproducibility gap, not a circular reduction, and it does not affect the circularity score.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new physical entities or mediators are introduced; 'skip mode' and 'prior set' are software/model components. The central claim depends on a small number of hand-chosen hyperparameters (M, temperatures, Top-K) and on domain assumptions about the latent distribution and training-inference consistency.

free parameters (4)
  • Number of priors M = 40 (default), 5-40 ablated
    Hand-chosen hyperparameter controlling the complexity/performance tradeoff of the prior set; not derived from theory.
  • Temperature schedule tau = tau = tau0 * exp(-0.01 * epoch), tau0 = 0.05 * M
    Annealing schedule hand-chosen to approach one-hot assignments; affects train-test consistency.
  • Skip-mode temperature t = t = 0.4 * exp(-0.01 * epoch)
    Gumbel-Softmax temperature schedule, hand-chosen.
  • Top-K = 2
    Number of nearest priors used in the training rate estimate, hand-chosen.
assumptions (4)
  • domain assumption The true distribution of latents can be well approximated by a finite set of M parametric distributions.
    The entire method rests on the existence of a small codebook that covers the latent distribution; failure would cause rate loss from index rounding. Tables III/VIII show small loss for M=40 on tested datasets, but this is not guaranteed in general.
  • domain assumption Weighted soft rate with temperature annealing (Eqs. 2-3) converges to the hard index rate (Eq. 4) at test.
    If the softmax weights do not converge to one-hot, the trained entropy model's index predictions may be mismatched at inference. The paper anneals temperature but provides no proof of convergence.
  • ad hoc to paper The decoder's reconstruction of skipped latents matches the training objective in Eq. (9).
    Eq. 9 uses gs(ŷ·b̃), i.e., skipped latents are set to zero; it is assumed the decoder does the same. If the decoder instead reconstructs skipped residuals as zero and adds the mean μ, training and test diverge.
  • standard math Standard rate-distortion Lagrangian and mixed quantization surrogate apply.
    Uses the established NIC training objective from [4], [5] without new justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Switchable Priors for Neural Image Compression." pith.science (2026). https://pith.science/paper/2F4PHJUK

@misc{pith2026250416586,
  author       = {Pith},
  title        = {Pith review of: Learning Switchable Priors for Neural Image Compression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2F4PHJUK}},
  note         = {Machine review of arXiv:2504.16586}
}
read the original abstract

Neural image compression (NIC) usually adopts a predefined family of probabilistic distributions as the prior of the latent variables, and meanwhile relies on entropy models to estimate the parameters for the probabilistic family. More complex probabilistic distributions may fit the latent variables more accurately, but also incur higher complexity of the entropy models, limiting their practical value. To address this dilemma, we propose a solution to decouple the entropy model complexity from the prior distributions. We use a finite set of trainable priors that correspond to samples of the parametric probabilistic distributions. We train the entropy model to predict the index of the appropriate prior within the set, rather than the specific parameters. Switching between the trained priors further enables us to embrace a skip mode into the prior set, which simply omits a latent variable during the entropy coding. To demonstrate the practical value of our solution, we present a lightweight NIC model, namely FastNIC, together with the learning of switchable priors. FastNIC obtains a better trade-off between compression efficiency and computational complexity for neural image compression. We also implanted the switchable priors into state-of-the-art NIC models and observed improved compression efficiency with a significant reduction of entropy coding complexity.

Figures

Figures reproduced from arXiv: 2504.16586 by the authors.

