Pith. sign in

REVIEW 2 major objections 6 minor 42 references

MultiTaskVIF: Segmentation-oriented visible and infrared image fusion via multi-task learning

T0 review · 2 major / 6 minor · reviewed 2026-08-15 · 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 replacing a visible–infrared fusion network's decoder with a two-branch Multi-Task Head turns it into a segmentation-aware fusion model in a single training stage, and that this single-model framework outperforms…

desk verdict Useful training-simplification idea for segmentation-oriented VIF, but the central claim that the Multi-Task Head drives the gains is not established because comparisons to prior methods use pretrained checkpoints under different training protocols. read the letter →

arxiv 2505.06665 v1 pith:EVQFBGF6 submitted 2025-05-10 cs.CV

Zixian Zhao , Andrew Howes , Xingchen Zhang This is my paper · ORCID
classification cs.CV
keywords visible-infraredimagefusionmulti-tasklearningsemanticsegmentationheaddecodercascadetrainingframeworkqualitymetricsRGB-channel
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

This paper tries to establish that a visible–infrared image fusion network can be made segmentation-aware without bolting a second, full segmentation model onto it. The proposed recipe is to replace the fusion model's decoder with a two-branch Multi-Task Head that outputs both the fused image and a segmentation map during training, so the segmentation loss supervises the shared fusion backbone directly. On the FMB and MSRS benchmarks, the resulting single-model, single-stage framework reports the top segmentation mIoU on fused images (65.8% on FMB, tying the top 76.2% on MSRS) while also improving fusion metrics such as mutual information, VIF, and color fidelity. If this holds, cascade training—the established way to inject semantic information into fusion models—would be replaceable by a simpler decoder swap that also reduces model count, parameters, and training memory.

What carries the argument

The central object is the Multi-Task Head (MTH), a dual-branch decoder that takes the deep fused features from a VIF backbone and outputs both the fused RGB image (fusion reconstruction branch) and a segmentation map (semantic segmentation branch). Within MTH, the HIA-F module (a Hierarchical Interactive Attention module for Fusion reconstruction) uses channel embedding and two cross-attention blocks to let high- and low-level semantic features interact with fusion reconstruction features, then aggregates them with an MLP and a residual connection. This mechanism is what makes the semantic supervision direct: the segmentation branch is part of the same model and its loss updates the shared backbone, in contrast to cascade frameworks where a separate segmentation model sits after the fusion model and supervises it indirectly.

What would settle it

Retrain the cascade baselines (e.g., SeAFusion and SegMiF) from scratch on FMB and MSRS using three-channel input, the same 256x256 crops, batch size 16, optimizer, early stopping, and the same unified SegFormer evaluator; if their mIoU and fusion metrics match or exceed MultiTaskVIF IV's 65.8/76.2 mIoU and MI/VIF/DeltaE values, the central claim that the decoder-swap framework is responsible for the gains would be contradicted.

Watch

Extended reading notes

Core claim

The central claim is that MultiTaskVIF, a universal training framework, lets an existing VIF model learn semantic features simply by replacing its decoder with the proposed Multi-Task Head (MTH), without joint training with a complete segmentation model. During training, MTH simultaneously reconstructs the fused image and produces a segmentation output; the total loss sums a fusion loss (intensity, gradient, SSIM, and color terms) and a segmentation loss (cross-entropy plus dice), and both branches share the fusion backbone. The paper demonstrates this on four backbones (SwinFusion, EMMA, SeAFusion, and SegMiF) and reports that the best variant, MultiTaskVIF IV, attains the top mIoU on FMB (65.8) and ties the top on MSRS (76.2) among all compared methods, while also delivering top or near-top fusion quality on metrics like MI, VIF, and DeltaE. The ablation study attributes the gain to the MTH dual-branch structure, the HIA-F module that cross-attends segmentation and fusion features, the segmentation loss, the color-preserving loss, and three-channel RGB fusion.

