Pith. sign in

REVIEW 5 major objections 4 minor 54 references

DuET: Dual Incremental Object Detection via Exemplar-Free Task Arithmetic

T0 review · 5 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Merging task vectors on shared weights lets one detector learn new object classes in new domains, without storing old images.

desk verdict A new dual-incremental detection setting and a simple, detector-agnostic task-arithmetic solution that mostly holds up; the headline RAI gains are plausible but rest on an unreported reference denominator that needs to be disclosed. read the letter →

arxiv 2506.21260 v2 pith:L5M7OLOE submitted 2025-06-26 cs.CV

Munish Monga , Vishal Chudasama , Pankaj Wasnik , Biplab Banerjee This is my paper · ORCID
classification cs.CV
keywords dualincrementalobjectdetectiontaskarithmeticmodelmergingcatastrophicforgettingdomainshiftexemplar-freelearningretention-adaptabilityindexdirectionalconsistencyloss
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 proposes a new learning setting, Dual Incremental Object Detection (DuIOD), in which a detector must add new object categories and adapt to a new visual domain at every task, without access to previous training data. The authors argue that existing class-incremental and domain-incremental detectors each miss one half of this problem, and they introduce DuET, a task-arithmetic-based model merging framework to handle both at once. DuET splits a detector into shared parameters (backbone and neck) and task-specific parameters (detection head), then fuses old and new task vectors—the weight differences from a pretrained model—layer by layer. A per-layer retention factor decides how much old knowledge to keep versus how much new signal to adopt, and a Directional Consistency Loss penalizes weight updates that oppose earlier ones. If the paper is right, DuET delivers the first exemplar-free way to make a standard real-time detector (YOLO11, RT-DETR) continuously learn both classes and domains, with a combined score, the Retention-Adaptability Index, to measure retention and adaptation together.

What carries the argument

The machinery is task-vector arithmetic on shared weights. A task vector $\tau = \theta_{\mathrm{finetuned}} - \theta_{\mathrm{pretrained}}$ records how a pretrained model's weights must shift for one task; DuET stores just two such vectors at a time, $\tau_{\mathrm{old}}$ and $\tau_{\mathrm{curr}}$, and fuses them with layer-wise weights $\alpha_l$, $\beta_l$ chosen by a $p$-factor $p_l = (\|\tau_{\mathrm{old}}^l\| - \|\tau_{\mathrm{curr}}^l\|) / (\|\tau_{\mathrm{old}}^l + \tau_{\mathrm{curr}}^l\| + \epsilon)$, mapped through $\gamma \tanh$ to keep $\alpha$ in $[\alpha_{\mathrm{base}} - \gamma, \alpha_{\mathrm{base}} + \gamma]$. The Directional Consistency Loss (ReLU of the negative dot product of successive shared-weight updates) stabilizes the trajectory of these vectors. The Incremental Head concatenates task-specific head weights so that new classes get their own parameters rather than overwriting old ones.

What would settle it

Train reference models for every unseen-class domain pair using the same pretrained checkpoint, epochs, and data splits used for DuET, then recompute Avg GI and RAI for all methods on a shared backbone such as YOLO11n; if the reported +13.12% and +11.39% RAI margins shrink substantially or reverse, the adaptability gain is an artifact of the normalization or backbone choice rather than the merging algorithm.

Watch

Extended reading notes

Core claim

The central claim is that simultaneous class and domain incremental learning can be handled by merging task vectors on the shared part of a detector, without exemplars. In DuET the shared parameters after the current task are rebuilt as $\theta_{s0} + \alpha_l \tau_{\mathrm{old}} + \beta_l \tau_{\mathrm{curr}}$ per layer, where $\tau_{\mathrm{old}}$ is the cumulative task vector from previous phases and $\tau_{\mathrm{curr}}$ is the current phase's vector; $\alpha_l$ and $\beta_l$ are derived from the ratio of the vectors' $\ell^1$ norms through a tanh-scaled $p$-factor, with $\beta_l = 1 - \alpha_l$. The task-specific head parameters are simply concatenated across tasks. A Directional Consistency Loss adds a ReLU penalty on negative dot products between consecutive shared-weight updates, reducing sign conflicts during merging. On the Pascal Series (4 tasks) the method reports 89.30% average retention and a +13.12% RAI improvement over baselines, and on the Diverse Weather Series (3 tasks) 88.57% Avg RI and +11.39% RAI, across detectors including YOLO11 and RT-DETR.

