Pith. sign in

REVIEW 4 major objections 6 minor 60 references

ProtoHGF-Net: Prototype HyperGraph Fusion with Intra-modal Calibration for RGBT Object Detection

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read ProtoHGF-Net claims that replacing dense cross-modal fusion with prototype-level hypergraph fusion, after teacher-mask calibration, delivers the best published RGBT object detection results on two benchmarks and the best overall mAP on a…

desk verdict A reasonable RGBT fusion and distillation paper whose two strong benchmark results are undercut by a false FLIR SOTA claim and a loss-ablation table that contradicts its own prose. read the letter →

arxiv 2608.11595 v1 pith:3PYZWZNH submitted 2026-08-12 cs.CV

classification cs.CV
keywords RGBTobjectdetectionprototypefusionhypergraphknowledgedistillationteacher-maskcalibrationcross-modalDroneVehicleYOLOv8
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

ProtoHGF-Net claims that RGB-thermal object detection works better when cross-modal fusion happens over a handful of semantic prototypes rather than over every spatial location in a full-resolution feature map. The paper reports 85.9% $mAP_{50}$ on DroneVehicle, 88.2% $mAP_{50}$ on DVTOD, and 79.1% $mAP_{50}$ on FLIR, with the best published $mAP$ on FLIR. The method first calibrates each modality with teacher-generated target masks, then builds a sparse hypergraph over six prototypes per modality to exchange intra- and cross-modal information, and finally uses the propagated prototype states to modulate the full feature maps. The sympathetic reading is that background interference, the main failure mode of dense fusion, is suppressed before it enters the fusion step.

What carries the argument

The load-bearing object is the prototype hypergraph: a $2K \times 2K$ block relation matrix built over $K=6$ prototypes per modality, where intra-modal blocks use KNN connectivity and cross-modal blocks use top-$k$ connectivity, propagated in a vertex-to-hyperedge-to-vertex pattern with a residual connection, then converted into channel-wise affine modulation parameters that recalibrate the full-resolution features. The second mechanism is Teacher-Mask Calibration Distillation, where frozen unimodal teachers emit channel-spatial masks that split features into foreground and background, enabling foreground alignment, background suppression, and orthogonality losses. These two mechanisms jointly implement the paper's principle: calibrate first to suppress background, then fuse selectively among target-relevant prototypes.

What would settle it

Re-run the released implementation on DroneVehicle with $K$ in {4, 8}, $k_{\text{intra}}$ and $k_{\text{cross}}$ in {1, 2, 5}, and $\lambda_{\text{orth}}$ in {0.0, 0.03, 0.15}; if $mAP_{50}$ stays near 85.9 across these settings, the configuration is not load-bearing, but if it falls toward the 84.5 baseline, the reported gains are tuned to these hyperparameters.

Watch

Extended reading notes

Core claim

The paper's central claim is that the dominant dense full-resolution cross-modal interaction paradigm in RGBT detection can be replaced by prototype-level semantic interaction, and that doing so after a target-aware pre-calibration yields state-of-the-art detection. Concretely, ProtoHGF-Net compresses each modality's feature map into $K=6$ prototypes via soft attention-weighted pooling, arranges the $2K$ prototypes in a block-structured relation matrix with intra-modal KNN edges and sparse cross-modal top-$k$ edges, and runs two-step vertex-to-hyperedge-to-vertex propagation with a residual connection. The updated prototypes generate channel-wise scale-and-shift modulation parameters that recalibrate the original feature maps, followed by global gating to produce the fused output. Separately, Teacher-Mask Calibration Distillation uses frozen RGB-only and thermal-only teachers to build channel-spatial foreground masks, decomposes student and teacher features into foreground and background parts, and applies three losses: foreground alignment, background suppression, and foreground-background orthogonality. Ablations attribute the reported gains to each component, with the hypergraph beating a plain graph and hard 0/1 connections beating soft relations.

