Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

BiDense: Binarization for Dense Prediction

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read BiDense claims that a binary neural network with adaptive binarization and full-precision bypasses can match full-precision accuracy on semantic segmentation and monocular depth estimation while using a small fraction of the memory and…

desk verdict Useful BNN module for dense prediction, but the 'comparable to full-precision' claim rests on a random-init baseline and one implausible table entry. read the letter →

arxiv 2411.10346 v2 pith:MKXBZ3TM submitted 2024-11-15 cs.CV

Rui Yin , Haotong Qin , Yulun Zhang , Wenbo Li , Yong Guo , Jianjun Zhu , Cheng Wang , Biao Jia This is my paper · ORCID
classification cs.CV
keywords binaryneuralnetworksdensepredictionsemanticsegmentationmonoculardepthestimationmodelbinarizationinformationentropyfull-precisionbypassadaptive
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

BiDense aims to show that binary neural networks, which store weights and activations as single signs, can be made accurate enough for dense prediction — semantic segmentation and monocular depth estimation — where every pixel matters and earlier binarization methods lost too much detail. Its strategy is to keep real-valued information flowing through the network in two places: the binarizer itself, which adaptively sets thresholds and scales from each input's mean and mean absolute deviation, and a full-precision bypass that routes real-valued features around binary convolutions when the channel count changes. Across ADE20K, PASCAL VOC, NYUv2, and KITTI, BiDense outperforms previous binary networks and lands close to its full-precision ConvNeXt-UPerNet baseline, and on NYUv2 it reports even better depth metrics than that baseline. The paper also positions itself as the first binarization method applied to monocular depth estimation. If the results hold, pixel-level prediction on memory- and power-limited devices becomes a realistic target for 1-bit networks.

What carries the argument

The load-bearing machinery is a redesigned binary convolutional layer with two additions. The Distribution-adaptive Binarizer (DAB) sets the binarization threshold as $\beta(x)=k\bar{x}+b$, with learnable $k$ and $b$ and $\bar{x}$ the input mean, then assigns a per-input scale $\alpha(x_s)=\exp(a(\|x_s\|_{\ell^1}/n_{x_s}-1))$, where $x_s=x-\beta(x)$, blending a learnable global baseline with the current input's mean and mean absolute deviation. The Channel-adaptive Full-precision Bypass (CFB) implements fusion-down, 1D adaptive average pooling along the channel dimension with kernel and stride equal to avoid overlap, and fusion-up, channel repetition plus pooling of the remainder, so a full-precision copy of any layer input can be reshaped to the binary output's channel count and added to it. Together they let the binary convolution learn residuals on top of a real-valued signal, while the entropy argument from earlier binarization work explains why the adaptive threshold preserves more information.

What would settle it

Run BiDense on a mobile or edge GPU and measure end-to-end latency, peak memory, and energy against the FP32 baseline and a BiDense variant with the CFB removed. If the full-precision bypasses make runtime or memory approach the FP32 model, the paper's core efficiency claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that the accuracy gap in binary dense prediction comes mostly from information discarded at two well-defined points, and that both can be patched cheaply. First, fixed binarization thresholds and scales, once trained, misfire on inputs whose distributions shift; the Distribution-adaptive Binarizer (DAB) replaces them with input-dependent values, $\beta(x)=k\bar{x}+b$ and $\alpha(x_s)=\exp(a(\|x_s\|_{\ell^1}/n_{x_s}-1))$, so that binarization keeps the entropy of the 1-bit signal near its maximum. Second, encoder-decoder architectures change channel counts constantly, which blocks full-precision skip connections; the Channel-adaptive Full-precision Bypass (CFB)'s fusion-down and fusion-up resize a full-precision copy of the input to match the binary convolution's output shape, letting the binary path learn residuals while the real-valued path carries fine detail. With both mechanisms, the paper reports 67.25% pixel accuracy and 18.75% mIoU on ADE20K, close to the 70.81% and 20.39% of its randomly initialized FP32 baseline, and 75.2% mIoU on PASCAL VOC against 84.6% for FP32, at roughly 4% of the parameters and 2% of the operations. On NYUv2 monocular depth, it reports better error and accuracy numbers than the same FP32 baseline on every metric. The authors take these results as evidence that 1-bit networks can serve generalized dense prediction rather than only classification.

Load-bearing premise

The efficiency claim assumes that carrying a full-precision copy of every activation through every binary convolution costs almost nothing in memory and runtime, but the paper reports no latency, energy, or peak-memory measurements.

Editorial extensions

If this is right

  • Binary dense prediction reaches within a few accuracy points of full precision on the reported benchmarks, suggesting 1-bit networks are a practical option for on-device segmentation and depth estimation.
  • DAB's input-dependent thresholds should keep working on unseen image distributions, since the threshold and scale track the statistics of whatever input arrives.
  • CFB applies to arbitrary channel-size changes, so the method transfers to any encoder-decoder or transformer-CNN hybrid beyond the ConvNeXt-UPerNet and DPT architectures tested.
  • The reported gains on a regression task (depth) as well as classification (segmentation) indicate that binarization need not be limited to label prediction.