Load-bearing premise

The reported adaptability advantage assumes that the reference mAP models used to normalize the Average Generalization Index are strong and fairly trained; the paper does not report their mAP values or training details, so weak reference models would inflate DuET's RAI advantage, and some headline comparisons also mix backbones across methods.

Editorial extensions

If this is right

  • A detector can be turned into an incremental one without changing its architecture; the same pipeline works for YOLO11-based and RT-DETR-based detectors and preserves real-time inference speed.
  • Retention stays high across multi-phase sequences: above 88% Avg RI on both the four-task Pascal sequence and the three-task weather sequence, compared with near-total forgetting for sequential fine-tuning.
  • The merged model generalizes to unseen class–domain pairs—for example, detecting VOC classes rendered in Clipart style and vice versa—which neither class-only nor domain-only incremental detectors provide.
  • The Retention-Adaptability Index offers a single number combining retention of old classes and adaptability to new ones, which could serve as a standard evaluation metric for this setting.
  • The method is exemplar-free and keeps memory footprint roughly constant by storing only two shared task vectors plus the pretrained weights, rather than a growing history of task vectors.

Reading between the lines

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

  • A testable consequence the authors do not pursue: the $p$-factor uses $\ell^1$ norms; variants using cosine similarity or per-channel statistics might shift the stability-plasticity trade-off, especially for extreme weather or style changes.
  • Because DuET requires a pretrained detector and a common initialization, it cannot start from scratch; extending the decomposition to a multi-stage pretrain-adapt loop would broaden its applicability.
  • The RAI metric's validity rests on the reference mAP models used to normalize the Average Generalization Index; publishing those reference values and their training protocols would make cross-paper comparisons meaningful.
  • The Incremental Head grows wider with each task; the paper does not study head-capacity saturation over long task sequences, which is a natural stress test for a ten- or twenty-task deployment.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 4 minor

Summary. The paper introduces DuIOD, an exemplar-free dual incremental object detection setting that combines class and domain shifts, and proposes DuET, a detector-agnostic task-arithmetic framework. DuET computes shared and task-specific parameters, merges shared task vectors through layer-wise retention and adaptation factors, concatenates task-specific head parameters in an Incremental Head, and adds a Directional Consistency Loss. A new metric, RAI, is defined as the average of Retention Index and Generalization Index. Experiments on Pascal Series and Diverse Weather Series report large RAI gains over sequential fine-tuning, LwF, ERD, LDB, and CL-DETR, with an ablation study showing each DuET component contributes.

Significance. If the empirical claims hold, DuET would be a meaningful contribution: it is the first task-merging approach for the combined dual incremental setting, it is detector-agnostic across YOLO11, RT-DETR, Deformable DETR, and ViTDet, and the ablations (Table 4, Appendix C) carefully demonstrate the role of each module and loss. The problem setting is timely and the core merging equations are simple and clearly presented. However, the central quantitative claims currently depend on an unreported reference denominator in the adaptability metric and on cross-backbone baseline comparisons, so the empirical significance is not yet established.