Load-bearing premise

The fixed prototype count, connectivity sparsity, and loss weights ($K=6$, $k_{\text{intra}}=3$, $k_{\text{cross}}=3$, $\lambda_{\text{fg}}=1.0$, $\lambda_{\text{bg}}=2.0$, $\lambda_{\text{orth}}=0.07$) are assumed to be the reason for the gains and to transfer across datasets, and the paper reports no sensitivity analysis for these choices.

Editorial extensions

If this is right

  • Dense full-resolution interaction is not necessary for strong RGBT fusion; prototype-level interaction with sparse hypergraph propagation suffices.
  • Pre-fusion calibration with teacher masks reduces background-driven false positives and improves both $mAP_{50}$ and $mAP$ across datasets.
  • Hypergraph propagation over prototypes outperforms plain graph propagation and soft relation matrices, indicating that sparse higher-order connectivity is a genuine source of gain.
  • The method achieves its results with 50.4M parameters and 133.38 GFLOPs, fewer parameters than several lower-scoring baselines, supporting a favorable accuracy-efficiency trade-off.
  • Combining foreground alignment, background suppression, and orthogonality regularization gives larger gains than any single distillation term, so the full loss is needed for best results.

Reading between the lines

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

  • The calibration-before-fusion ordering is likely the transferable insight: applying TM-Calib's decomposition to any fusion backbone, even dense ones, might close much of the gap; a direct test would be adding TM-Calib to a Concat or LIFAdd baseline.
  • Because $K=6$ and $k=3$ are arbitrary defaults, the method probably has a robustness plateau; sweeping $K$ from 4 to 10 would reveal whether prototype count is a sensitive knob or a stable choice.
  • The hard-top-$k$ hypergraph beating soft relations suggests that sparsity, not just higher-order modeling, is doing much of the work; sweeping $k_{\text{intra}}$ and $k_{\text{cross}}$ across values would separate the two effects.
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 / 6 minor

Summary. The paper proposes ProtoHGF-Net, a two-stream RGBT object detector built on a dual-branch YOLOv8m backbone. The method has two main components. First, ProtoHG-Fusion compresses per-modality features into K semantic prototypes, builds a block-structured relation matrix H over the 2K prototypes with intra-modal and cross-modal top-k connections, performs a two-step propagation, and uses the propagated prototypes to modulate the original full-resolution feature maps before a global gated fusion. Second, TM-Calib uses frozen modality-specific teachers to generate target-aware masks, decomposes teacher and student features into foreground and background components, and applies foreground alignment, background suppression, and orthogonality losses. The paper reports experiments on DroneVehicle, DVTOD, and FLIR, claiming state-of-the-art mAP50 of 85.9%, 88.2%, and 79.1%, respectively, along with ablations on components, losses, distillation strategies, fusion designs, and computational cost.

Significance. If the reported results hold, the paper makes a useful contribution by showing that cross-modal fusion can be performed at a compact prototype level rather than through dense full-resolution interaction, which is a plausible way to reduce background interference and computational overhead. The teacher-mask calibration idea, using frozen unimodal teachers to produce target-aware masks for pre-fusion feature calibration, is also interesting and is backed by ablations. The release of code is a strength. However, the FLIR state-of-the-art claim is contradicted by the paper's own Table 3, the loss ablation in Table 5 contradicts the surrounding text, and the hypergraph formulation as written appears to be a k-NN graph, which undermines the core novelty. The absence of sensitivity analysis for the many hyperparameters further limits the strength of the generalization claims.

