Pith. sign in

REVIEW 5 major objections 6 minor 1 cited by

An Efficient Adaptive Compression Method for Human Perception and Machine Vision Tasks

T0 review · 5 major / 6 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 The paper claims that a lightweight add-on, EAC, can partition a neural codec's latent features into task-specific subsets so that multiple machine vision tasks run at over 33% lower bitrate while human-viewable reconstruction quality is…

desk verdict A plausible incremental extension of the authors' ICMH-Net to multi-task and video settings, but the described architecture doesn't actually enforce the latent partition that the bitrate-savings and human-parity claims rest on. read the letter →

arxiv 2501.04329 v1 pith:WCCUWXYK submitted 2025-01-08 cs.CV

Lei Liu , Zhenghao Chen , Zhihao Hu , Dong Xu This is my paper · ORCID
classification cs.CV
keywords neuralimagecompressionvideomachinevisionlatentsubsetselectionbinarymaskstask-specificadapterparameter-efficientfine-tuninghuman-machine
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 Efficient Adaptive Compression (EAC), a lightweight add-on that makes existing neural image and video codecs serve both human viewers and multiple machine vision tasks at once. The core claim is that a codec's quantized latent features can be partitioned, by learned binary masks, into task-specific subsets, so that each machine task receives only the bits it needs while human vision can still request the full latent. The authors report that this saves more than 33% bitrate over Ballé2018-based baselines at matched segmentation and detection accuracy, and that even with classic neural codecs the results beat the modern standard codec VTM across most bitrates. A second component, a task-specific adapter trained with parameter-efficient delta-tuning, lets downstream networks (classification, segmentation, detection, action recognition, video object segmentation) be adapted cheaply without retraining them. If correct, existing learned codecs can be upgraded for machine-centric pipelines without redesigning them.

What carries the argument

The adaptive compression module: it multiplies the quantized latent ŷ elementwise by n binary masks (generated from hyperprior μ, σ and, in video, reference-frame features via Gumbel-softmax), losslessly transmits only the selected subsets, then reconstructs and aggregates them, filling unselected elements with the predicted mean μ to form task-specific latents. The task-specific adapter is a small residual-block network inserted before a frozen task network; it projects the reconstructed image or frame (plus, for video, several reconstructed reference frames) into a feature map that nudges the task network toward the correct output. These two pieces together carry the argument: mask selection buys the bitrate savings, and μ-filling plus the adapter preserves task accuracy.

What would settle it

Take a trained EAC (NIC) model and replace its learned binary masks with random binary masks of the same density on a detection benchmark: if mAP@0.5 stays within a small margin of the reported level, then the mask predictor is not the source of the task accuracy. Alternatively, replace the μ-filling of unselected latent entries with zeros: if task accuracy is unchanged, the aggregation step's dependence on the hyperprior mean is not load-bearing.

Watch

Extended reading notes

Core claim

EAC claims that the information a downstream vision task needs is a subset of the quantized latent, and that subset can be identified from the hyperprior statistics (μ, σ) — plus, for video, from reference frames — by a small predictor network that outputs binary masks. The selected subset is transmitted and decoded; the unselected positions are filled with the predicted mean μ before reconstruction. For the i-th task, reconstructed subsets are summed and unselected entries filled with μ to yield a task-specific reconstruction. The full latent is transmitted only when a high-quality human-viewable reconstruction is required. With codec weights frozen, only the masks and a tiny adapter are trained, so the scheme is a drop-in extension of Ballé2018, Cheng2020, DVC, and FVC.

Load-bearing premise

The whole scheme rests on the assumption that the information a downstream vision task needs lives in a subset of the codec's latent features that can be predicted from the hyperprior statistics (plus reference frames in video), and that filling the discarded positions with the average value μ leaves reconstructions accurate enough for the task.

Editorial extensions

If this is right

  • Any existing neural codec with a hyperprior (e.g., Ballé2018, Cheng2020) can be wrapped by EAC to serve multiple machine tasks without retraining the codec, so the reported savings should transfer to newer codecs that expose latent statistics.
  • Machine-vision clients can operate at a fraction of the bitrate by receiving only their task's subset; the same bitstream can still be upgraded to a full human-quality reconstruction when needed, enabling progressive transmission.
  • For video, using reference frames as additional mask input improves the selection, so temporal redundancy is exploited for machine tasks the same way it is for human viewing.
  • Because only the adapter (about 0.2M parameters) is optimized, adapting a codec to a new task costs orders of magnitude less than full fine-tuning of the task network.
  • The two-stage training (predictors first, then adapter) lets the mask and the adapter be optimized separately, so the scheme is compatible with frozen codecs and frozen task networks.