Figure 1
Figure 1. Diagram of coding process of the proposed switchable priors method. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Diagram of a typical neural image compression model. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Diagram of training and testing stages for dynamically computing CDF, LUTs-based method, and the proposed switchable priors method. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Frequency of the rate of latent variables collected from the Kodak set [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Framework of the proposed FastNIC model. (a) shows the diagram of FastNIC. In the entropy model, the module hm(mean) is used to predict the mean parameter, while the module he(entropy) predicts other parameters of the probabilistic models. The detailed network structur…
Figure 6
Figure 6. Figure 6: Performance of our proposed FastNIC+Switch. (a) shows the performance of FastNIC+Switch trained for MSE. (b) shows the performance of FastNIC+Switch trained for MS-SSIM. “Switch” represents our proposed switchable priors method. • Cheng-ckbd (CVPR2021) [37] denotes the…
Figure 7
Figure 7. Figure 7: Rate-distortion performance of different methods on the Kodak dataset. All models are trained for MSE and evaluated with PSNR. “Switch” represents [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Performance comparison of the proposed FastNIC+Switch model with other NIC models on the Kodak dataset. All models are trained for MSE and [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Skip ratio of latents in FastNIC model. (a) and (b) shows the frequency [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Performance of incorporating skip mode concerning the number of [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: Rate saving at different bitrates and skip ratios. The results are [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Relationship between coding time and skip ratio. The results are [PITH_FULL_IMAGE:figures/full_fig_p012_12.png]
Figure 13
Figure 13. Figure 13: Visualization of the learned prior set with [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 14
Figure 14. Figure 14: Visualization of the learned prior set with [PITH_FULL_IMAGE:figures/full_fig_p013_14.png]
Figure 15
Figure 15. Figure 15: Visualization of learned prior set of GMM on Cheng-ckbd. Each [PITH_FULL_IMAGE:figures/full_fig_p014_15.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 19 canonical work pages

  1. [22]

    Generalized Gaussian model for learned image compression,

    H. Zhang, L. Li, and D. Liu, “Generalized Gaussian model for learned image compression,” IEEE Transactions on Image Processing , vol. 34, pp. 1950–1965, 2025

    work page 1950

  2. [1]

    Theoretical foundations of transform coding,

    V . Goyal, “Theoretical foundations of transform coding,” IEEE Signal Processing Magazine, vol. 18, no. 5, pp. 9–21, 2001

    2001

  3. [2]

    The JPEG still picture compression standard,

    G. K. Wallace, “The JPEG still picture compression standard,” Commu- nications of the ACM , vol. 34, no. 4, pp. 30–44, 1991

    work page 1991

  4. [3]

    Overview of the versatile video coding (VVC) standard and its applications,

    B. Bross, Y .-K. Wang, Y . Ye, S. Liu, J. Chen, G. J. Sullivan, and J.- R. Ohm, “Overview of the versatile video coding (VVC) standard and its applications,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 31, no. 10, pp. 3736–3764, 2021

    2021

  5. [4]

    End-to-end optimized image compression,

    J. Ball ´e, V . Laparra, and E. P. Simoncelli, “End-to-end optimized image compression,” arXiv:1611.01704, 2016

    arXiv 2016

  6. [5]

    Channel-wise autoregressive entropy models for learned image compression,

    D. Minnen and S. Singh, “Channel-wise autoregressive entropy models for learned image compression,” in IEEE International Conference on Image Processing (ICIP) , 2020, pp. 3339–3343

    2020

  7. [6]

    ELIC: Efficient learned image compression with unevenly grouped space-channel con- textual adaptive coding,

    D. He, Z. Yang, W. Peng, R. Ma, H. Qin, and Y . Wang, “ELIC: Efficient learned image compression with unevenly grouped space-channel con- textual adaptive coding,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2022, pp. 5708–5717

    2022

  8. [7]

    M2T: Masking Transform- ers twice for faster decoding,

    F. Mentzer, E. Agustson, and M. Tschannen, “M2T: Masking Transform- ers twice for faster decoding,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2023, pp. 5317–5326

    2023

Show all 51 references
  1. [8]

    Flexible coding order for learned image compression,

    Y . Li, H. Zhang, and D. Liu, “Flexible coding order for learned image compression,” in IEEE International Conference on Visual Communica- tions and Image Processing (VCIP) , 2023, pp. 1–5

    2023

  2. [9]

    Learned image compression with Gaussian- Laplacian-Logistic mixture model and concatenated residual modules,