major comments (4)
  1. [Abstract and Section 4.2.3, Table 3] The abstract's claim of state-of-the-art performance on FLIR is not supported by Table 3: ProtoHGF-Net obtains 79.1% mAP50, while CMX obtains 82.2% and LRAF-Net obtains 80.5%. Section 4.2.3 itself concedes that the mAP50 is lower than LRAF-Net and CMX, then pivots to the mAP metric. The abstract and Section 4.2.3 must be revised to present the FLIR result as competitive rather than state-of-the-art on mAP50, or the metric in the abstract must be changed explicitly.
  2. [Section 4.3.2, Table 5] The prose states that 'using any single loss term alone improves upon the baseline,' but Table 5 shows that each single-loss row (84.9, 84.8, and 85.0 mAP50) is below the baseline of 85.1 mAP50. This is a direct factual contradiction. The text should be corrected to reflect that individual loss terms give performance close to or slightly below the baseline, while combinations yield the reported gains.
  3. [Section 3.2.2, Eqs. (2)-(5)] The matrix H is defined as a relation matrix whose entries indicate message passing from source prototype j to target prototype i, and Eq. (5) is described as 'vertex→hyperedge→vertex' propagation. As written, no hyperedge incidence matrix is defined; the operations in Eq. (5) are two graph propagations on a k-NN adjacency matrix, not hypergraph propagation. The terminology 'hypergraph' and the claim in Section 4.3.4 that gains stem from 'higher-order dependencies' are therefore not supported by the formulation. Please either define each row of H as a hyperedge and give the correct incidence interpretation, or revise the terminology and the claims about higher-order modeling.
  4. [Sections 3.2.1, 3.2.2, 3.3.3] No sensitivity analysis is reported for the key hyperparameters K=6, k_intra=3, k_cross=3, lambda_fg=1.0, lambda_bg=2.0, and lambda_orth=0.07. Given that the reported margins over prior methods are modest (e.g., +2.9 mAP50 over UAVD on DroneVehicle and +3.2 mAP50 over CMA on DVTOD), it is not clear that the improvements are robust to these choices rather than tuned per dataset. Please provide a small sweep or a clear justification that these values are not dataset-specific.
minor comments (6)
  1. [Section 4.2.2] The phrase 'Under the thermal-only setting, ProtoHGF-Net surpasses...' is misleading because ProtoHGF-Net is a multimodal method; the sentence should be reworded to 'Compared to thermal-only methods...'.
  2. [Table 8 and Section 4.4] The table column is labeled mAP50, but the text says 'highest mAP of 85.9%'. Please make the metric consistent between the table and the prose.
  3. [Abstract] There is a typo in the abstract: 'backgrou- nd' should be 'background'.
  4. [Section 3.2.2] The notation 'TopK(·, k)' is introduced with a generic k, but the text immediately uses k_intra and k_cross; please align the notation for clarity.
  5. [References] Reference [40] is cited for the Ultralytics framework; consider citing the original Ultralytics repository or documentation rather than a YOLOv11-specific arXiv paper.
  6. [Section 3.3.2, Eq. (11)] The region-weighting map W^m is used in the foreground alignment loss, but its exact construction is not defined; a sentence specifying how W^m is derived would improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's claims are empirical benchmark results measured against external datasets and prior methods, with self-citations used only as baselines.

full rationale

ProtoHGF-Net's central assertion is an empirical performance comparison on DroneVehicle, DVTOD, and FLIR. The proposed components (ProtoHG-Fusion, TM-Calib) are defined by their own equations (Eqs. 1-14) and are not defined in terms of the reported mAP values, nor are the reported numbers fitted parameters renamed as predictions. The cited works by the same authors, MGFF [5] and IGIANet [7], appear only as comparison baselines in Tables 1 and 8, not as load-bearing justifications for the framework or for the claimed results. No uniqueness theorem or prior design choice by the same authors is invoked to force the architecture. The fixed hyperparameters (K=6, k_intra=3, k_cross=3, loss weights) are asserted without sensitivity analysis, which is a robustness gap but not circularity. The paper does contain an internal factual inconsistency: the abstract claims state-of-the-art 79.1% mAP50 on FLIR, while Table 3 and Section 4.2.3 concede that this is below LRAF-Net (80.5%) and CMX (82.2%). That is a correctness/consistency problem, not a circular-reasoning problem, because the claim is contradicted by external baselines rather than derived from the paper's own assumptions. For DVTOD and DroneVehicle, the reported improvements are benchmark results against external methods and are not forced by construction. Therefore no circular step is exhibited, and the circularity score is 0.