Reading between the lines

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

  • The entropy-maximization story suggests DAB would transfer to 2-bit and 4-bit quantization, where threshold placement also controls information retention; the paper does not test this.
  • Since CFB runs a full-precision copy alongside every binary convolution, the real-world speedup depends on how cheaply that path can be fused into a hardware pipeline; an actual latency and energy profile would settle whether the reported operation counts translate into wall-clock savings.
  • A natural next experiment is to apply BiDense to other dense tasks like keypoint detection or optical flow, where fine-grained localization should magnify the benefit of retaining full-precision detail.
  • The 'comparable to full precision' phrasing leans on the randomly initialized FP32 baseline; against the ImageNet-pretrained baseline on ADE20K the gap is larger, so the strength of the headline claim depends on which baseline is meant.
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 BiDense, a binary neural network (BNN) framework for dense prediction tasks, specifically semantic segmentation and monocular depth estimation. BiDense introduces two components: a Distribution-adaptive Binarizer (DAB), which computes binarization thresholds and scaling factors from the mean and mean absolute deviation of the input activations, and a Channel-adaptive Full-precision Bypass (CFB), which provides full-precision skip paths when a binary convolution changes the number of channels. The authors evaluate BiDense against several prior BNN methods on ADE20K, PASCAL VOC, NYUv2, and KITTI, and report that it outperforms existing BNNs and approaches the accuracy of full-precision models at a fraction of the parameters and operations. The paper also includes ablations, an extension to a DPT-style hybrid architecture, and visualizations of information entropy.

Significance. If the central claim were fully supported, BiDense would be the first binary network to closely match full-precision accuracy on both semantic segmentation and monocular depth estimation, which would be a meaningful advance for efficient dense prediction. The two proposed components are simple and potentially generalizable, and the ablation study (Table 4) shows that the CFB contributes a large accuracy gain (mIoU from 9.78 to 18.75 on ADE20K). The paper also provides a public code link and reports comparisons against multiple BNN baselines in a unified experimental setup. However, the headline claim of 'comparable to full-precision' is only established against a randomly initialized full-precision baseline, not the standard ImageNet-pretrained baseline, and the OPs definition in Section 4 is ambiguous. These issues materially weaken the significance as currently stated; the contribution is better characterized as a strong BNN for dense prediction trained from random initialization, with the full-precision comparison requiring substantial qualification.

major comments (4)
  1. [Abstract, §4.1, Tables 1–3] The central claim that BiDense achieves 'performance levels comparable to full-precision models' is supported only by comparison with the randomly initialized FP32 baseline (ConvNeXt), not the ImageNet-pretrained baseline (ConvNeXt*) that the paper itself lists. On ADE20K, the gap to ConvNeXt* is 22.21 mIoU (18.75 vs 40.96), whereas the gap to the random ConvNeXt is only 1.64 mIoU (18.75 vs 20.39). Similarly, on NYUv2 AbsRel BiDense is 0.189 vs 0.117 for ConvNeXt* and 0.201 for random ConvNeXt, and on KITTI AbsRel it is 0.095 vs 0.065 for ConvNeXt*. Since dense prediction models in practice are initialized from ImageNet, the 'comparable to full-precision' statement in the abstract and the text in §4.1 and §4.2 is unsupported. The authors should either compare against the pretrained baseline for all methods and report those numbers prominently, or explicitly narrow the claim to 'models trained from random initialization.'
  2. [Table 2] The ConvNeXt* row on PASCAL VOC reports pixAcc 98.58 and mIoU 99.72. An mIoU of 99.72 for ConvNeXt-Tiny with UPerNet on PASCAL VOC is implausible, and the mIoU being higher than pixAcc is anomalous for a 21-class segmentation benchmark. This appears to be a data error or a typo, but it is used in the running text to assert that BiDense 'approaches the FP32 baseline.' The row must be corrected or removed before any claims about VOC are made; as printed, it undermines the credibility of the comparison.
  3. [§4, OPs definition] The computational-cost formula at the start of §4 is internally inconsistent with the reported numbers. If OPs_f denotes the full-precision model's total operations, then OPs = OPs_b + OPs_f with OPs_b = OPs_f/64 would give roughly 1.016×OPs_f for any BNN, contradicting Table 1, where BiDense is listed at 5.37 G vs 288.37 G for the FP32 model. The formula only makes sense if OPs_f is reinterpreted as the remaining full-precision operations inside the binary network, but that interpretation is never stated. Furthermore, the 'significantly reducing computational costs' claim rests entirely on this FLOPs proxy: no latency, energy, or memory measurements are provided, and the CFB bypass carries full-precision activations through the network, whose memory footprint is not accounted for. Please clarify the definition and, ideally, add at least one hardware measurement (latency or energy) to support the efficiency claim.
  4. [§3.3, Eqs. (12)–(13)] The fusion-up procedure is not fully specified. Eq. (13) defines K = floor(Cin / (Cout mod Cin)), which divides by zero when Cout is an exact multiple of Cin — a common case in dense prediction architectures. The text also does not specify how the 'remainder' channels are merged when the remainder is zero, nor how the pooled output is concatenated with the repeated channels in that case. Please provide the complete algorithm, including the zero-remainder case, so that the implementation is reproducible without recourse to the code repository.
minor comments (5)
  1. [§3.2, Eq. (10)] The approximation 'lim_{x→1} exp(a(x−1)) ≈ ax' is inaccurate; the correct first-order approximation near x=1 is exp(a(x−1)) ≈ 1 + a(x−1), not ax. This does not affect the experiments but should be corrected.
  2. [§4.1] The sentence 'only 3.56% and 1.64% reductions of pixAcc and mIoU' compares BiDense to ConvNeXt, the random baseline, without saying so. Please state explicitly which baseline is being used in each comparison, since the pretrained ConvNeXt* has much larger gaps.
  3. [Supplementary, §A and §C] The supplementary's limitation statement that DAB is not suited for transformer/self-attention modules is important and should be in the main text, because it materially qualifies the claim of a 'generalized' binary framework for dense prediction.
  4. [References] Reference [3] is cited as BiSRNet in the text, but the listed title is 'Binarized spectral compressive imaging,' which does not match a semantic segmentation method called BiSRNet. Please verify that the correct reference is cited.
  5. [Table 4, item 7] Setting α(xs) = ||xs||_1 / n_xs yields a dramatic drop to 4.97 mIoU, which suggests training instability. It would be helpful to report the training configuration for this variant (same learning rate, epochs, seeds) to confirm the comparison is not affected by a single unlucky run.