Load-bearing premise

The load-bearing premise is that the public checkpoints of the cascade baselines are a fair comparison to MultiTaskVIF, which is trained from scratch with different input channels, crop sizes, batch sizes, and loss terms; if retraining the baselines under the same protocol closes the performance gap, the framework's claimed advantage is not established.

Editorial extensions

If this is right

  • Any VIF backbone whose decoder can be detached can be turned into a segmentation-oriented model with one training stage and one set of parameters, so the framework is a drop-in alternative to cascade training rather than a new fusion architecture.
  • Fused images from MultiTaskVIF carry more segmentable structure: variant IV reports 65.8 mIoU on FMB and 76.2 on MSRS, with improved mutual information and color fidelity over the same-backbone cascade baselines.
  • Training memory and parameters drop sharply compared with cascade competitors: variant IV reports 1.28M parameters and 214.52 MB memory versus SegMiF's 45.25M parameters and 430.90 MB, enabling larger batches or crops under fixed GPU memory.
  • The positive gradient projections between fusion and segmentation losses (1.9e-1 and 2.7e-3) indicate the two tasks reinforce each other, so segmentation utility and fusion fidelity need not be traded off.
  • Ablations show each component carries weight: removing HIA-F, the segmentation loss, the color loss, or switching to single-channel input each degrades mIoU or color quality, and the full configuration is best.

Reading between the lines

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

  • Beyond the paper, the decoder-swap recipe should transfer to other downstream heads (detection, instance segmentation, depth) since MTH only requires a task head with labels; a test would be swapping in a detection head on the same backbone.
  • Beyond the paper, the three-channel versus single-channel ablation suggests that the common VIF practice of fusing only the luminance channel may discard color cues that help dense prediction; a cross-dataset comparison would tell whether this holds generally.
  • Beyond the paper, the reported gradient projections suggest an automatic scheme for weighting the fusion and segmentation losses could replace the hand-set beta and lambda values and possibly improve stability.
  • Beyond the paper, since the unified segmentation evaluator is fine-tuned on averaged fused images, the absolute mIoU numbers might be evaluator-specific; testing with a segmentation model trained solely on source images would probe whether the semantic gain is intrinsic to the fused images.
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

2 major / 6 minor

Summary. The paper proposes MultiTaskVIF, a training framework for visible-infrared image fusion (VIF) that replaces a fusion model's decoder with a multi-task head (MTH) which simultaneously outputs the fused image and a semantic segmentation map. This is intended to replace the cascade structure used in prior application-oriented VIF methods, where a separate segmentation model is trained on the fused output. The authors instantiate four variants using the backbones of SwinFusion, EMMA, SeAFusion, and SegMiF, all trained from scratch on the FMB and MSRS datasets. They report improved mIoU for downstream segmentation of the fused images and competitive fusion quality metrics, as well as lower memory usage during training. The paper also introduces a Hierarchical Interactive Attention module for fusion (HIA-F) and a loss combination that includes intensity, gradient, SSIM, and color-preservation terms.

Significance. If the central claim holds, the framework is a useful simplification for segmentation-oriented VIF: it avoids training a separate segmentation model and reduces memory and multi-stage training costs. The evaluation uses a unified external segmentation model for all fused images, which is a good practice for comparing fusion methods fairly. The paper also provides ablations for HIA-F, individual loss terms, and input channels. The main significance is limited by the lack of a controlled comparison isolating the effect of the proposed MTH from the accompanying training-protocol changes; the reported gains over baselines are small and within the range of effects that the protocol changes themselves can produce.