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

The central claim is empirical; the method rests on several unverified design assumptions and hand-set hyperparameters. No new physical entities are introduced.

free parameters (4)
  • K (number of prototypes) = 6 (default)
    Hand-chosen default in Sec. 3.2.1; no sensitivity analysis. If K varies across datasets, the prototype compression may not be the cause of the gains.
  • k_intra (intra-modal neighbors) = 3
    Set in Sec. 3.2.2; controls sparsity of intra-modal connections. No ablation is reported for this value.
  • k_cross (cross-modal neighbors) = 3
    Set in Sec. 3.2.2; controls the number of cross-modal prototype links. No ablation is reported for this value.
  • lambda_fg, lambda_bg, lambda_orth = 1.0, 2.0, 0.07
    Loss weights in Eq. 14, default values in Sec. 3.3.3. Ablations show the combined weighting matters, but no sensitivity to the weights is reported.
assumptions (4)
  • domain assumption Pixel-level features can be compressed into K=6 semantic prototypes without losing target-relevant detection information.
    Introduced in Sec. 3.2.1; no information-theoretic or empirical guarantee, and no sensitivity to K.
  • domain assumption Top-k binary relations among prototypes provide a structural prior that reduces noise and over-smoothing better than soft relations.
    Sec. 3.2.2 and Table 7 support this empirically on two datasets, but the mechanism is asserted rather than demonstrated.
  • domain assumption Frozen unimodal teachers generate target-aware masks that reliably separate foreground from background for both modalities.
    Sec. 3.3.1, Eq. 9. If the teacher itself has background-dominant features, the masks propagate that bias into the student.
  • ad hoc to paper The block relation matrix H in Eq. 2 defines a hypergraph, and Eq. 5 is a vertex-hyperedge-vertex propagation.
    H is a binary top-k adjacency matrix, and Agg(H^T) then Agg(H) is ordinary two-step graph message passing. No hyperedge set is constructed, so the 'hypergraph' label is an asserted naming rather than a demonstrated structure.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ProtoHGF-Net: Prototype HyperGraph Fusion with Intra-modal Calibration for RGBT Object Detection." pith.science (2026). https://pith.science/paper/3PYZWZNH

@misc{pith2026260811595,
  author       = {Pith},
  title        = {Pith review of: ProtoHGF-Net: Prototype HyperGraph Fusion with Intra-modal Calibration for RGBT Object Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3PYZWZNH}},
  note         = {Machine review of arXiv:2608.11595}
}
abstract

RGB-Thermal (RGBT) object detection enables robust perception in complex scenes by leveraging the complementary strengths of visible textures and thermal cues. However, existing methods mainly rely on dense cross-modal interactions over full-resolution features, which inevitably introduce background interference and hinder the learning of target-relevant representations. In this paper, we propose the Prototype HyperGraph Fusion Network (ProtoHGF-Net), a novel framework that redefines cross-modal fusion as prototype-level semantic interaction rather than the dense cross-modal interaction paradigm. Specifically, we design Prototype HyperGraph Fusion to perform cross-modal interaction in a compact prototype-level semantic space. This design enables more selective fusion among target-relevant prototypes. To support this prototype-level fusion, we propose Teacher-Mask Calibration Distillation, which calibrates modality features before fusion using modality-specific teachers and target-aware masks. This strategy suppresses backgrou- nd-dominant responses and produces more target-focused features. Extensive experiments on DroneVehicle, DVTOD, and FLIR demonstrate that ProtoHGF-Net achieves state-of-the-art performance with 85.9\% $mAP_{50}$, 88.2\% $mAP_{50}$, and 79.1\% $mAP_{50}$, respectively. Our code is available at \href{https://github.com/ZiMo-Chen/ProtoHGF}{GitHub}.

Figures

Figures reproduced from arXiv: 2608.11595 by the authors.