Circularity Check

1 steps flagged · score 2.0 of 10

Central DAB/CFB claims are externally benchmarked; only the entropy diagnostic is self-confirming by construction.

  1. other [Sec. 3.2, Eqs. (4)-(5); Fig. 3; Fig. 6d]
    "As indicated in Fig. 3 by information entropy, adaptively determining thresholds effectively enhances information richness after binarization in BiDense."

    DAB's threshold is defined as β(x)=k·x̄+b (Eq. 5), i.e., it is built to place the binarization cut near the center of each input's distribution. The paper's information-richness metric is H(Bx)=−p ln p−(1−p)ln(1−p) (Eq. 2), which is maximized when p approaches 0.5, i.e., when the threshold is at the distribution center. Therefore the higher entropy reported for BiDense in Fig. 3 and Fig. 6d follows from the construction of the binarizer itself; it is a restatement of the design objective rather than an independent measurement that DAB 'preserves more real-valued information.' This diagnostic does not predetermine the dense-prediction accuracy numbers, which come from external benchmarks, so the circularity is minor and non-central.

full rationale

BiDense's load-bearing derivation chain is not circular. DAB is defined in Eqs. (4)-(10) as a mean/MAD-based adaptive threshold and scaling rule, with no dense-prediction accuracy term appearing in its construction. CFB is a deterministic channel-resizing bypass (fusion-down/fusion-up) whose operations are specified independently of the target metrics. The headline accuracy and efficiency claims are supported by comparisons on ADE20K, PASCAL VOC, NYUv2, and KITTI against published BNN baselines (BNN, ReActNet, AdaBin, BiSRNet) and two FP32 baselines; those results are external to the method's construction. Self-citations, e.g., IR-Net [39] for mean thresholds and BiSRNet [3] for channel-mismatch bypasses, point to published and independently benchmarked prior work and are used as design inspirations, not as forbidden uniqueness premises. The one genuinely self-confirming element is the information-entropy evidence: since Eq. (5) sets the threshold from the input mean and Eq. (2) defines entropy as maximized at a balanced split, the higher H(Bx) shown for BiDense in Figs. 3 and 6d is a mathematical consequence of the binarizer's design, not an independent confirmation that more real-valued information is preserved. This diagnostic does not force the dense-prediction results, so the overall circularity score is low. Separately, the choice of randomly initialized ConvNeXt as the main FP32 comparison point is a benchmarking and claim-scoping concern, not a circularity issue.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim depends on a few heuristic assumptions about binarization and on the experimental baseline choice. The learnable parameters k, b, and a are fitted to data, and several domain assumptions (entropy-performance link, mean-as-median, CFB channel operations preserving information) are not proven. No invented entities, particles, or forces are introduced.