major comments (2)
  1. [Sec. 4.5, Exp. i and Sec. 4.2] The ablation intended to verify the effectiveness of MTH does not control for training protocol. The paper states in Sec. 4.5 that "due to the inability to maintain consistent loss functions and training settings across different network architectures, we directly compared our models with existing VIF models that share the same backbone." Consequently, the comparison in Table 6 between the CNN decoder (original methods) and the MTH variants changes not only the decoder but also the input channels (1-channel vs. 3-channel), the loss functions (the original methods do not use L_seg or L_color), and the training settings (crop size, batch size, learning rate, early stopping, training from scratch vs. pretrained checkpoints). The reported mIoU improvements (e.g., 65.7 vs. 64.6 for SwinFusion and 65.8 vs. 65.4 for SegMiF) are of the same magnitude as the effects of the protocol changes shown in the paper's own ablations: Exp. v (3-channel to 1-channel) changes mIoU by -1.2 on FMB, Exp. iv (removing L_color) by -0.5, and Exp. iii (removing L_seg) by -0.6. Therefore, the central claim that the MTH architecture itself is responsible for the improvements is not established.
  2. [Sec. 4.3, Tables 2-3] The comparison against SOTA methods is not controlled. All baseline methods are used via their public pretrained checkpoints (Sec. 4.2), while MultiTaskVIF I-IV are trained from scratch on FMB and MSRS with a different protocol (Sec. 4.1): 3-channel input, new loss terms (including L_color and L_seg), different crop sizes, batch sizes, and learning rates. The headline advantages over the strongest baselines are small: on FMB, MultiTaskVIF IV achieves mIoU 65.8 vs. SegMiF's 65.4 (a 0.4-point gain), and on MSRS, MultiTaskVIF II/IV tie with CDDFuse and EMMA at 76.2. Given that the ablation results in Table 6 show protocol-induced changes of 0.5-1.2 mIoU points, the reported gains cannot be unambiguously attributed to the MultiTaskVIF framework. A controlled experiment that trains the baseline backbones under the identical protocol (same input channels, loss terms, optimizer, and data augmentation) while only swapping the decoder is needed to support the paper's central claim.
minor comments (6)
  1. [Sec. 3.4] The text contains a typo: "In summery" should be "In summary".
  2. [Sec. 3.1, Eq. (1)] The notation in Eq. (1), Ff(f, Nm(v,i; ωb, ωf)) and Fs(s, Nm(v,i; ωb, ωs)), is ambiguous; these are likely loss functions rather than fidelity terms. Please clarify the notation and the meaning of the two F terms.
  3. [Sec. 4.4 vs. Tables 4-5] The claim that "Our method has excellent performance on almost all metrics" is too broad. For instance, on FMB the EN of MultiTaskVIF I (6.63) is lower than CDDFuse (6.78), and on MSRS the SSIM of MultiTaskVIF II (0.99) is lower than SeAFusion's 0.99 (though similar). Please phrase the summary of fusion results more precisely.
  4. [Sec. 4.6, Table 7] The efficiency comparison in Table 7 is not apples-to-apples: MultiTaskVIF variants use only the backbone of the original methods (e.g., MultiTaskVIF III has 0.93M parameters vs. SeAFusion's 13.06M, which includes its full architecture with the segmentation model). The text should explicitly state that the comparison isolates the fusion backbone and that the parameter/memory reductions come from removing the separate segmentation model, rather than from the MTH itself.
  5. [Sec. 4.1 and references] There is a citation inconsistency: in Sec. 4.1 the paper refers to "SeAFusion [9]", but SeAFusion is correctly cited as [26] elsewhere (e.g., Sec. 2.2). Reference [9] is a different paper (Liu et al., CVPR 2022). Please correct the citation.
  6. [Sec. 4.7] The gradient analysis reports projections of 1.9e-1 for the fusion task onto the segmentation task and 2.7e-3 for the segmentation task onto the fusion task. This asymmetry is large and is not discussed; the claim that "both positive values confirm that the segmentation branch positively contributes to fusion model training" is weakened by the near-zero second value. Please analyze or qualify this result.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the framework claim is supported by independent benchmark evaluations and no equation reduces to its own inputs.