Figure 1
Figure 1. Comparison of cross-modal fusion paradigms for [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. Overview of the ProtoHGF-Net framework, which integrates a ProtoHG-Fusion, and TM-Calib for RGBT object [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Comparison of different detection results, where [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 43 canonical work pages

  1. [1]

    Haowen Bai, Jiangshe Zhang, Zixiang Zhao, Yichen Wu, Lilun Deng, Yukun Cui, Tao Feng, and Shuang Xu. 2025. Task-driven Image Fusion with Learnable Fusion Loss. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 7457–7468

  2. [2]

    Weihan Cao, Yifan Zhang, Jianfei Gao, Anda Cheng, Ke Cheng, and Jian Cheng

  3. [3]

    Chen Chen, Jiahao Qi, Xingyue Liu, Kangcheng Bin, Ruigang Fu, Xikun Hu, and Ping Zhong. 2024. Weakly misalignment-free adaptive feature alignment for uavs- based multimodal object detection. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 26836–26845

  4. [4]

    Guobin Chen, Wongun Choi, Xiang Yu, Tony Han, and Manmohan Chandraker

  5. [5]

    Xiangqi Chen, Shan Jin, Li Zhao, Chengzhuan Yang, Dawei Zhang, Xiao Wang, Xiaowei He, Hua Wang, Zhongyu Chen, and Zhonglong Zheng. 2025. Mask- Guided Frequency Feature Fusion for Visible–Infrared Remote Sensing Object Detection.IEEE Transactions on Geoscience and Remote Sensing63 (2025), 1–15. doi:10.1109/TGRS.2025.3612495

  6. [6]

    Xiangqi Chen, Chengzhuan Yang, Jiashuaizi Mo, Yaxin Sun, Li Zhao, Hao Chen, and Zhonglong Zheng. 2025. A Cross-domain Feature Fusion Network for Night- time Drone-view Object Detection.Pattern Recognition(2025), 112635

  7. [7]

    Xiangqi Chen, Dawei Zhang, Li Zhao, Chengzhuan Yang, Zhongyu Chen, Jun- gang Lou, Zhonglong Zheng, Sang-Woon Jeon, and Hua Wang. 2026. IGIANet: Illumination Guided Implicit Alignment Network for Infrared–Visible UAV De- tection. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 40. 3056–3064

  8. [8]

    Zizhao Chen, Yeqiang Qian, Xiaoxiao Yang, Chunxiang Wang, and Ming Yang

Show all 60 references
  1. [9]

    Xiaolong Cheng, Keke Geng, Ziwei Wang, Jinhu Wang, Yuxiao Sun, and Pengbo Ding. 2023. SLBAF-Net: Super-Lightweight bimodal adaptive fusion network for UAV detection in low recognition environment.Multimedia Tools and Applications 82, 30 (2023), 47773–47792

  2. [10]

    Jian Ding, Nan Xue, Yang Long, Gui-Song Xia, and Qikai Lu. 2019. Learning RoI transformer for oriented object detection in aerial images. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2849–2858

  3. [11]

    Yihe Dong, Will Sawin, and Yoshua Bengio. 2020. Hnhn: Hypergraph networks with hyperedge neurons.arXiv preprint arXiv:2006.12278(2020)

  4. [12]

    Qingyun Fang, Dapeng Han, and Zhaokui Wang. 2021. Cross-modality fusion transformer for multispectral object detection.arXiv preprint arXiv:2111.00273 (2021)

  5. [13]

    Yifan Feng, Jiangang Huang, Shaoyi Du, Shihui Ying, Jun-Hai Yong, Yipeng Li, Guiguang Ding, Rongrong Ji, and Yue Gao. 2024. Hyper-yolo: When visual object detection meets hypergraph computation.IEEE Transactions on Pattern Analysis and Machine Intelligence47, 4 (2024), 2388–2401

  6. [14]

    Yifan Feng, Haoxuan You, Zizhao Zhang, Rongrong Ji, and Yue Gao. 2019. Hy- pergraph neural networks. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 33. 3558–3565

  7. [15]

    Haolong Fu, Shixun Wang, Puhong Duan, Changyan Xiao, Renwei Dian, Shutao Li, and Zhiyong Li. 2024. LRAF-Net: Long-Range Attention Fusion Network for Visible–Infrared Object Detection.IEEE Transactions on Neural Networks and Learning Systems35, 10 (2024), 13232–13245

  8. [16]