free parameters (3)
  • k
    Layer-wise learnable coefficient in the adaptive threshold beta(x) = k * mean(x) + b. Fitted on the training data; values are shown as distributions in Fig. 6 but not reported numerically.
  • b
    Layer-wise learnable bias in the adaptive threshold. Fitted on the training data; values are shown as distributions in Fig. 6.
  • a
    Layer-wise learnable exponent controlling the adaptive scaling factor alpha(xs) = exp(a * (||xs||_1 / n - 1)). Fitted on the training data; values are shown as distributions in Fig. 6.
assumptions (4)
  • domain assumption Maximizing information entropy of binarized activations improves task performance
    Used to justify DAB's threshold choice in Sections 3.1 and 3.2; not proven, and the paper only shows a correlation via entropy plots.
  • domain assumption The mean of activations approximates the median, so using the mean as threshold approaches entropy-optimal binarization
    Section 3.1, drawing on IR-Net [39]; activations are not guaranteed symmetric, so the mean may be a poor median proxy.
  • domain assumption Channel-wise average pooling and repetition in CFB preserve the information needed for dense prediction
    Section 3.3; no analysis of information loss from these operations is provided.
  • ad hoc to paper Randomly initialized full-precision models trained for 50 epochs are an appropriate 'full-precision baseline' for the comparable-accuracy claim
    Section 4.1 and Table 1 pick ConvNeXt (random init) for the headline comparison, not the ImageNet-pretrained ConvNeXt*, which is substantially stronger.

how reviews work

0 comments
Cite this review

Pith. "Pith review of BiDense: Binarization for Dense Prediction." pith.science (2026). https://pith.science/paper/MKXBZ3TM

@misc{pith2026241110346,
  author       = {Pith},
  title        = {Pith review of: BiDense: Binarization for Dense Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MKXBZ3TM}},
  note         = {Machine review of arXiv:2411.10346}
}
read the original abstract

Dense prediction is a critical task in computer vision. However, previous methods often require extensive computational resources, which hinders their real-world application. In this paper, we propose BiDense, a generalized binary neural network (BNN) designed for efficient and accurate dense prediction tasks. BiDense incorporates two key techniques: the Distribution-adaptive Binarizer (DAB) and the Channel-adaptive Full-precision Bypass (CFB). The DAB adaptively calculates thresholds and scaling factors for binarization, effectively retaining more information within BNNs. Meanwhile, the CFB facilitates full-precision bypassing for binary convolutional layers undergoing various channel size transformations, which enhances the propagation of real-valued signals and minimizes information loss. By leveraging these techniques, BiDense preserves more real-valued information, enabling more accurate and detailed dense predictions in BNNs. Extensive experiments demonstrate that our framework achieves performance levels comparable to full-precision models while significantly reducing memory usage and computational costs.

Figures

Figures reproduced from arXiv: 2411.10346 by the authors.

Figure 1
Figure 1. BiDense approaches the performance of full-precision 32-bit networks in dense prediction tasks while significantly re￾ducing memory and computational costs. methods have been proposed [20, 21, 28, 29, 37, 47, 51, 54, 63]. Among these, model quantization [9, 23] stands out for its ability to compress the weights and activations of networks into low-bit representations. As the most extreme form, model binarization res… view at source ↗
Figure 2
Figure 2. Overview of a BiDense convolutional layer. The BiDense layer integrates two techniques: 1) The Distribution-adaptive Binarizer (DAB) adaptively binarizes activations based on input distributions, allowing it to retain more real-valued information. 2) The Channel-Adaptive Full-precision Bypass (CFB) employs fusion-down and fusion-up to dynamically adjust the channel size of full￾precision features, ensuring they alig… view at source ↗
Figure 3
Figure 3. Average channel-wise information entropy and fea￾ture visualization of activations before and after the first binariza￾tion in networks on the ADE20K [67] validation set. The feature visualization is represented by averaging the activations. As indi￾cated by information entropy, the fixed thresholds in BNN and Re￾ActNet can result in suboptimal binarization and information loss, whereas BiDense retains more informat… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Qualitative Results of semantic segmentation on the ADE20K [67] validation set. BiDense preserves full-precision semantic information, yielding more fine-grained and accurate segmentation than previous BNNs. Method Bit Topology pixAcc (%) mIoU (%) ConvNeXt* [33] 32 Con…
Figure 5
Figure 5. Figure 5: Qualitative results of monocular depth estimation results on the NYUv2 [48] and KITTI [16] depth datasets. BiDense retains real-valued information, resulting in more accurate performance and finer-grained depth predictions compared to previous BNNs. Method Bit NYUv2 [4…
Figure 6
Figure 6. Figure 6: Statistics of DAB on the ADE20K [67] validation set. 1) Panels (a), (b), and (c) illustrate the statics of three learnable parameters, k, b, and a, from three different convolutional modules across the 18 ConvNeXt blocks, ranging from shallow to deep layers. The colors…
Figure 7
Figure 7. Figure 7: Qualitative semantic segmentation results on the PAS [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Qualitative results of monocular depth estimation results on the NYUv2 [ [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Qualitative Results of semantic segmentation on the ADE20K [ [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Information-Bottleneck Driven Binary Neural Network for Change Detection

    cs.CV 2025-07 conditional novelty 6.0 of 10

    BiCD is a 1-bit change detection network whose auxiliary IB-style losses improve F1 by about 1 to 3 points over other binary networks, with no extra inference cost.