major comments (5)
  1. [Sec. 4.2 and Appendix A.2, Eq. (18)/(24)] The central RAI claim is not reproducible because the reference mAP values mAPref(Di[Cunseen]) used as denominators in Avg GI are never reported, and no training details (epochs, schedule, backbone, hyperparameters) are given for the reference models. Since RAI is the headline metric in the abstract and Tables 1 and 2, and several reference domains are small (Clipart has 500 training images, Watercolor/Comic 1000, Foggy Cityscapes 1829), weakly trained reference models would inflate Avg GI and make the reported +13.12% and +11.39% RAI improvements uninterpretable. The authors should add a table of all mAPref values and a precise description of how those models were trained, or replace the ratio with raw unseen-class mAP.
  2. [Tables 1 and 2 and Sec. 5.1] The main comparisons are confounded by backbone choice: DuET is evaluated on YOLO11n (2.58M trainable parameters) while LDB uses ViTDet (110.52M) and CL-DETR uses Deformable DETR (39.85M). Consequently, the headline superiority cannot be attributed to the method rather than the architecture. Appendix E.1 partially addresses this with same-backbone results in Tables S7-S14, but the main text should either present controlled comparisons on shared backbones or explicitly restrict the claimed improvement to the YOLO11n comparisons.
  3. [Sec. 3.4, Eq. (13) and Fig. 3] The Incremental Head concatenates task-specific parameters from past and current tasks, but the paper never specifies how the detection head output dimension is expanded for new classes, how new-class weights are initialized, or how class-ID alignment is maintained across tasks. This is essential for reproducibility, especially because YOLO11 head output channels depend on the cumulative class count, and the ablation in Table 4 attributes substantial gains to the Incremental Head.
  4. [Appendix C.2, Fig. S2] The key hyperparameters alpha_base=0.5, gamma=0.1, lambda_Distill=0.01 and lambda_DC=0.01 are selected by maximizing RAI on the Pascal Series and Diverse Weather Series, which are the same datasets used for the final reported results. No held-out validation split or nested selection procedure is described. This risks overfitting the evaluation protocol and makes the reported gains less trustworthy as a fair model-selection outcome.
  5. [Appendix E.1, Tables S7-S14] The statement that DuET outperforms CL-DETR and LDB on their own backbones with +5.97% and +11.98% RAI gains does not match the data. Averaging the seven paired two-phase and multi-phase experiments gives approximately +5.2 RAI for Deformable DETR and +8.4 RAI for ViTDet. Moreover, in Table S12, DuET on ViTDet has RAI 53.01 vs. LDB's 52.83 but Avg RI drops from 86.08 to 65.57, so the claim of consistent superiority across all experiments and backbones is not supported.
minor comments (4)
  1. [Sec. 3.5, Eq. (16)] The Directional Consistency Loss uses tau^(i)_{st-2}, but for t=2 this term is not defined in the main text; the supplementary should clarify that tau_{s0} = 0, since theta_{s0} - theta_{s0} = 0.
  2. [Table 1] The column header 'mAP@0.5%' should read 'mAP@0.5' or 'mAP@0.5 (%)' for consistency with the metric definition.
  3. [Fig. S5 caption] The caption reads 'Diverse Series' but should be 'Diverse Weather Series' to match the terminology used elsewhere.
  4. [Sec. 3.4, Eq. (10)-(11)] The clamping operation on delta_l is redundant because delta_l = gamma * tanh(p_l) already lies within [-gamma, gamma]; consider removing it or explaining why it is retained.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DuET's task-vector merging, losses, and RAI metric are evaluated from measured mAP against external baselines; the unreported reference mAP is a reproducibility issue, not a definitional shortcut.