    Yue Gao, Yifan Feng, Shuyi Ji, and Rongrong Ji. 2022. Hgnn+: General hypergraph neural networks.IEEE Transactions on Pattern Analysis and Machine Intelligence 45, 3 (2022), 3181–3199

  9. [17]

    Zhinan Gao, Dongdong Li, Yangliu Kuai, Rui Chen, and Gongjian Wen. 2025. Visible-Infrared Image Alignment for UAVs: Benchmark and New Baseline.IEEE Transactions on Geoscience and Remote Sensing63 (2025), 1–14. doi:10.1109/TGRS. 2025.3528634

  10. [18]

    Jiaming Han, Jian Ding, Jie Li, and Gui-Song Xia. 2021. Align deep features for oriented object detection.IEEE Transactions on Geoscience and Remote Sensing60 (2021), 1–11

  11. [19]

    Yan Han, Peihao Wang, Souvik Kundu, Ying Ding, and Zhangyang Wang. 2023. Vision hgnn: An image is more than a graph of nodes. InProceedings of the IEEE/CVF International Conference on Computer Vision. 19878–19888

  12. [20]

    Xiao He, Chang Tang, Xin Zou, and Wei Zhang. 2023. Multispectral object detection via cross-modal conflict-aware learning. InProceedings of the ACM International Conference on Multimedia. 1465–1474

  13. [21]

    Jie Hu, Li Shen, and Gang Sun. 2018. Squeeze-and-excitation networks. InPro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 7132– 7141

  14. [22]

    Fushuo Huo, Wenchao Xu, Jingcai Guo, Haozhao Wang, and Song Guo. 2024. C2kd: Bridging the modality gap for cross-modal knowledge distillation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 16006–16015

  15. [23]

    Guyue Jin, Tianming Zhao, Jiacan Yan, and Tian Tian. 2025. Contextually-Guided State Space Fusion for Misaligned Multi-Spectral Object Detection. InProceedings of the ACM International Conference on Multimedia. 2526–2535

  16. [24]

    2020.Ultralytics YOLOv5

    Glenn Jocher. 2020.Ultralytics YOLOv5. doi:10.5281/zenodo.3908559

  17. [25]

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al

  18. [26]

    Quanquan Li, Shengying Jin, and Junjie Yan. 2017. Mimicking very efficient network for object detection. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 6356–6364

  19. [27]

    Ting Li, Songtao Li, Shuaifeng Li, Xiaolin Qin, Maoyuan Zhao, Luping Ji, and Mao Ye. 2025. SAM-Guided Semantic Knowledge Fusion for Visible-Infrared Object Detection. InProceedings of the ACM International Conference on Multimedia. 8835–8844

  20. [28]

    Wei Li, Jiaman Tang, Yang Li, Beihao Xia, Ligang Tan, and Hongmao Qin. 2025. UAVD-Mamba: Deformable Token Fusion Vision Mamba for Multimodal UAV Detection. InProceedings of the IEEE Intelligent Vehicles Symposium. 561–568. doi:10.1109/IV64158.2025.11097442

  21. [29]

    Jinyuan Liu, Bowei Zhang, Qingyun Mei, Xingyuan Li, Yang Zou, Zhiying Jiang, Long Ma, Risheng Liu, and Xin Fan. 2025. DCEvo: Discriminative Cross- Dimensional Evolutionary Learning for Infrared and Visible Image Fusion. In Proceedings of the Computer Vision and Pattern Recogni...

  22. [30]

    Kewei Liu, Tao Li, and Dongliang Peng. 2025. Aerial Image Object Detection Based on RGB-Infrared Multibranch Progressive Fusion.IEEE Transactions on Geoscience and Remote Sensing63 (2025), 1–14. doi:10.1109/TGRS.2025.3552787

  23. [31]

    Wenxuan Ma, Shuang Li, Lincan Cai, and Jingxuan Kang. 2024. Learning modality knowledge alignment for cross-modality transfer.arXiv preprint arXiv:2406.18864 (2024)

  24. [32]