full rationale

The paper's central claim is an empirical architectural one: replacing a VIF decoder with the proposed Multi-Task Head (MTH) provides a single-model, single-stage training framework that improves both fusion quality and downstream segmentation utility. Nothing in the derivation chain reduces to the paper's own inputs by construction. The loss function L_total = beta_1 L_fusion + beta_2 L_seg combines standard intensity, gradient, SSIM, color, cross-entropy, and dice terms; these are not fitted to the evaluation metrics and are not renamed predictions. The segmentation evaluation uses external benchmark datasets (FMB, MSRS) and a unified Segformer-based model fine-tuned outside the proposed framework, rather than the paper's own segmentation branch, so the reported mIoU values are not self-defined. The MTH and HIA-F are described as building on prior HIA work [10], but this is a design inspiration, not a load-bearing uniqueness theorem or an ansatz smuggled in via self-citation. The authors do cite their own prior works CDDFuse [41] and EMMA [42], but only as VIF methods whose backbones are borrowed for generalizability experiments; the central framework claim does not rest on those citations. The ablation in Sec. 4.5 compares against pretrained checkpoints while MultiTaskVIF is trained from scratch with different input channels, losses, and hyperparameters, and the paper explicitly concedes this limitation: 'due to the inability to maintain consistent loss functions and training settings across different network architectures, we directly compared our models with existing VIF models that share the same backbone.' That is a comparison-validity confound, not a circular derivation; it does not make the framework's output equivalent to its input by construction. The gradient analysis in Sec. 4.7 is post hoc evidence rather than a derivation. Accordingly, no circular step is present, and the honest finding is a score of 0.

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

The central claim rests on three domain assumptions: that segmentation supervision transfers to fusion quality, that the unified segmentation evaluator is unbiased, and that pretrained baselines are comparable. The loss weights and training hyperparameters are fitted by hand to the evaluation datasets. No new physical entities are introduced; MTH and HIA-F are architectural components within the model, not independent entities with external falsifiable handles.

free parameters (3)
  • Loss weights (beta1, beta2, lambda1, lambda2, lambda3, lambda4) = (1, 1, 20, 20, 10, 20)
    Sec. 3.4 sets these by manual tuning; they balance fusion and segmentation losses. The reported mIoU and fusion metrics depend on them.
  • SSIM term weights w_j = 0.5 per source image
    Sec. 3.4 sets w_j = 0.5 for infrared and visible, borrowed from prior work.
  • Training hyperparameters (learning rate, crop size, batch size, early stopping patience) = Model I: 1e-4, 160x160, batch 10; Models II-IV: 0.5e-4, 256x256, batch 16; patience 10
    Sec. 4.1 states different values per backbone complexity; these choices affect optimization and could contribute to performance differences versus baselines.
assumptions (3)
  • domain assumption Segmentation labels provide a useful auxiliary learning signal for fusion reconstruction.
    The method adds L_seg to train fusion; ablation Exp. iii (Table 6) shows removing it lowers mIoU from 65.8 to 65.2, so the assumption is empirical rather than derived.
  • domain assumption The unified Segformer-based segmentation model fine-tuned on average-fused images gives unbiased mIoU comparisons across all fused images.
    Sec. 4.3 introduces this evaluator; if it favors outputs distributionally similar to average fusion or to the proposed method, the ranking of methods could change.
  • domain assumption Public pretrained checkpoints of compared methods are representative of those methods' performance and are directly comparable to MultiTaskVIF models trained from scratch.
    Sec. 4.2 uses default models from public code; training protocol differences (channels, crop, batch, epochs) are not controlled.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MultiTaskVIF: Segmentation-oriented visible and infrared image fusion via multi-task learning." pith.science (2026). https://pith.science/paper/EVQFBGF6

@misc{pith2026250506665,
  author       = {Pith},
  title        = {Pith review of: MultiTaskVIF: Segmentation-oriented visible and infrared image fusion via multi-task learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EVQFBGF6}},
  note         = {Machine review of arXiv:2505.06665}
}
read the original abstract