Reading between the lines

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

  • The μ-filling trick suggests a general recipe for any autoencoder with a tractable prior: unselected latent entries can be replaced by their conditional mean at no extra bit cost, and testing it on other modalities (audio, point clouds, medical images) would show whether the recipe is codec-specific.
  • The mask predictor learns a task-relevance map over latent channels and positions; equipping it with uncertainty estimates could let a system decide per-image whether to request the full latent, an implicit rate-control policy the paper does not explore.
  • A natural stress test is distribution shift: since masks are predicted from hyperprior statistics, images whose statistics differ from training (e.g., night scenes, medical scans) may need more subsets or a fallback to full transmission, a condition the paper does not report.
  • The comparison to VTM is at the system level (codec plus frozen task network); a more complete machine-vision benchmark would tune VTM's bitrate allocation for task features, a direction the paper leaves open.
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

5 major / 6 minor

Summary. This paper proposes Efficient Adaptive Compression (EAC), a plug-in module for existing learned image and video codecs (Ballé2018, Cheng2020, DVC, FVC) to serve multiple downstream machine vision tasks (classification, segmentation, detection, action recognition, video object segmentation) while preserving human-viewing quality. The core idea is to partition the quantized latent into n subsets via binary masks generated from hyperprior statistics (and, for video, reference frames), transmit only the subsets needed by each task, and fill unselected entries with the predicted mean μ. A second contribution is a lightweight task-specific adapter, trained with the task network frozen, that adapts the reconstructed frame for each task. Training is two-stage: predictors are trained with rate–distortion–task losses, then adapters are trained alone. Experiments on VOC, COCO, ILSVRC2012, UCF101, and DAVIS report bitrate savings over the base codecs plus task-network pipelines and claim parity with the base codecs for human viewing.

Significance. If the results hold, EAC offers a simple and broadly applicable way to redirect established compression architectures toward machine tasks, with a very small parameter overhead (adapter ~0.2% of the task network) and a clean separation of codec and task-network tuning. The adapter-based delta-tuning is a sensible, efficient strategy. However, the central mechanism's correctness depends on an unstated and unenforced partition property of the learned masks, and the experimental evidence is presented entirely through unreadable RD subfigures without tables, error bars, or code release. The claimed 33% bitrate savings and human-vision parity cannot currently be verified. The idea is promising and the architectural modifications are local, so the paper is a candidate for major revision rather than rejection.

major comments (5)
  1. [Section III-A, Fig. 1, Eqs. (3)–(4)] The paper describes the adaptive compression module as partitioning the quantized latent ŷ into n subsets via binary masks, and claims that for human vision 'we add all reconstructed latent features ... obtaining full latent feature ŷ' (Section III-A). This is only correct if the masks are a partition (disjoint and covering). However, each mask is produced by an independent predictor (with Gumbel-Softmax/Max) and the losses in Eqs. (3)–(4) contain no term enforcing disjointness or full coverage. Nothing prevents two predictors from selecting the same latent elements (inflating total rate) or all predictors from omitting some elements (degrading the human reconstruction). Please add an explicit partition constraint (e.g., a single softmax over n+1 classes, or a regularizer) and, in the experiments, report the overlap/coverage statistics of the learned masks.
  2. [Section III-A, Section IV-C] The paper does not state how the binary masks are encoded and transmitted. If the masks are regenerated at the decoder, the predictor must be deterministic and the paper should say so; if they are encoded, their bitrate must be added to the reported savings. In addition, because the masks may overlap, the sum R(ŷ1)+...+R(ŷn) is not guaranteed to equal R(ŷ), so the claimed bitrate savings are not well defined. Please clarify the mask transmission scheme and provide a breakdown of bitrate: mask overhead versus latent subvectors.
  3. [Section IV-C, Fig. 5] The central quantitative claims, including 'saves more than 33% bit-rate' at 0.79 mIoU and at 0.67 mAP@0.5, are supported only by RD subfigures with unreadable labels and no numerical tables. No error bars, standard deviations, or number of runs are given, and no code or data are released. The 33% figure cannot be checked from the text. Please add tables with the actual bpp, task-metric, and PSNR values for at least the key operating points, and report variance or standard deviations across at least three runs.
  4. [Section IV-C, 'Human Vision Results'] The claim that human-vision performance is maintained is circular: it asserts that because the same pre-trained codec and the 'full latent feature' are used, performance is similar to the baseline. This is only true under the partition assumption of Comment 1, and even then, the full-latent reconstruction would be identical to the baseline only if the masks perfectly reconstruct ŷ, which is not guaranteed given that unselected elements are filled with μ for the machine branches. Moreover, no numerical PSNR values are reported; the paper should compare PSNR (or MS-SSIM) of 'Ours' and the baseline at matched bitrates.
  5. [Section III-E, Eq. (4)] The paper states that a cumulative objective over T frames mitigates error propagation caused by transmitting only subsets of motion and residual latents in video. No ablation is provided: there is no experiment varying T, no comparison with single-frame training, and no measurement of reconstruction drift over time. Please either add evidence for this claim or soften the claim to what the experiments actually support.
minor comments (6)
  1. [Section IV-C] 'It is worse noting' should be 'It is worth noting' (Section IV-C). Section III-C misspells 'predictor' as 'Preditor'; Section III-D misspells 'reconstructed' as 'reconsturcted'; Section III-C has 'hyperperior' for 'hyperprior'.
  2. [Figures 5–8] The figure captions contain garbled Unicode tokens (e.g., '/uni00000013/...') that obscure the figure content; these artifacts must be cleaned in the final version.
  3. [Eq. (4)] Eq. (4) has an unbalanced parenthesis in the displayed formula; please fix the typesetting.
  4. [Notation] The notation is inconsistent: 'Ballé2018' and 'Cheng2020' are used both as abbreviations and as reference names (e.g., in captions and text). Use 'Ballé et al. [15]' and 'Cheng et al. [16]' consistently.
  5. [Section IV-B] The paper claims to outperform VTM 'across most bit-rates' but does not specify the bitrate range or whether VTM is used in intra-only mode for image comparison; please state the exact configuration.
  6. [Table I] Table I gives parameter counts but does not specify what is included in 'Adapter in EAC (NIC)' and 'Adapter in EAC (NVC)'; please clarify the adapter architecture and parameter count.