    Yalong Pi, Nipun D Nath, and Amir H Behzadan. 2020. Convolutional neural networks for object detection in aerial imagery for disaster response and recovery. Advanced Engineering Informatics43 (2020), 101009

  25. [33]

    Liuxiang Qiu, Si Chen, Jing-Hao Xue, Da-Han Wang, Shunzhi Zhu, and Yan Yan

  26. [34]

    Herbert Robbins and Sutton Monro. 1951. A stochastic approximation method. The Annals of Mathematical Statistics(1951), 400–407

  27. [35]

    Jifeng Shen, Yifei Chen, Yue Liu, Xin Zuo, Heng Fan, and Wankou Yang. 2024. ICAFusion: Iterative cross-attention guided feature fusion for multispectral object detection.Pattern Recognition145 (2024), 109913

  28. [36]

    Changyong Shu, Yifan Liu, Jianfei Gao, Zheng Yan, and Chunhua Shen. 2021. Channel-wise knowledge distillation for dense prediction. InProceedings of the IEEE/CVF International Conference on Computer Vision. 5311–5320

  29. [37]

    Kechen Song, Xiaotong Xue, Hongwei Wen, Yingying Ji, Yunhui Yan, and Qing- gang Meng. 2024. Misaligned Visible-Thermal Object Detection: A Drone-Based Benchmark and Baseline.IEEE Transactions on Intelligent Vehicles9, 11 (2024), 7449–7460. doi:10.1109/TIV.2024.3398429

  30. [38]

    Yiming Sun, Bing Cao, Pengfei Zhu, and Qinghua Hu. 2022. Drone-Based RGB- Infrared Cross-Modality Vehicle Detection Via Uncertainty-Aware Learning.IEEE Transactions on Circuits and Systems for Video Technology32, 10 (2022), 6700–6713. doi:10.1109/TCSVT.2022.3168279

  31. [39]

    Xiaozhong Tong, Xiaojun Guo, Xiaoyong Sun, Runze Guo, Shaojing Su, and Zhen Zuo. 2025. CMDistill: Cross-Modal Distillation Framework for AAV Image Object Detection.IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing18 (2025), 1395–1409. doi:10.1109...

  32. [40]

    Dahang Wan, Rongsheng Lu, Yang Fang, Xianli Lang, Shuangbao Shu, Jingjing Chen, Siyuan Shen, Ting Xu, and Zecong Ye. 2025. YOLOv11-RGBT: Towards a Comprehensive Single-Stage Multispectral Object Detection Framework.arXiv preprint arXiv:2506.14696(2025). MM ’26, November 10–14,...

  33. [41]

    Ao Wang, Hui Chen, Lihao Liu, Kai Chen, Zijia Lin, Jungong Han, et al . 2024. Yolov10: Real-time end-to-end object detection.Advances in Neural Information Processing Systems37 (2024), 107984–108011

  34. [42]

    Xingxing Xie, Gong Cheng, Jiabao Wang, Xiwen Yao, and Junwei Han. 2021. Oriented R-CNN for object detection. InProceedings of the IEEE/CVF International Conference on Computer Vision. 3520–3529

  35. [43]

    Muhammad Yaseen. 2024. What is YOLOv8: An In-Depth Exploration of the Internal Features of the Next-Generation Object Detector.arXiv preprint arXiv:2408.15857(2024)

  36. [44]

    Maoxun Yuan, Yinyan Wang, and Xingxing Wei. 2022. Translation, scale and rota- tion: cross-modal alignment meets RGB-infrared vehicle detection. InProceedings of the European Conference on Computer Vision. Springer, 509–525

  37. [45]

    Maoxun Yuan and Xingxing Wei. 2024. C2former: Calibrated and complementary transformer for rgb-infrared object detection.IEEE Transactions on Geoscience and Remote Sensing62 (2024), 1–12

  38. [46]

    Heng Zhang, Elisa Fromont, Sébastien Lefevre, and Bruno Avignon. 2020. Multi- spectral Fusion for Object Detection with Cyclic Fuse-and-Refine Blocks. InIEEE International Conference on Image Processing. 276–280. doi:10.1109/ICIP40778. 2020.9191080