Visible and infrared image fusion (VIF) has attracted significant attention in recent years. Traditional VIF methods primarily focus on generating fused images with high visual quality, while recent advancements increasingly emphasize incorporating semantic information into the fusion model during training. However, most existing segmentation-oriented VIF methods adopt a cascade structure comprising separate fusion and segmentation models, leading to increased network complexity and redundancy. This raises a critical question: can we design a more concise and efficient structure to integrate semantic information directly into the fusion model during training-Inspired by multi-task learning, we propose a concise and universal training framework, MultiTaskVIF, for segmentation-oriented VIF models. In this framework, we introduce a multi-task head decoder (MTH) to simultaneously output both the fused image and the segmentation result during training. Unlike previous cascade training frameworks that necessitate joint training with a complete segmentation model, MultiTaskVIF enables the fusion model to learn semantic features by simply replacing its decoder with MTH. Extensive experimental evaluations validate the effectiveness of the proposed method. Our code will be released upon acceptance.

Figures

Figures reproduced from arXiv: 2505.06665 by the authors.

Figure 1
Figure 1. Existing cascade training framework vs. MultiTaskVIF. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The workflow of proposed MultiTaskVIF training framework for AO VIF. Notably, MultiTaskVIF exhibits strong generalizability [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Qualitative segmentation on the FMB dataset. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative segmentation on the MSRS dataset. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Feature maps of the two branches within MTH. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 36 canonical work pages

  1. [1]

    Loss func- tions in the era of semantic segmentation: A survey and out- look

    Reza Azad, Moein Heidary, Kadir Yilmaz, et al. Loss func- tions in the era of semantic segmentation: A survey and out- look. arXiv preprint arXiv:2312.05391, 2023. v

    arXiv 2023

  2. [2]

    A unified architecture for natural language processing: Deep neural networks with multitask learning

    Ronan Collobert and Jason Weston. A unified architecture for natural language processing: Deep neural networks with multitask learning. In Proceedings of the 25th international conference on Machine learning, pages 160–167, 2008. iii

    work page 2008

  3. [3]

    New types of deep neural network learning for speech recognition and related applications: An overview

    Li Deng, Geoffrey Hinton, and Brian Kingsbury. New types of deep neural network learning for speech recognition and related applications: An overview. In Proceedings of 2013 IEEE international conference on acoustics, speech and sig- nal processing, pages 8599–8603. IEEE, 2013. iii

    work page 2013

  4. [4]

    The ciede2000 color-difference formula: Implementation notes, supplementary test data, and mathe- matical observations

    Sharma Gaurav. The ciede2000 color-difference formula: Implementation notes, supplementary test data, and mathe- matical observations. COLOR research and application, 30 (1):21–30, 2005. vi

    work page 2005

  5. [5]

    Fast R-CNN

    Ross Girshick. Fast R-CNN. In Proceedings of the IEEE international conference on computer vision , pages 1440– 1448, 2015. iii

    work page 2015

  6. [6]

    R-cnns for pose estimation and action detec- tion

    Georgia Gkioxari, Bharath Hariharan, Ross Girshick, and Ji- tendra Malik. R-cnns for pose estimation and action detec- tion. arXiv preprint arXiv:1406.5212, 2014. iii

    arXiv 2014

  7. [7]

    Multi-task learning using uncertainty to weigh losses for scene geome- try and semantics

    Alex Kendall, Yarin Gal, and Roberto Cipolla. Multi-task learning using uncertainty to weigh losses for scene geome- try and semantics. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 7482–7491,

  8. [8]

    DenseFuse: A fusion approach to infrared and visible images

    Hui Li and Xiaojun Wu. DenseFuse: A fusion approach to infrared and visible images. IEEE Transactions on Image Processing, 28(5):2614–2623, 2019. i

    work page 2019

Show all 42 references
  1. [9]