Circularity Check

1 steps flagged · score 1.0 of 10

Only minor by-construction element is human-vision parity, which the paper itself states; machine-vision claims are empirical and non-circular.

  1. self definitional [Section IV-C, 'Human Vision Results']
    "As we adopt the same pre-trained codecs as our baseline method, and use the full latent feature for human vision. So the human vision performance in our methods (e.g., 'Ours (Cheng2020)') is similar to our baseline methods (e.g., 'Cheng2020')."

    The paper presents 'maintaining the quality of human vision' as a result, but this result is entailed by the design choice of transmitting the full latent through the same frozen codec as the baseline. No independent prediction or fitted quantity is involved; the human-vision RD curve coincides with the baseline curve by construction. This is a minor self-definitional element and does not bear on the central machine-vision claims, which are empirically measured on held-out splits.

full rationale

The paper's central machine-vision claims are empirical systems results: the predictors and task-specific adapters are trained with rate-distortion and task losses (Eqs. 3-5) on training splits and evaluated on held-out test splits (VOC, COCO, UCF101, DAVIS). No fitted parameter is renamed as a prediction; the reported >33% bitrate savings and task-accuracy gains are measured RD points, not identities. Citations to the authors' conference version [25] provide predictor architecture details and a comparison baseline, but the multi-task improvements are supported by this paper's own ablations and test-set measurements, so the self-citations are not load-bearing. The skeptic's concern that the independently predicted binary masks may not form a true partition of the latent is a correctness or robustness risk, not a circularity, because the paper's rate and reconstruction claims are not defined to be true by the mask equations; they are asserted and then empirically evaluated. The only by-construction element is the human-vision parity, which the paper openly and correctly attributes to using the same pre-trained codec and the full latent for human vision. Overall, no significant circularity.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The listed free parameters and domain assumptions are the load-bearing choices that the experiments do not independently justify. The method introduces no new physical entities; the adaptive compression module and task-specific adapter are architectural components built from existing operations.

free parameters (3)
  • rate-distortion trade-off λ = unreported
    λ in Eq. (3) and Eq. (4) balances bit-rate and task loss; values are chosen per operating point and not stated, so the rate-distortion curves cannot be reproduced.
  • number of latent subsets n = unreported
    The paper defines n subsets and masks but never states how n is set for the image or video experiments, despite it directly controlling bit-rate and task multiplexing.
  • video context length T for cumulative loss = unreported
    Eq. (4) sums over T consecutive frames to counter accumulative errors in motion and residual compression; T is not reported, which affects the video results.
assumptions (5)
  • standard math Arithmetic encoding and decoding are lossless.
    Standard assumption in neural compression, invoked in the transmission module of Section III-A; not a source of error.
  • domain assumption Pre-trained codec latents contain task-relevant structure separable by learned binary masks.
    Section III-A assumes partitioning ŷ into mask-selected subsets preserves per-task information; no analysis of mask information content or capacity is provided.
  • domain assumption Replacing unselected latent elements with the predicted mean μ produces reconstructions adequate for downstream task networks.
    Section III-A aggregation module fills unselected elements with μ; this is asserted to improve performance without excess bit-rate, but no distortion analysis is given.
  • ad hoc to paper Frozen codec and task network parameters suffice; only predictors and adapters need training.
    The two-stage optimization in Section III-E holds all base networks fixed; if the frozen codec features are not separable, the method underperforms.
  • ad hoc to paper Cumulative training over T frames eliminates error propagation in video coding.
    Section III-E uses a cumulative loss as in [40] to mitigate accumulative errors, but no evidence is shown that T-frame training generalizes to arbitrary sequence lengths.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An Efficient Adaptive Compression Method for Human Perception and Machine Vision Tasks." pith.science (2026). https://pith.science/paper/WCCUWXYK