  2. High-Fidelity Differential-information Driven Binary Vision Transformer

    cs.CV 2025-07 conditional novelty 6.0 of 10

    DIDB-ViT combines differential attention, Haar-wavelet frequency decomposition, and token-wise activation shifts to improve binary vision transformers, achieving state-of-the-art results on several benchmarks.

Reference graph

Works this paper leans on

70 extracted references · 51 canonical work pages · cited by 2 Pith papers

  1. [1]

    Beit: Bert pre-training of image transformers

    Hangbo Bao, Li Dong, Songhao Piao, and Furu Wei. Beit: Bert pre-training of image transformers. arXiv preprint arXiv:2106.08254, 2021. 2

    arXiv 2021

  2. [2]

    Zoedepth: Zero-shot trans- fer by combining relative and metric depth

    Shariq Farooq Bhat, Reiner Birkl, Diana Wofk, Peter Wonka, and Matthias M ¨uller. Zoedepth: Zero-shot trans- fer by combining relative and metric depth. arXiv preprint arXiv:2302.12288, 2023. 2

    arXiv 2023

  3. [3]

    Binarized spectral compressive imaging

    Yuanhao Cai, Yuxin Zheng, Jing Lin, Xin Yuan, Yulun Zhang, and Haoqian Wang. Binarized spectral compressive imaging. In NeurIPS, 2024. 1, 2, 5, 6, 7, 12, 13, 14

    work page 2024

  4. [4]

    Rethinking atrous convolution for semantic image segmentation

    Liang-Chieh Chen. Rethinking atrous convolution for semantic image segmentation. arXiv preprint arXiv:1706.05587, 2017. 2

    arXiv 2017

  5. [5]

    Encoder-decoder with atrous separable convolution for semantic image segmentation

    Liang-Chieh Chen, Yukun Zhu, George Papandreou, Florian Schroff, and Hartwig Adam. Encoder-decoder with atrous separable convolution for semantic image segmentation. In ECCV, 2018. 2

    2018

  6. [6]

    Vision transformer adapter for dense predictions

    Zhe Chen, Yuchen Duan, Wenhai Wang, Junjun He, Tong Lu, Jifeng Dai, and Yu Qiao. Vision transformer adapter for dense predictions. arXiv preprint arXiv:2205.08534, 2022. 2, 12

    arXiv 2022

  7. [7]

    Per- pixel classification is not all you need for semantic segmen- tation

    Bowen Cheng, Alex Schwing, and Alexander Kirillov. Per- pixel classification is not all you need for semantic segmen- tation. In NeurIPS, 2021. 2

    work page 2021

  8. [8]

    Masked-attention mask transformer for universal image segmentation

    Bowen Cheng, Ishan Misra, Alexander G Schwing, Alexan- der Kirillov, and Rohit Girdhar. Masked-attention mask transformer for universal image segmentation. In CVPR,

Show all 70 references
  1. [9]

    Binaryconnect: Training deep neural networks with binary weights during propagations

    Matthieu Courbariaux, Yoshua Bengio, and Jean-Pierre David. Binaryconnect: Training deep neural networks with binary weights during propagations. In NeurIPS, 2015. 1, 2, 3

    2015

  2. [10]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In CVPR, 2009. 5

    2009

  3. [11]

    Superpoint: Self-supervised interest point detection and description

    Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabi- novich. Superpoint: Self-supervised interest point detection and description. In CVPRW, 2018. 2

    2018

  4. [12]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020. 2, 12

    2010 arXiv

  5. [13]

    Depth map prediction from a single image using a multi-scale deep net- work

    David Eigen, Christian Puhrsch, and Rob Fergus. Depth map prediction from a single image using a multi-scale deep net- work. In NeurIPS, 2014. 6

    2014

  6. [14]

    The pascal visual object classes (voc) challenge

    Mark Everingham, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. The pascal visual object classes (voc) challenge. In ICCV, 2010. 5, 6, 12

    2010

  7. [15]

    Dual attention network for scene segmentation

    Jun Fu, Jing Liu, Haijie Tian, Yong Li, Yongjun Bao, Zhiwei Fang, and Hanqing Lu. Dual attention network for scene segmentation. In CVPR, 2019. 2, 5

    2019

  8. [16]

    Are we ready for autonomous driving? the kitti vision benchmark suite

    Andreas Geiger, Philip Lenz, and Raquel Urtasun. Are we ready for autonomous driving? the kitti vision benchmark suite. In CVPR, 2012. 6, 7, 13