    Target- aware dual adversarial Learning and a multi-scenario multi- modality benchmark to fuse infrared and visible for object detection

    Jinyuan Liu, Xin Fan, Zhanbo Huang, et al. Target- aware dual adversarial Learning and a multi-scenario multi- modality benchmark to fuse infrared and visible for object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. i, v

    2022

  2. [10]

    Multi-interactive feature learning and a full-time multi-modality benchmark for image fusion and segmentation

    Jinyuan Liu, Zhu Liu, Guanyao Wu, Long Ma, Risheng Liu, Wei Zhong, Zhongxuan Luo, and Xin Fan. Multi-interactive feature learning and a full-time multi-modality benchmark for image fusion and segmentation. In International Confer- ence on Computer Vision, 2023. i, ii, iii, iv,...

    2023

  3. [11]

    A task-guided, implicitly-searched and metainitialized deep model for image fusion

    Risheng Liu, Zhu Liu, Jinyuan Liu, Xin Fan, and Zhongxuan Luo. A task-guided, implicitly-searched and metainitialized deep model for image fusion. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024. vi

    2024

  4. [12]

    Infrared and visible image fusion with convolutional neural networks

    Yu Liu, Xun Chen, Juan Cheng, Hu Peng, and Zengfu Wang. Infrared and visible image fusion with convolutional neural networks. International Journal of Wavelets, Multiresolution and Information Processing, 16(03):1850018, 2018. ii

    2018

  5. [13]

    Paif: Perception-aware infrared-visible im- age fusion for attack-tolerant semantic segmentation

    Zhu Liu, Jinyuan Liu, Benzhuang Zhang, Long Ma, Xin Fan, and Risheng Liu. Paif: Perception-aware infrared-visible im- age fusion for attack-tolerant semantic segmentation. InPro- ceedings of the 31st ACM international conference on multi- media, pages 3706–3714, 2023. ii, vi

    2023

  6. [14]

    Infrared and visible im- age fusion methods and applications: A survey

    Jiayi Ma, Yong Ma, and Chang Li. Infrared and visible im- age fusion methods and applications: A survey. Information Fusion, 45:153–178, 2019. i, vi

    2019

  7. [15]

    Fusiongan: A generative adversarial network for in- frared and visible image fusion

    Jiayi Ma, Wei Yu, Pengwei Liang, Chang Li, and Junjun Jiang. Fusiongan: A generative adversarial network for in- frared and visible image fusion. Information fusion, 48:11– 26, 2019. ii

    2019

  8. [16]

    SwinFusion: Cross- domain Long-range Learning for General Image Fusion via Swin Transformer

    Jiayi Ma, Linfeng Tang, Fan Fan, et al. SwinFusion: Cross- domain Long-range Learning for General Image Fusion via Swin Transformer. IEEE/CAA Journal of Automatica Sinica, 9(7):1200–1217, 2022. iii, iv, v, vi

    2022

  9. [17]

    AT- GAN: A generative adversarial network with attention and transition for infrared and visible image fusion

    Yujing Rao, Dan Wu, Mina Han, Ting Wang, Yang Yang, Tao Lei, Chengjiang Zhou, Haicheng Bai, and Lin Xing. AT- GAN: A generative adversarial network with attention and transition for infrared and visible image fusion. Information Fusion, 92:336–349, 2023. i

    2023

  10. [18]

    Learning multiple visual domains with residual adapters.Ad- vances in neural information processing systems , 30, 2017

    Sylvestre-Alvise Rebuffi, Hakan Bilen, and Andrea Vedaldi. Learning multiple visual domains with residual adapters.Ad- vances in neural information processing systems , 30, 2017. iii

    2017

  11. [19]

    ” grabcut” interactive foreground extraction using iterated graph cuts

    Carsten Rother, Vladimir Kolmogorov, and Andrew Blake. ” grabcut” interactive foreground extraction using iterated graph cuts. ACM transactions on graphics (TOG) , 23(3): 309–314, 2004. viii