@misc{pith2026250104329,
  author       = {Pith},
  title        = {Pith review of: An Efficient Adaptive Compression Method for Human Perception and Machine Vision Tasks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WCCUWXYK}},
  note         = {Machine review of arXiv:2501.04329}
}
read the original abstract

While most existing neural image compression (NIC) and neural video compression (NVC) methodologies have achieved remarkable success, their optimization is primarily focused on human visual perception. However, with the rapid development of artificial intelligence, many images and videos will be used for various machine vision tasks. Consequently, such existing compression methodologies cannot achieve competitive performance in machine vision. In this work, we introduce an efficient adaptive compression (EAC) method tailored for both human perception and multiple machine vision tasks. Our method involves two key modules: 1), an adaptive compression mechanism, that adaptively selects several subsets from latent features to balance the optimizations for multiple machine vision tasks (e.g., segmentation, and detection) and human vision. 2), a task-specific adapter, that uses the parameter-efficient delta-tuning strategy to stimulate the comprehensive downstream analytical networks for specific machine vision tasks. By using the above two modules, we can optimize the bit-rate costs and improve machine vision performance. In general, our proposed EAC can seamlessly integrate with existing NIC (i.e., Ball\'e2018, and Cheng2020) and NVC (i.e., DVC, and FVC) methods. Extensive evaluation on various benchmark datasets (i.e., VOC2007, ILSVRC2012, VOC2012, COCO, UCF101, and DAVIS) shows that our method enhances performance for multiple machine vision tasks while maintaining the quality of human vision.

Figures

Figures reproduced from arXiv: 2501.04329 by the authors.

Figure 1
Figure 1. (a) The overview of our adaptive compression module, where we simultaneously balance the optimizations for the multiple machine and human vision [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The details of how we implement the adapter with the task-specific [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. (a) The overview of “EAC (NVC)”, where we incorporate our efficient adaptive compression method in neural video compression network. (b) The [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: The multi-tasks (i.e., the segmentation task, the detection task, and the human vision) results for our “EAC (NIC)” compared to the baseline methods on the VOC2007 dataset. We report mIoU, mAP@0.5, and PSNR results. For all codecs, we use the PSANet as the segmentation…
Figure 6
Figure 6. Figure 6: The machine vision tasks (i.e., the classification task, the segmentation task, and the detection task) results for our “EAC (NIC)” compared to ICMH-Net on the ILSVRC2012, VOC2012, and COCO dataset. We report accuracy, mIoU, and mAP results. For all codecs, we use the …
Figure 7
Figure 7. Figure 7: Video action recognition results (a), (b) and video object segmentation results (c) for our “EAC (NVC)” compared to the baseline methods on the [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: (a) Ablation study of our adaptive motion and residual compression [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Visualization of (a) the input frame, (b) the selected feature map [PITH_FULL_IMAGE:figures/full_fig_p009_9.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. HEMGS: A Hybrid Entropy Model for 3D Gaussian Splatting Data Compression

    cs.CV 2024-11 conditional novelty 6.0 of 10

    HEMGS compresses 3D Gaussian Splatting data with a single variable-rate model and a joint autoregressive/hyperprior entropy model, achieving roughly 40% smaller files than HAC at similar quality.

Reference graph

Works this paper leans on

72 extracted references · 69 canonical work pages · cited by 1 Pith paper

  1. [1]

    Semantics-to-signal scal- able image compression with learned revertible representations,

    K. Liu, D. Liu, L. Li, N. Yan, and H. Li, “Semantics-to-signal scal- able image compression with learned revertible representations,” Int. J. Comput. Vision, vol. 129, no. 9, pp. 2605–2621, 2021

    work page 2021

  2. [2]

    Towards image understanding from deep compression without decoding,

    R. Torfason, F. Mentzer, E. Agustsson, M. Tschannen, R. Timofte, and L. Van Gool, “Towards image understanding from deep compression without decoding,” Proc. Int. Conf. Learn. Representations , 2018

    work page 2018

  3. [4]

    End-to-end compression towards machine vision: Network architecture design and optimization,

    S. Wang, Z. Wang, S. Wang, and Y . Ye, “End-to-end compression towards machine vision: Network architecture design and optimization,” IEEE Open J. Circuits Syst. , vol. 2, pp. 675–685, 2021

    work page 2021

  4. [5]

    Learn a compression for objection detection-vae with a bridge,

    Y . Mei, F. Li, L. Li, and Z. Li, “Learn a compression for objection detection-vae with a bridge,” in IEEE Int. Conf. Vis. Commun. Image Process. IEEE, 2021, pp. 1–5

    work page 2021

  5. [6]