    H. Fu, F. Liang, J. Lin, B. Li, M. Akbari, J. Liang, G. Zhang, D. Liu, C. Tu, and J. Han, “Learned image compression with Gaussian- Laplacian-Logistic mixture model and concatenated residual modules,” IEEE Transactions on Image Processing, vol. 32, pp. 2063–2076, 2023

    2023

  3. [10]

    MLIC++: Linear complexity multi-reference entropy modeling for learned image compression,

    W. Jiang and R. Wang, “MLIC++: Linear complexity multi-reference entropy modeling for learned image compression,” https://openreview. net/forum?id=hxIpcSoz2t, 2023

    2023

  4. [11]

    Frequency- aware transformer for learned image compression,

    H. Li, S. Li, W. Dai, C. Li, J. Zou, and H. Xiong, “Frequency- aware transformer for learned image compression,” https://openreview. net/forum?id=UVf72oSajj, 2024

    2024

  5. [12]

    GroupedMixer: An entropy model with group-wise token-mixers for learned image compression,

    D. Li, Y . Bai, K. Wang, J. Jiang, X. Liu, and W. Gao, “GroupedMixer: An entropy model with group-wise token-mixers for learned image compression,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 10, pp. 9606–9619, 2024

    2024

  6. [13]

    Causal context adjustment loss for learned image compression,

    M. Han, S. Jiang, S. Li, X. Deng, M. Xu, C. Zhu, and S. Gu, “Causal context adjustment loss for learned image compression,” in Advances in Neural Information Processing Systems , 2024

    2024

  7. [14]

    End- to-end learning-based image compression with a decoupled framework,

    Z. Zhang, S. Esenlik, Y . Wu, M. Wang, K. Zhang, and L. Zhang, “End- to-end learning-based image compression with a decoupled framework,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 34, no. 5, pp. 3067–3081, 2024

    2024

  8. [15]

    Causal contextual prediction for learned image compression,

    Z. Guo, Z. Zhang, R. Feng, and Z. Chen, “Causal contextual prediction for learned image compression,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 4, pp. 2329–2341, 2022

    2022

  9. [16]

    Multirate progres- sive entropy model for learned image compression,

    C. Li, S. Yin, C. Jia, F. Meng, Y . Tian, and Y . Liang, “Multirate progres- sive entropy model for learned image compression,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 8, pp. 7725– 7741, 2024

    2024

  10. [17]

    NLIC: Non-uniform quantization-based learned image compression,

    Z. Ge, S. Ma, W. Gao, J. Pan, and C. Jia, “NLIC: Non-uniform quantization-based learned image compression,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 34, no. 10, pp. 9647– 9663, 2024

    2024

  11. [18]

    Efficient contextformer: Spatio-channel window attention for fast context model- ing in learned image compression,

    A. B. Koyuncu, P. Jia, A. Boev, E. Alshina, and E. Steinbach, “Efficient contextformer: Spatio-channel window attention for fast context model- ing in learned image compression,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 8, pp. 7498–7511, 2024

    2024

  12. [19]

    Variational image compression with a scale hyperprior,

    J. Ball ´e, D. Minnen, S. Singh, S. J. Hwang, and N. Johnston, “Variational image compression with a scale hyperprior,” arXiv:1802.01436, 2018

    2018 arXiv

  13. [20]

    Joint autoregressive and hier- archical priors for learned image compression,

    D. Minnen, J. Ball ´e, and G. Toderici, “Joint autoregressive and hier- archical priors for learned image compression,” in Advances in Neural Information Processing Systems , 2018, pp. 10 794–10 803

    2018

  14. [21]

    Learned image com- pression with discretized Gaussian mixture likelihoods and attention modules,

    Z. Cheng, H. Sun, M. Takeuchi, and J. Katto, “Learned image com- pression with discretized Gaussian mixture likelihoods and attention modules,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, pp. 7936–7945

    2020

  15. [23]

    An introduction to arithmetic coding,

    G. G. Langdon, “An introduction to arithmetic coding,” IBM Journal of Research and Development , vol. 28, no. 2, pp. 135–149, 1984

    1984

  16. [24]