    2004

  12. [20]

    An overview of multi-task learning in deep neural networks

    Sebastian Ruder. An overview of multi-task learning in deep neural networks. arXiv preprint arXiv:1706.05098, 2017. iii

    2017 arXiv

  13. [21]

    Overfeat: Integrated recognition, localization and detection using convolutional networks

    Pierre Sermanet, David Eigen, Xiang Zhang, Micha ¨el Math- ieu, Rob Fergus, and Yann LeCun. Overfeat: Integrated recognition, localization and detection using convolutional networks. arXiv preprint arXiv:1312.6229, 2013. iii, iv

    2013 arXiv

  14. [22]

    Deep visible and thermal image fusion for enhanced pedes- trian visibility

    Ivana Shopovska, Ljubomir Jovanov, and Wilfried Philips. Deep visible and thermal image fusion for enhanced pedes- trian visibility. Sensors, 19(17):3727, 2019. i

    2019

  15. [23]

    Generalised dice overlap as a deep learning loss function for highly unbalanced seg- mentations

    Carole H Sudre, Wenqi Li, Tom Vercauteren, Sebastien Ourselin, and M Jorge Cardoso. Generalised dice overlap as a deep learning loss function for highly unbalanced seg- mentations. In Proceedings of Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical D...

    2017

  16. [24]

    Det- fusion: A detection-driven infrared and visible image fusion network

    Yiming Sun, Bing Cao, Pengfei Zhu, and Qinghua Hu. Det- fusion: A detection-driven infrared and visible image fusion network. In Proceedings of the 30th ACM International Con- ference on Multimedia, pages 4003–4011, 2022. ii

    2022

  17. [25]

    SuperFusion: A versatile image registration and fusion ix network with semantic awareness

    Linfeng Tang, Yuxin Deng, Yong Ma, Jun Huang, and Jiayi Ma. SuperFusion: A versatile image registration and fusion ix network with semantic awareness. IEEE/CAA Journal of Au- tomatica Sinica, 9(12):2121–2137, 2022. i

    2022

  18. [26]

    Image fusion in the loop of high-level vision tasks: A semantic-aware real- time infrared and visible image fusion network

    Linfeng Tang, Jiteng Yuan, and Jiayi Ma. Image fusion in the loop of high-level vision tasks: A semantic-aware real- time infrared and visible image fusion network. Information Fusion, 82:28–42, 2022. i, ii, iii, iv, vi

    2022

  19. [27]

    Piafusion: A progressive infrared and visible im- age fusion network based on illumination aware.Information Fusion, 83:79–92, 2022

    Linfeng Tang, Jiteng Yuan, Hao Zhang, Xingyu Jiang, and Jiayi Ma. Piafusion: A progressive infrared and visible im- age fusion network based on illumination aware.Information Fusion, 83:79–92, 2022. v, vii

    2022

  20. [28]

    YDTR: infrared and visible image fusion via y-shape dynamic transformer

    Wei Tang, Fazhi He, and Yu Liu. YDTR: infrared and visible image fusion via y-shape dynamic transformer. IEEE Trans. Multimedia, 2022. i

    2022

  21. [29]

    An inter- actively reinforced paradigm for joint infrared-visible image fusion and saliency object detection

    Di Wang, Jinyuan Liu, Risheng Liu, and Xin Fan. An inter- actively reinforced paradigm for joint infrared-visible image fusion and saliency object detection. Information Fusion , page 101828, 2023. iii

    2023

  22. [30]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, Eero P Simon- celli, et al. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Process- ing, 13(4):600–612, 2004. v

    2004

  23. [31]

    Segformer: Simple and efficient design for semantic segmentation with transform- ers

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and efficient design for semantic segmentation with transform- ers. Advances in neural information processing systems, 34: 12077–12090, 2021. vii