    Image coding for machines: an end-to-end learned approach,

    N. Le, H. Zhang, F. Cricri, R. Ghaznavi-Youvalari, and E. Rahtu, “Image coding for machines: an end-to-end learned approach,” in IEEE Int. Conf. Acoust. Speech Signal Process. , 2021, pp. 1590–1594

    work page 2021

  6. [7]

    Transtic: Transferring transformer-based image compression from human perception to machine perception,

    Y .-H. Chen, Y .-C. Weng, C.-H. Kao, C. Chien, W.-C. Chiu, and W.- H. Peng, “Transtic: Transferring transformer-based image compression from human perception to machine perception,” in Proc. IEEE Int. Conf. Comp. Vis., 2023, pp. 23 297–23 307

    work page 2023

  7. [8]

    Scalable image coding for humans and machines,

    H. Choi and I. V . Baji ´c, “Scalable image coding for humans and machines,” IEEE Trans. Image Process., vol. 31, pp. 2739–2754, 2022

    work page 2022

  8. [9]

    Towards end-to-end image compression and analysis with transformers,

    Y . Bai, X. Yang, X. Liu, J. Jiang, Y . Wang, X. Ji, and W. Gao, “Towards end-to-end image compression and analysis with transformers,” in Proc. Conf. AAAI, vol. 36, no. 1, 2022, pp. 104–112

    work page 2022

Show all 72 references
  1. [10]

    Towards coding for human and machine vision: Scalable face image coding,

    S. Yang, Y . Hu, W. Yang, L.-Y . Duan, and J. Liu, “Towards coding for human and machine vision: Scalable face image coding,” IEEE Trans. Multimedia, pp. 1–1, 2021

    2021

  2. [11]

    Towards analysis-friendly face representation with scalable feature and texture compression,

    S. Wang, S. Wang, W. Yang, X. Zhang, S. Wang, S. Ma, and W. Gao, “Towards analysis-friendly face representation with scalable feature and texture compression,” IEEE Trans. Multimedia , 2021

    2021

  3. [12]

    Deepsvc: Deep scalable video coding for both machine and human vision,

    H. Lin, B. Chen, Z. Zhang, J. Lin, X. Wang, and T. Zhao, “Deepsvc: Deep scalable video coding for both machine and human vision,” in Proc. ACM Int. Conf. Multimedia , 2023, pp. 9205–9214

    2023

  4. [13]

    Non-semantics suppressed mask learning for unsupervised video semantic compression,

    Y . Tian, G. Lu, G. Zhai, and Z. Gao, “Non-semantics suppressed mask learning for unsupervised video semantic compression,” in Proc. IEEE Int. Conf. Comp. Vis. , 2023, pp. 13 610–13 622

    2023

  5. [14]

    Parameter-efficient fine-tuning of large- scale pre-trained language models,

    N. Ding, Y . Qin, G. Yang, F. Wei, Z. Yang, Y . Su, S. Hu, Y . Chen, C.-M. Chan, W. Chen et al. , “Parameter-efficient fine-tuning of large- scale pre-trained language models,” Natu. Mach. Inte., vol. 5, no. 3, pp. 220–235, 2023

    2023

  6. [15]

    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,” in Proc. Int. Conf. Learn. Representations, 2018

    2018

  7. [16]

    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 Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2020, pp. 7939–7948

    2020

  8. [17]

    Dvc: An end-to-end deep video compression framework,

    G. Lu, W. Ouyang, D. Xu, X. Zhang, C. Cai, and Z. Gao, “Dvc: An end-to-end deep video compression framework,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2019, pp. 11 006–11 015

    2019

  9. [18]

    Fvc: A new framework towards deep video compression in feature space,

    Z. Hu, G. Lu, and D. Xu, “Fvc: A new framework towards deep video compression in feature space,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2021, pp. 1502–1511

    2021

  10. [19]

    Imagenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale hierarchical image database,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2009, pp. 248–255

    2009

  11. [20]

    The pascal visual object classes (voc) challenge,

    M. Everingham, L. Van Gool, C. K. Williams, J. Winn, and A. Zisser- man, “The pascal visual object classes (voc) challenge,” Int. J. Comput. Vision, vol. 88, pp. 303–338, 2010

    2010

  12. [21]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Proc. Eur. Conf. Comp. Vis. Springer, 2014, pp. 740–755

    2014

  13. [22]

    Ucf101: A dataset of 101 human actions classes from videos in the wild,

    K. Soomro, A. R. Zamir, and M. Shah, “Ucf101: A dataset of 101 human actions classes from videos in the wild,”arXiv preprint arXiv:1212.0402, 2012

    2012 arXiv

  14. [24]

    Vvc test model (vtm),

    “Vvc test model (vtm),” 2022, https://jvet.hhi.fraunhofer.de/, ac- cessed:2024. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 10