    2012

  9. [17]

    Silk: Simple learned keypoints

    Pierre Gleize, Weiyao Wang, and Matt Feiszli. Silk: Simple learned keypoints. In ICCV, 2023. 2

    2023

  10. [18]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR,

  11. [19]

    Bivit: Extremely com- pressed binary vision transformers

    Yefei He, Zhenyu Lou, Luoming Zhang, Jing Liu, Weijia Wu, Hong Zhou, and Bohan Zhuang. Bivit: Extremely com- pressed binary vision transformers. In ICCV, 2023. 1, 2, 3, 12

    2023

  12. [20]

    Distilling the knowledge in a neural network

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network. In NeurIPSW, 2015. 1

    2015

  13. [21]

    Mobilenets: Efficient convolu- tional neural networks for mobile vision applications

    Andrew G Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco An- dreetto, and Hartwig Adam. Mobilenets: Efficient convolu- tional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861, 2017. 1

    2017 arXiv

  14. [22]

    Densely connected convolutional net- works

    Gao Huang, Zhuang Liu, Laurens Van Der Maaten, and Kil- ian Q Weinberger. Densely connected convolutional net- works. In CVPR, 2017. 2

    2017

  15. [23]

    Binarized neural networks

    Itay Hubara, Matthieu Courbariaux, Daniel Soudry, Ran El- Yaniv, and Yoshua Bengio. Binarized neural networks. In NeurIPS, 2016. 1, 2, 3, 4, 5, 6, 7, 12, 13, 14

    2016

  16. [24]

    Guiding monocular depth estimation using depth-attention volume

    Lam Huynh, Phong Nguyen-Ha, Jiri Matas, Esa Rahtu, and Janne Heikkil¨a. Guiding monocular depth estimation using depth-attention volume. In ECCV, 2020. 2

    2020

  17. [25]

    Oneformer: One transformer to rule universal image segmentation

    Jitesh Jain, Jiachen Li, Mang Tik Chiu, Ali Hassani, Nikita Orlov, and Humphrey Shi. Oneformer: One transformer to rule universal image segmentation. In CVPR, 2023. 2

    2023

  18. [26]

    Training binary neural network without batch normalization for image super-resolution

    Xinrui Jiang, Nannan Wang, Jingwei Xin, Keyu Li, Xi Yang, and Xinbo Gao. Training binary neural network without batch normalization for image super-resolution. In AAAI,

  19. [27]

    From big to small: Multi-scale local planar guidance for monocular depth estimation

    Jin Han Lee, Myung-Kyu Han, Dong Wook Ko, and Il Hong Suh. From big to small: Multi-scale local planar guidance for monocular depth estimation. arXiv preprint arXiv:1907.10326, 2019. 2, 6

    1907 arXiv

  20. [28]

    Pruning filters for efficient convnets

    Hao Li, Asim Kadav, Igor Durdanovic, Hanan Samet, and Hans Peter Graf. Pruning filters for efficient convnets. In ICLR, 2017. 1

    2017

  21. [29]

    Bi-real net: Enhancing the perfor- mance of 1-bit cnns with improved representational capabil- ity and advanced training algorithm

    Zechun Liu, Baoyuan Wu, Wenhan Luo, Xin Yang, Wei Liu, and Kwang-Ting Cheng. Bi-real net: Enhancing the perfor- mance of 1-bit cnns with improved representational capabil- ity and advanced training algorithm. In ECCV, 2018. 1

    2018

  22. [30]

    Reactnet: Towards precise binary neural net- work with generalized activation functions

    Zechun Liu, Zhiqiang Shen, Marios Savvides, and Kwang- Ting Cheng. Reactnet: Towards precise binary neural net- work with generalized activation functions. In ECCV, 2020. 1, 2, 3, 4, 5, 6, 7, 12, 13, 14

    2020

  23. [31]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In ICCV, 2021. 2

    2021

  24. [32]

    Swin transformer v2: Scaling up capacity and resolution

    Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, et al. Swin transformer v2: Scaling up capacity and resolution. In CVPR, 2022. 2

    2022

  25. [33]

    A convnet for the 2020s

    Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feicht- enhofer, Trevor Darrell, and Saining Xie. A convnet for the 2020s. In CVPR, 2022. 1, 2, 5, 6, 7, 12, 13, 14 9

    2022

  26. [34]

    Bit: Robustly binarized multi-distilled transformer

    Zechun Liu, Barlas Oguz, Aasish Pappu, Lin Xiao, Scott Yih, Meng Li, Raghuraman Krishnamoorthi, and Yashar Mehdad. Bit: Robustly binarized multi-distilled transformer. In NeurIPS, 2022. 1, 2, 3, 8, 12

    2022

  27. [35]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. In CVPR, 2015. 2

    2015

  28. [36]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In ICLR, 2018. 5

    2018

  29. [37]