    2021

  24. [32]

    U2fusion: A unified unsupervised image fusion net- work

    Han Xu, Jiayi Ma, Junjun Jiang, Xiaojie Guo, and Haibin Ling. U2fusion: A unified unsupervised image fusion net- work. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(1):502–518, 2020. ii

    2020

  25. [33]

    Advanced corrective training strategy for surrogating complex hysteretic behavior

    Yongjia Xu, Yifan Fei, Yuli Huang, Yuan Tian, and Xinzheng Lu. Advanced corrective training strategy for surrogating complex hysteretic behavior. In Structures, pages 1792–

  26. [34]

    Demt: De- formable mixer transformer for multi-task learning of dense prediction

    Yangyang Xu, Yibo Yang, and Lefei Zhang. Demt: De- formable mixer transformer for multi-task learning of dense prediction. In Proceedings of the AAAI conference on artifi- cial intelligence, pages 3072–3080, 2023. iii

    2023

  27. [35]

    Dif-fusion: Towards high color fidelity in infrared and visi- ble image fusion with diffusion models

    Jun Yue, Leyuan Fang, Shaobo Xia, Yue Deng, and Jiayi Ma. Dif-fusion: Towards high color fidelity in infrared and visi- ble image fusion with diffusion models. IEEE Transactions on Image Processing, 2023. i, ii

    2023

  28. [36]

    MRFS: Mutually Reinforcing Image Fusion and Segmentation

    Hao Zhang, Xuhui Zuo, Jie Jiang, Chunchao Guo, and Ji- ayi Ma. MRFS: Mutually Reinforcing Image Fusion and Segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26974– 26983, 2024. i, iii, vi

    2024

  29. [37]

    Transformer based conditional gan for multimodal image fusion

    Jun Zhang, Licheng Jiao, Wenping Ma, Fang Liu, Xu Liu, Lingling Li, Puhua Chen, and Shuyuan Yang. Transformer based conditional gan for multimodal image fusion. IEEE Transactions on Multimedia, 25:8988–9001, 2023. i

    2023

  30. [38]

    Visible and infrared image fusion using deep learning.IEEE Transactions on Pat- tern Analysis and Machine Intelligence, 45(8):10535–10554,

    Xingchen Zhang and Yiannis Demiris. Visible and infrared image fusion using deep learning.IEEE Transactions on Pat- tern Analysis and Machine Intelligence, 45(8):10535–10554,

  31. [39]

    Object fusion tracking based on visible and infrared images using fully convolutional siamese networks

    Xingchen Zhang, Gang Xiao, Ping Ye, Dan Qiao, Junhao Zhao, and Shengyun Peng. Object fusion tracking based on visible and infrared images using fully convolutional siamese networks. In Proceedings of the 22nd International Confer- ence on Information Fusion. IEEE, 2019. i

    2019

  32. [40]

    VIFB: A visi- ble and infrared image fusion benchmark

    Xingchen Zhang, Ping Ye, and Gang Xiao. VIFB: A visi- ble and infrared image fusion benchmark. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pages 104–105, 2020. vi

    2020

  33. [41]

    Cddfuse: Correlation-driven dual-branch feature decompo- sition for multi-modality image fusion

    Zixiang Zhao, Haowen Bai, Jiangshe Zhang, Yulun Zhang, Shuang Xu, Zudi Lin, Radu Timofte, and Luc Van Gool. Cddfuse: Correlation-driven dual-branch feature decompo- sition for multi-modality image fusion. In Proceedings of the IEEE/CVF conference on computer vision and pattern...

    2023

  34. [42]

    Equivariant multi-modality image fusion

    Zixiang Zhao, Haowen Bai, Jiangshe Zhang, Yulun Zhang, Kai Zhang, Shuang Xu, Dongdong Chen, Radu Timofte, and Luc Van Gool. Equivariant multi-modality image fusion. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 25912–25921, 2024....

    2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.