    2022

  15. [25]

    Icmh-net: Neural image compres- sion towards both machine vision and human vision,

    L. Liu, Z. Hu, Z. Chen, and D. Xu, “Icmh-net: Neural image compres- sion towards both machine vision and human vision,” in Proc. ACM Int. Conf. Multimedia, 2023, pp. 8047–8056

    2023

  16. [26]

    End-to-end optimized image compression,

    J. Ball ´e, V . Laparra, and E. P. Simoncelli, “End-to-end optimized image compression,” in Proc. Int. Conf. Learn. Representations , 2017

    2017

  17. [27]

    Joint autoregressive and hierarchical priors for learned image compression,

    D. Minnen, J. Ball ´e, and G. D. Toderici, “Joint autoregressive and hierarchical priors for learned image compression,” Advances in Neural Inf. Process. Syst. , vol. 31, 2018

    2018

  18. [28]

    Neural data-dependent transform for learned image compression,

    D. Wang, W. Yang, Y . Hu, and J. Liu, “Neural data-dependent transform for learned image compression,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2022, pp. 17 379–17 388

    2022

  19. [29]

    The devil is in the details: Window- based attention for image compression,

    R. Zou, C. Song, and Z. Zhang, “The devil is in the details: Window- based attention for image compression,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2022, pp. 17 492–17 501

    2022

  20. [30]

    Unified multivariate gaussian mixture for efficient neural image compression,

    X. Zhu, J. Song, L. Gao, F. Zheng, and H. T. Shen, “Unified multivariate gaussian mixture for efficient neural image compression,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2022, pp. 17 612–17 621

    2022

  21. [31]

    Variable-rate deep image compression through spatially-adaptive feature transform,

    M. Song, J. Choi, and B. Han, “Variable-rate deep image compression through spatially-adaptive feature transform,” in Proc. IEEE Int. Conf. Comp. Vis., 2021, pp. 2380–2389

    2021

  22. [32]

    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 Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2021, pp. 4998–5007

    2021

  23. [33]

    Learning accurate entropy model with global reference for image compression,

    Y . Qian, Z. Tan, X. Sun, M. Lin, D. Li, Z. Sun, L. Hao, and R. Jin, “Learning accurate entropy model with global reference for image compression,” in Proc. Int. Conf. Learn. Representations , 2021

    2021

  24. [34]

    Learned block-based hybrid image compression,

    Y . Wu, X. Li, Z. Zhang, X. Jin, and Z. Chen, “Learned block-based hybrid image compression,” IEEE Trans. Circuits Syst. Video Technol. , vol. 32, no. 6, pp. 3978–3990, 2021

    2021

  25. [35]

    Asymmetric gained deep image compression with continuous rate adaptation,

    Z. Cui, J. Wang, S. Gao, T. Guo, Y . Feng, and B. Bai, “Asymmetric gained deep image compression with continuous rate adaptation,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2021, pp. 10 532–10 541

    2021

  26. [36]

    Content adaptive latents and decoder for neural image compression,

    G. Pan, G. Lu, Z. Hu, and D. Xu, “Content adaptive latents and decoder for neural image compression,” in Proc. Eur. Conf. Comp. Vis. Springer, 2022, pp. 556–573

    2022

  27. [37]

    Elic: Efficient learned image compression with unevenly grouped space- channel contextual 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 contextual adaptive coding,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2022, pp. 5718–5727

    2022

  28. [38]

    Cra5: Extreme com- pression of era5 for portable global climate and weather research via an efficient variational transformer,

    T. Han, Z. Chen, S. Guo, W. Xu, and L. Bai, “Cra5: Extreme com- pression of era5 for portable global climate and weather research via an efficient variational transformer,” arXiv preprint arXiv:2405.03376 , 2024

    2024 arXiv

  29. [39]

    Pchm-net: A new point cloud compression framework for both human vision and machine vision,

    L. Liu, Z. Hu, and J. Zhang, “Pchm-net: A new point cloud compression framework for both human vision and machine vision,” Proc. IEEE Int. Conf. Multimedia Expo , 2023

    2023

  30. [40]

    M-lvc: Multiple frames prediction for learned video compression,

    J. Lin, D. Liu, H. Li, and F. Wu, “M-lvc: Multiple frames prediction for learned video compression,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2020, pp. 3546–3554

    2020

  31. [41]

    Scale-space flow for end-to-end optimized video compres- sion,

    E. Agustsson, D. Minnen, N. Johnston, J. Balle, S. J. Hwang, and G. Toderici, “Scale-space flow for end-to-end optimized video compres- sion,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2020, pp. 8503– 8512

    2020

  32. [42]

    Neural inter-frame compression for video coding,

    A. Djelouah, J. Campos, S. Schaub-Meyer, and C. Schroers, “Neural inter-frame compression for video coding,” in Proc. IEEE Int. Conf. Comp. Vis., 2019, pp. 6421–6429