    Shufflenet v2: Practical guidelines for efficient cnn architec- ture design

    Ningning Ma, Xiangyu Zhang, Hai-Tao Zheng, and Jian Sun. Shufflenet v2: Practical guidelines for efficient cnn architec- ture design. In ECCV, 2018. 1

    2018

  30. [38]

    Dinov2: Learning robust visual features without supervision

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193, 2023. 2, 12

    2023 arXiv

  31. [39]

    Forward and backward information retention for accurate binary neural networks

    Haotong Qin, Ruihao Gong, Xianglong Liu, Mingzhu Shen, Ziran Wei, Fengwei Yu, and Jingkuan Song. Forward and backward information retention for accurate binary neural networks. In CVPR, 2020. 2, 3

    2020

  32. [40]

    Bibert: Accurate fully binarized bert

    Haotong Qin, Yifu Ding, Mingyuan Zhang, Y AN Qinghua, Aishan Liu, Qingqing Dang, Ziwei Liu, and Xianglong Liu. Bibert: Accurate fully binarized bert. In ICLR, 2022. 12

    2022

  33. [41]

    Bibert: Accurate fully binarized bert

    Haotong Qin, Yifu Ding, Mingyuan Zhang, Qinghua Yan, Aishan Liu, Qingqing Dang, Ziwei Liu, and Xianglong Liu. Bibert: Accurate fully binarized bert. arXiv preprint arXiv:2203.06390, 2022. 12

    2022 arXiv

  34. [42]

    Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer

    Ren ´e Ranftl, Katrin Lasinger, David Hafner, Konrad Schindler, and Vladlen Koltun. Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer. IEEE transactions on pattern analysis and machine intelligence, 2020. 6

    2020

  35. [43]

    Vi- sion transformers for dense prediction

    Ren ´e Ranftl, Alexey Bochkovskiy, and Vladlen Koltun. Vi- sion transformers for dense prediction. In ICCV, 2021. 2, 5, 6, 12

    2021

  36. [44]

    Xnor-net: Imagenet classification using bi- nary convolutional neural networks

    Mohammad Rastegari, Vicente Ordonez, Joseph Redmon, and Ali Farhadi. Xnor-net: Imagenet classification using bi- nary convolutional neural networks. In ECCV, 2016. 1, 2

    2016

  37. [45]

    R2d2: Reliable and repeatable detec- tor and descriptor

    Jerome Revaud, Cesar De Souza, Martin Humenberger, and Philippe Weinzaepfel. R2d2: Reliable and repeatable detec- tor and descriptor. In NeurIPS, 2019. 2

    2019

  38. [46]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In MICCAI, 2015. 1

    2015

  39. [47]

    Mobilenetv2: Inverted residuals and linear bottlenecks

    Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zh- moginov, and Liang-Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In CVPR, 2018. 1

    2018

  40. [48]

    Indoor segmentation and support inference from rgbd images

    Nathan Silberman, Derek Hoiem, Pushmeet Kohli, and Rob Fergus. Indoor segmentation and support inference from rgbd images. In ECCV, 2012. 6, 7, 13

    2012

  41. [49]

    Super-convergence: Very fast training of neural networks using large learning rates

    Leslie N Smith and Nicholay Topin. Super-convergence: Very fast training of neural networks using large learning rates. In Artificial intelligence and machine learning for multi-domain operations applications, 2019. 5

    2019

  42. [50]

    Segmenter: Transformer for semantic segmenta- tion

    Robin Strudel, Ricardo Garcia, Ivan Laptev, and Cordelia Schmid. Segmenter: Transformer for semantic segmenta- tion. In ICCV, 2021. 2

    2021

  43. [51]

    Efficient processing of deep neural networks: A tutorial and survey

    Vivienne Sze, Yu-Hsin Chen, Tien-Ju Yang, and Joel S Emer. Efficient processing of deep neural networks: A tutorial and survey. Proceedings of the IEEE, 2017. 1

    2017

  44. [52]

    Adabin: Improving binary neural networks with adaptive bi- nary sets

    Zhijun Tu, Xinghao Chen, Pengju Ren, and Yunhe Wang. Adabin: Improving binary neural networks with adaptive bi- nary sets. In ECCV, 2022. 1, 2, 5, 6, 7, 12, 14

    2022

  45. [53]

    Disk: Learning local features with policy gradient

    Michał Tyszkiewicz, Pascal Fua, and Eduard Trulls. Disk: Learning local features with policy gradient. In NeurIPS,

  46. [54]

    Trainability preserving neural prun- ing

    Huan Wang and Yun Fu. Trainability preserving neural prun- ing. In ICLR, 2023. 1

    2023

  47. [55]

    Image as a foreign language: Beit pretraining for vision and vision- language tasks