  39. [47]

    Jiaqing Zhang, Jie Lei, Weiying Xie, Zhenman Fang, Yunsong Li, and Qian Du

  40. [48]

    Jiaming Zhang, Huayao Liu, Kailun Yang, Xinxin Hu, Ruiping Liu, and Rainer Stiefelhagen. 2023. CMX: Cross-modal fusion for RGB-X semantic segmentation with transformers.IEEE Transactions on Intelligent Transportation Systems24, 12 (2023), 14679–14694

  41. [49]

    Lu Zhang, Zhiyong Liu, Shifeng Zhang, Xu Yang, Hong Qiao, Kaizhu Huang, and Amir Hussain. 2019. Cross-modality interactive attention network for multispec- tral pedestrian detection.Information Fusion50 (2019), 20–29

  42. [50]

    Tianyi Zhao, Boyang Liu, Yanglei Gao, Yiming Sun, Maoxun Yuan, and Xingxing Wei. 2025. Rethinking Multi-modal Object Detection from the Perspective of Mono-Modality Feature Learning. InProceedings of the IEEE/CVF International Conference on Computer Vision. 6364–6373

  43. [51]

    Zhicheng Zhao, Wei Zhang, Yun Xiao, Chenglong Li, and Jin Tang. 2025. Reflectance-Guided Progressive Feature Alignment Network for All-Day UAV Object Detection.IEEE Transactions on Geoscience and Remote Sensing63 (2025), 1–15. doi:10.1109/TGRS.2025.3574963

  44. [52]

    Minghang Zhou, Tianyu Li, Chaofan Qiao, Dongyu Xie, Guoqing Wang, Ningjuan Ruan, Lin Mei, Yang Yang, and Heng Tao Shen. 2025. DMM: Disparity-guided multispectral mamba for oriented object detection in remote sensing.IEEE Transactions on Geoscience and Remote Sensing(2025), 1–13

  45. [53]

    SuperYOLO: Super resolution assisted object detection in multimodal remote sensing imagery.IEEE Transactions on Geoscience and Remote Sensing61 (2023), 1–15

  46. [54]

    Liu Zongzhen, Luo Hui, Wang Zhixing, Wei Yuxing, Zuo Haorui, and Zhang Jianlin. 2025. Cross-modal Offset-guided Dynamic Alignment and Fusion for Weakly Aligned UAV Object Detection.arXiv preprint arXiv:2506.16737(2025)

  47. [59]

    Yue Zhou, Xue Yang, Gefan Zhang, Jiabao Wang, Yanyi Liu, Liping Hou, Xue Jiang, Xingzhao Liu, Junchi Yan, Chengqi Lyu, et al. 2022. Mmrotate: A rotated object detection benchmark using pytorch. InProceedings of the ACM International Conference on Multimedia. 7331–7334

  48. [2017]

    Advances in Neural Information Processing Systems30 (2017), 742–751

    Learning efficient object detection models with knowledge distillation. Advances in Neural Information Processing Systems30 (2017), 742–751

  49. [2022]

    Pkd: General distillation framework for object detectors via pearson corre- lation coefficient.Advances in Neural Information Processing Systems35 (2022), 15394–15406

  50. [2023]

    InProceedings of the IEEE/CVF International Conference on Computer Vision

    Segment anything. InProceedings of the IEEE/CVF International Conference on Computer Vision. 4015–4026

  51. [2025]

    doi:10.1109/TMM.2025.3604937

    AMFD: Distillation via Adaptive Multimodal Fusion for Multispectral Pedestrian Detection.IEEE Transactions on Multimedia27 (2025), 8298–8310. doi:10.1109/TMM.2025.3604937

  52. [2026]

    doi:10.1109/TCSVT.2025

    HOH-Net: High-Order Hierarchical Middle-Feature Learning Network for Visible-Infrared Person Re-Identification.IEEE Transactions on Circuits and Systems for Video Technology36, 2 (2026), 2607–2622. doi:10.1109/TCSVT.2025. 3609840

Pith tools

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