    2019

  33. [43]

    Video compression with rate-distortion autoencoders,

    A. Habibian, T. v. Rozendaal, J. M. Tomczak, and T. S. Cohen, “Video compression with rate-distortion autoencoders,” in Proc. IEEE Int. Conf. Comp. Vis., 2019, pp. 7033–7042

    2019

  34. [44]

    Improving deep video compression by resolution-adaptive flow coding,

    Z. Hu, Z. Chen, D. Xu, G. Lu, W. Ouyang, and S. Gu, “Improving deep video compression by resolution-adaptive flow coding,” in Proc. Eur. Conf. Comp. Vis. Springer, 2020, pp. 193–209

    2020

  35. [45]

    Exploiting intra-slice and inter-slice redundancy for learning-based lossless volumetric image compression,

    Z. Chen, S. Gu, G. Lu, and D. Xu, “Exploiting intra-slice and inter-slice redundancy for learning-based lossless volumetric image compression,” IEEE Trans. Image Process. , vol. 31, pp. 1697–1707, 2022

    2022

  36. [46]

    Coarse-to-fine deep video coding with hyperprior-guided mode prediction,

    Z. Hu, G. Lu, J. Guo, S. Liu, W. Jiang, and D. Xu, “Coarse-to-fine deep video coding with hyperprior-guided mode prediction,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2022, pp. 5921–5930

    2022

  37. [47]

    Content adaptive and error propagation aware deep video compression,

    G. Lu, C. Cai, X. Zhang, L. Chen, W. Ouyang, D. Xu, and Z. Gao, “Content adaptive and error propagation aware deep video compression,” in Proc. Eur. Conf. Comp. Vis. Springer, 2020, pp. 456–472

    2020

  38. [48]

    An end- to-end learning framework for video compression,

    G. Lu, X. Zhang, W. Ouyang, L. Chen, Z. Gao, and D. Xu, “An end- to-end learning framework for video compression,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 43, no. 10, pp. 3292–3308, 2020

    2020

  39. [49]

    Learning for video compression with hierarchical quality and recurrent enhancement,

    R. Yang, F. Mentzer, L. V . Gool, and R. Timofte, “Learning for video compression with hierarchical quality and recurrent enhancement,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2020, pp. 6628–6637

    2020

  40. [50]

    Improving facial attribute recognition by group and graph learning,

    Z. Chen, S. Gu, F. Zhu, J. Xu, and R. Zhao, “Improving facial attribute recognition by group and graph learning,” in Proc. IEEE Int. Conf. Multimedia Expo. IEEE, 2021, pp. 1–6

    2021

  41. [51]

    Towards point cloud compression for machine perception: A simple and strong baseline by learning the octree depth level predictor,

    L. Liu, Z. Hu, and Z. Chen, “Towards point cloud compression for machine perception: A simple and strong baseline by learning the octree depth level predictor,” arXiv preprint arXiv:2406.00791 , 2024

    2024 arXiv

  42. [52]

    Deep contextual video compression,

    J. Li, B. Li, and Y . Lu, “Deep contextual video compression,” Advances in Neural Inf. Process. Syst. , vol. 34, pp. 18 114–18 125, 2021

    2021

  43. [53]

    Lsvc: A learning- based stereo video compression framework. in 2022 ieee,

    Z. Chen, G. Lu, Z. Hu, S. Liu, W. Jiang, and D. Xu, “Lsvc: A learning- based stereo video compression framework. in 2022 ieee,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2022, pp. 6063–6072

    2022

  44. [54]

    Temporal context mining for learned video compression,

    X. Sheng, J. Li, B. Li, L. Li, D. Liu, and Y . Lu, “Temporal context mining for learned video compression,” IEEE Trans. Multimedia, 2022

    2022

  45. [55]

    Hybrid spatial-temporal entropy modelling for neural video compression,

    J. Li, B. Li, and Y . Lu, “Hybrid spatial-temporal entropy modelling for neural video compression,” in Proc. ACM Int. Conf. Multimedia , 2022, pp. 1503–1511

    2022

  46. [56]

    Neural video compression with spatio-temporal cross- covariance transformers,

    Z. Chen, L. Relic, R. Azevedo, Y . Zhang, M. Gross, D. Xu, L. Zhou, and C. Schroers, “Neural video compression with spatio-temporal cross- covariance transformers,” in Proc. ACM Int. Conf. Multimedia, 2023, pp. 8543–8551

    2023

  47. [57]

    Group-aware parameter-efficient updating for content-adaptive neural video compression,

    Z. Chen, L. Zhou, Z. Hu, and D. Xu, “Group-aware parameter-efficient updating for content-adaptive neural video compression,” in Proceedings of the 32nd ACM International Conference on Multimedia , 2024, pp. 11 022–11 031