    Asymmetric numeral systems: Entropy coding combining speed of Huffman coding with compression rate of arithmetic coding,

    J. Duda, “Asymmetric numeral systems: Entropy coding combining speed of Huffman coding with compression rate of arithmetic coding,” arXiv preprint arXiv:1311.2540 , 2013

    2013 arXiv

  17. [25]

    Integer networks for data compression with latent-variable models,

    J. Ball ´e, N. Johnston, and D. Minnen, “Integer networks for data compression with latent-variable models,” https://openreview.net/forum? id=S1zz2i0cY7, 2019

    2019

  18. [26]

    Computationally-efficient neural image com- pression with shallow decoders,

    Y . Yang and S. Mandt, “Computationally-efficient neural image com- pression with shallow decoders,” in IEEE/CVF International Conference on Computer Vision (ICCV) , 2023, pp. 530–540

    2023

  19. [27]

    Learned image compression with mixed Transformer-CNN architectures,

    J. Liu, H. Sun, and J. Katto, “Learned image compression with mixed Transformer-CNN architectures,” inIEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2023, pp. 14 388–14 397

    2023

  20. [28]

    Neural video compression with feature modulation,

    J. Li, B. Li, and Y . Lu, “Neural video compression with feature modulation,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 26 099–26 108

    2024

  21. [29]

    End-to-end optimized versatile image compression with wavelet-like transform,

    H. Ma, D. Liu, N. Yan, H. Li, and F. Wu, “End-to-end optimized versatile image compression with wavelet-like transform,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 3, pp. 1247– 1263, 2022

    2022

  22. [30]

    Wavelet-like transform-based technology in response to the call for proposals on neural network-based image coding,

    C. Dong, H. Ma, H. Zhang, C. Gao, L. Li, and D. Liu, “Wavelet-like transform-based technology in response to the call for proposals on neural network-based image coding,” arXiv preprint arXiv:2403.05937 , 2024

    2024 arXiv

  23. [31]

    Estimating or propagating gradients through stochastic neurons for conditional computation,

    Y . Bengio, N. L ´eonard, and A. Courville, “Estimating or propagating gradients through stochastic neurons for conditional computation,” arXiv preprint arXiv:1308.3432, 2013

    2013 arXiv

  24. [32]

    Advancing the rate-distortion-computation frontier for neural image compression,

    D. Minnen and N. Johnston, “Advancing the rate-distortion-computation frontier for neural image compression,” in IEEE International Confer- ence on Image Processing (ICIP) , 2023, pp. 2940–2944

    2023

  25. [33]

    Practical learned image compression with online encoder optimization,

    H. Zhang, F. Mei, J. Liao, L. Li, H. Li, and D. Liu, “Practical learned image compression with online encoder optimization,” in Picture Coding Symposium (PCS), 2024, pp. 1–5

    2024

  26. [34]

    Slimmable compressive autoencoders for practical neural image compression,

    F. Yang, L. Herranz, Y . Cheng, and M. G. Mozerov, “Slimmable compressive autoencoders for practical neural image compression,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 4996–5005

    2021

  27. [35]

    Memory-efficient learned image compression with pruned hyperprior module,

    A. Luo, H. Sun, J. Liu, and J. Katto, “Memory-efficient learned image compression with pruned hyperprior module,” in IEEE International Conference on Image Processing (ICIP) , 2022, pp. 3061–3065

    2022

  28. [36]

    EVC: Towards real-time neural image compression with mask decay,

    G.-H. Wang, J. Li, B. Li, and Y . Lu, “EVC: Towards real-time neural image compression with mask decay,” https://openreview.net/forum?id= XUxad2Gj40n, 2023

    2023

  29. [37]

    Checkerboard context model for efficient learned image compression,

    D. He, Y . Zheng, B. Sun, Y . Wang, and H. Qin, “Checkerboard context model for efficient learned image compression,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 14 766–14 775

    2021

  30. [38]

    AlphaVC: High-performance and efficient learned video compression,

    Y . Shi, Y . Ge, J. Wang, and J. Mao, “AlphaVC: High-performance and efficient learned video compression,” in European Conference on Computer Vision (ECCV) , 2022, pp. 616–631