    Wenhui Wang, Hangbo Bao, Li Dong, Johan Bjorck, Zhil- iang Peng, Qiang Liu, Kriti Aggarwal, Owais Khan Mo- hammed, Saksham Singhal, Subhojit Som, et al. Image as a foreign language: Beit pretraining for vision and vision- language tasks. In CVPR, 2023. 2

    2023

  48. [56]

    Basic binary convolution unit for binarized image restoration network

    Bin Xia, Yulun Zhang, Yitong Wang, Yapeng Tian, Wen- ming Yang, Radu Timofte, and Luc Van Gool. Basic binary convolution unit for binarized image restoration network. In ICLR, 2023. 2

    2023

  49. [57]

    Unified perceptual parsing for scene understand- ing

    Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, and Jian Sun. Unified perceptual parsing for scene understand- ing. In ECCV, 2018. 2, 5

    2018

  50. [58]

    Segformer: Simple and ef- ficient design for semantic segmentation with transformers

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and ef- ficient design for semantic segmentation with transformers. In NeurIPS, 2021. 2

    2021

  51. [59]

    Aggregated residual transformations for deep neural networks

    Saining Xie, Ross Girshick, Piotr Doll ´ar, Zhuowen Tu, and Kaiming He. Aggregated residual transformations for deep neural networks. In CVPR, 2017. 2

    2017

  52. [60]

    Depth anything: Unleashing the power of large-scale unlabeled data

    Lihe Yang, Bingyi Kang, Zilong Huang, Xiaogang Xu, Jiashi Feng, and Hengshuang Zhao. Depth anything: Unleashing the power of large-scale unlabeled data. In CVPR, 2024. 2

    2024

  53. [61]

    Lift: Learned invariant feature transform

    Kwang Moo Yi, Eduard Trulls, Vincent Lepetit, and Pascal Fua. Lift: Learned invariant feature transform. In ECCV,

  54. [62]

    Lq-nets: Learned quantization for highly accurate and compact deep neural networks

    Dongqing Zhang, Jiaolong Yang, Dongqiangzi Ye, and Gang Hua. Lq-nets: Learned quantization for highly accurate and compact deep neural networks. In ECCV, 2018. 6

    2018

  55. [63]

    Shufflenet: An extremely efficient convolutional neural net- work for mobile devices

    Xiangyu Zhang, Xinyu Zhou, Mengxiao Lin, and Jian Sun. Shufflenet: An extremely efficient convolutional neural net- work for mobile devices. In CVPR, 2018. 1

    2018

  56. [64]

    Flexible residual binariza- tion for image super-resolution

    Yulun Zhang, Haotong Qin, Zixiang Zhao, Xianglong Liu, Martin Danelljan, and Fisher Yu. Flexible residual binariza- tion for image super-resolution. In ICML, 2024. 2

    2024

  57. [65]

    Pyramid scene parsing network

    Hengshuang Zhao, Jianping Shi, Xiaojuan Qi, Xiaogang Wang, and Jiaya Jia. Pyramid scene parsing network. In CVPR, 2017. 2

    2017

  58. [66]

    Alike: Accurate and lightweight keypoint detection and descriptor extraction

    Xiaoming Zhao, Xingming Wu, Jinyu Miao, Weihai Chen, Peter CY Chen, and Zhengguo Li. Alike: Accurate and lightweight keypoint detection and descriptor extraction. IEEE Transactions on Multimedia, 2022. 2

    2022

  59. [67]

    Scene parsing through ade20k dataset

    Bolei Zhou, Hang Zhao, Xavier Puig, Sanja Fidler, Adela Barriuso, and Antonio Torralba. Scene parsing through ade20k dataset. In CVPR, 2017. 4, 5, 6, 7, 8, 12, 14

    2017

  60. [68]

    Cellular binary neural network for accurate im- 10 age classification and semantic segmentation

    Xichuan Zhou, Rui Ding, Yuxiao Wang, Wenjia Wei, and Haijun Liu. Cellular binary neural network for accurate im- 10 age classification and semantic segmentation. IEEE Trans- actions on Multimedia, 2022. 2, 6

    2022

  61. [69]

    Structured binary neural networks for accurate image classification and semantic segmentation

    Bohan Zhuang, Chunhua Shen, Mingkui Tan, Lingqiao Liu, and Ian Reid. Structured binary neural networks for accurate image classification and semantic segmentation. In CVPR,

  62. [2019]

    Semantic segmentation results of the DPT architecture on ADE20K [67] val, expanding upon Tab

    2, 6 11 BiDense: Binarization for Dense Prediction Supplementary Material Method Bit Params (K) OPs (G) pixAcc (%) mIoU (%) ConvNeXt* [33] 32 40,020 288.36 79.48 40.96 ConvNeXt [33] 32 40,023 288.37 70.81 20.39 BNN [23] 1 1,470 4.84 61.69 8.68 ReActNet [30] 1 1,531 4.98 62.77 ...

    2012

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.