    2024

  48. [58]

    Video coding for machines: A paradigm of collaborative compression and intelligent analytics,

    L. Duan, J. Liu, W. Yang, T. Huang, and W. Gao, “Video coding for machines: A paradigm of collaborative compression and intelligent analytics,” IEEE Trans. Image Process. , vol. 29, pp. 8680–8695, 2020

    2020

  49. [59]

    AdapterHub: A framework for adapting transformers,

    J. Pfeiffer, A. R ¨uckl´e, C. Poth, A. Kamath, I. Vuli ´c, S. Ruder, K. Cho, and I. Gurevych, “AdapterHub: A framework for adapting transformers,” in Proc. Conf. Empirical Methods in Natural Language Processing, Oct. 2020, pp. 46–54

    2020

  50. [60]

    Towards a unified view of parameter-efficient transfer learning,

    J. He, C. Zhou, X. Ma, T. Berg-Kirkpatrick, and G. Neubig, “Towards a unified view of parameter-efficient transfer learning,” in Proc. Int. Conf. Learn. Representations, 2022

    2022

  51. [61]

    Counter-interference adapter for multilingual machine translation,

    Y . Zhu, J. Feng, C. Zhao, M. Wang, and L. Li, “Counter-interference adapter for multilingual machine translation,” in Proc. Conf. Empirical Methods in Natural Language Processing , 2021, pp. 2812–2823

    2021

  52. [62]

    SparseAdapter: An easy approach for improving the parameter-efficiency of adapters,

    S. He, L. Ding, D. Dong, J. Zhang, and D. Tao, “SparseAdapter: An easy approach for improving the parameter-efficiency of adapters,” in Proc. Conf. Empirical Methods in Natural Language Processing , Dec. 2022, pp. 2184–2190

    2022

  53. [63]

    Compacter: Efficient low-rank hypercomplex adapter layers,

    R. Karimi Mahabadi, J. Henderson, and S. Ruder, “Compacter: Efficient low-rank hypercomplex adapter layers,”Advances in Neural Inf. Process. Syst., vol. 34, pp. 1022–1035, 2021

    2021

  54. [64]

    Krona: Parameter efficient tuning with kronecker adapter,

    A. Edalati, M. Tahaei, I. Kobyzev, V . P. Nia, J. J. Clark, and M. Reza- gholizadeh, “Krona: Parameter efficient tuning with kronecker adapter,” Advances in Neural Inf. Process. Syst. , 2022

    2022

  55. [65]

    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,” Advances in Neural Inf. Process. Syst. , vol. 35, pp. 13 146–13 157, 2022

    2022

  56. [66]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn. , 2016, pp. 770–778

    2016

  57. [67]

    Psanet: Point-wise spatial attention network for scene parsing,

    H. Zhao, Y . Zhang, S. Liu, J. Shi, C. C. Loy, D. Lin, and J. Jia, “Psanet: Point-wise spatial attention network for scene parsing,” in Proc. Eur. Conf. Comp. Vis., 2018, pp. 267–283

    2018

  58. [68]

    Faster r-cnn: Towards real-time object detection with region proposal networks,

    S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object detection with region proposal networks,” Advances in Neural Inf. Process. Syst., vol. 28, 2015

    2015

  59. [69]

    Categorical reparametrization with gumble-softmax,

    E. Jang, S. Gu, and B. Poole, “Categorical reparametrization with gumble-softmax,” in Proc. Int. Conf. Learn. Representations , 2017

    2017

  60. [70]

    Temporal segment networks: Towards good practices for deep action recognition,

    L. Wang, Y . Xiong, Z. Wang, Y . Qiao, D. Lin, X. Tang, and L. Van Gool, “Temporal segment networks: Towards good practices for deep action recognition,” in Proc. Eur. Conf. Comp. Vis. Springer, 2016, pp. 20–36

    2016

  61. [71]

    Xmem: Long-term video object segmentation with an atkinson-shiffrin memory model,

    H. K. Cheng and A. G. Schwing, “Xmem: Long-term video object segmentation with an atkinson-shiffrin memory model,” in Proc. Eur. Conf. Comp. Vis. Springer, 2022, pp. 640–658

    2022

  62. [72]

    A benchmark dataset and evaluation methodology for video object segmentation,

    F. Perazzi, J. Pont-Tuset, B. McWilliams, L. Van Gool, M. Gross, and A. Sorkine-Hornung, “A benchmark dataset and evaluation methodology for video object segmentation,” in Proc. IEEE Conf. Comp. Vis. Patt. Recogn., 2016, pp. 724–732

    2016

  63. [73]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in Proc. Int. Conf. Learn. Representations , 2015

    2015

  64. [74]

    Bpg image format,

    B. Fabrice, “Bpg image format,” 2018, http://bellard.org/bpg/, ac- cessed:2022

    2018

Pith tools

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