    2022

  31. [39]

    Selective compression learning of latent representations for variable-rate image compression,

    J. Lee, S. Jeong, and M. Kim, “Selective compression learning of latent representations for variable-rate image compression,” in Advances in Neural Information Processing Systems , 2022

    2022

  32. [40]

    CompressAI: A PyTorch library and evaluation platform for end-to-end compression research,

    J. B ´egaint, F. Racap ´e, S. Feltman, and A. Pushparaja, “CompressAI: A PyTorch library and evaluation platform for end-to-end compression research,” arXiv preprint arXiv:2011.03029 , 2020

    2011 arXiv

  33. [41]

    Learned image compression with fixed- point arithmetic,

    H. Sun, L. Yu, and J. Katto, “Learned image compression with fixed- point arithmetic,” in Picture Coding Symposium (PCS) , 2021, pp. 1–5. 16

    2021

  34. [42]

    Post- training quantization for cross-platform learned image compression,

    D. He, Z. Yang, Y . Chen, Q. Zhang, H. Qin, and Y . Wang, “Post- training quantization for cross-platform learned image compression,” arXiv preprint arXiv:2202.07513 , 2022

    2022 arXiv

  35. [43]

    Nonlinear transform coding,

    J. Ball ´e, P. A. Chou, D. Minnen, S. Singh, N. Johnston, E. Agustsson, S. J. Hwang, and G. Toderici, “Nonlinear transform coding,” IEEE Journal of Selected Topics in Signal Processing , vol. 15, no. 2, pp. 339–353, 2021

    2021

  36. [44]

    Categorical reparameterization with Gumbel-softmax,

    E. Jang, S. Gu, and B. Poole, “Categorical reparameterization with Gumbel-softmax,” https://openreview.net/forum?id=rkE3y85ee, 2017

    2017

  37. [45]

    Run, don’t walk: Chasing higher FLOPS for faster neural networks,

    J. Chen, S.-h. Kao, H. He, W. Zhuo, S. Wen, C.-H. Lee, and S.- H. G. Chan, “Run, don’t walk: Chasing higher FLOPS for faster neural networks,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023, pp. 12 021–12 031

    2023

  38. [46]

    A unified end-to-end framework for efficient deep image compression,

    J. Liu, G. Lu, Z. Hu, and D. Xu, “A unified end-to-end framework for efficient deep image compression,” arXiv preprint arXiv:2002.03370 , 2020

    2002 arXiv

  39. [47]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv:1412.6980, 2015

    2015 arXiv

  40. [48]

    TESTIMAGES: A large-scale archive for testing visual devices and basic image processing algorithms,

    N. Asuni and A. Giachetti, “TESTIMAGES: A large-scale archive for testing visual devices and basic image processing algorithms,” in Smart Tools and Apps for Graphics - Eurographics Italian Chapter Conference, 2014, pp. 63–70

    2014

  41. [49]

    USTC-TD: A test dataset and benchmark for image and video coding in 2020s,

    Z. Li, J. Liao, C. Tang, H. Zhang, Y . Li, Y . Bian, X. Sheng, X. Feng, Y . Li, C. Gao, L. Li, D. Liu, and F. Wu, “USTC-TD: A test dataset and benchmark for image and video coding in 2020s,” accepted to IEEE Transactions on Multimedia , 2025

    2025

  42. [50]

    Calculation of average PSNR differences between RD- curves,

    G. Bjontegaard, “Calculation of average PSNR differences between RD- curves,” ITU SG16, Tech. Rep. VCEG-M33, 2001

    2001

  43. [51]

    On uniform scalar quantization for learned image compression,

    H. Zhang, L. Li, and D. Liu, “On uniform scalar quantization for learned image compression,” arXiv preprint arXiv:2309.17051 , 2023. APPENDIX A COMPARISON OF SKIPPING MECHANISM First, we summarize the implementation of previous ap- proaches. The skipping mechanism proposed in ...

    2023 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.