full rationale

The paper's derivation chain is self-contained and does not reduce to its inputs by construction. The task vectors in Eqs. 5 and 8 are computed from real fine-tuning runs on external benchmark data, and the merged shared weights in Eq. 12 are a convex combination of those measured vectors with coefficients from layer-wise norms; this is standard model merging, not a self-referential definition. The losses in Eq. 15 combine the detector loss, a modified distillation loss, and the Directional Consistency Loss; the latter penalizes sign conflicts among consecutive shared-weight updates and is evaluated through ablations, so it is an empirical component rather than a tautology. The RAI metric (Eq. 19) combines Avg RI and Avg GI, both of which are ratios of measured mAP values. Although the paper does not report the reference mAP values used in Eq. 18, that is an unreported experimental detail and a reproducibility concern, not a circular step: mAP_ref is an independent normalization anchor trained on unseen classes, and DuET's scores are not derived from it. Hyperparameters such as alpha_base, gamma, and the loss weights are selected via sensitivity analysis on the same evaluation protocol, which is in-sample tuning but not a fitted quantity renamed as a prediction; the final tables report measured performance, not a forecast from the tuned parameters. The cross-backbone comparisons are questioned, but the paper's supplementary Appendix E.1 provides same-backbone comparisons, and this is a fairness/correctness issue rather than circularity. There are no self-citations used as load-bearing evidence and no imported uniqueness theorems. Overall, the central claims are empirical evaluations against external baselines, so no circularity score is warranted.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced. The framework's load-bearing items are the heuristic merging coefficients, the incremental head concatenation, and the RAI metric's reference-model normalization; these are justified experimentally rather than derived.

free parameters (4)
  • alpha_base (base scaling coefficient) = 0.5
    Controls the default balance between the old task vector and the new task vector in Eq. 11; tuned via sensitivity analysis on the evaluation datasets (Fig. S2a).
  • gamma (limiting factor) = 0.1
    Scales the tanh in Eq. 10 to bound per-layer adaptation; tuned via sensitivity analysis on the evaluation datasets (Fig. S2b).
  • lambda_Distill = 0.01
    Weight for the modified distillation loss in Eq. 15; tuned via sensitivity analysis on the evaluation datasets (Fig. S2c).
  • lambda_DC = 0.01
    Weight for the Directional Consistency Loss in Eq. 15; tuned via sensitivity analysis on the evaluation datasets (Fig. S2d).
assumptions (5)
  • domain assumption A pre-trained object detector is available and used to compute task vectors at every incremental task.
    Stated in the footnote to Eq. 8; standard for task-vector methods, but it is a prerequisite for the framework.
  • ad hoc to paper Model parameters can be decomposed into shared (backbone and neck) and task-specific (detection head) components, and this decomposition transfers across detectors.
    Used in Section 3.3 for YOLO11 and RT-DETR; no analysis of how well this captures class versus domain knowledge.
  • ad hoc to paper The p-factor based on L1 norms of task vectors is a valid per-layer indicator of retention versus adaptation need.
    Equation 9 and surrounding text call the factors 'empirically motivated'; no independent justification is provided.
  • ad hoc to paper Weighted linear interpolation of task vectors (Eq. 12) is an effective merging operation for object detectors.
    Core of the DuET Module; inherits task-arithmetic assumptions without detector-specific validation.
  • domain assumption Reference models for Avg GI, trained solely on each unseen class-domain pair, are strong enough to be fair normalizers.
    Appendix A.2 defines Avg GI without reporting reference mAPs or reference training budgets; weak references would inflate GI.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DuET: Dual Incremental Object Detection via Exemplar-Free Task Arithmetic." pith.science (2026). https://pith.science/paper/L5M7OLOE

@misc{pith2026250621260,
  author       = {Pith},
  title        = {Pith review of: DuET: Dual Incremental Object Detection via Exemplar-Free Task Arithmetic},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L5M7OLOE}},
  note         = {Machine review of arXiv:2506.21260}
}
read the original abstract

Real-world object detection systems, such as those in autonomous driving and surveillance, must continuously learn new object categories and simultaneously adapt to changing environmental conditions. Existing approaches, Class Incremental Object Detection (CIOD) and Domain Incremental Object Detection (DIOD) only address one aspect of this challenge. CIOD struggles in unseen domains, while DIOD suffers from catastrophic forgetting when learning new classes, limiting their real-world applicability. To overcome these limitations, we introduce Dual Incremental Object Detection (DuIOD), a more practical setting that simultaneously handles class and domain shifts in an exemplar-free manner. We propose DuET, a Task Arithmetic-based model merging framework that enables stable incremental learning while mitigating sign conflicts through a novel Directional Consistency Loss. Unlike prior methods, DuET is detector-agnostic, allowing models like YOLO11 and RT-DETR to function as real-time incremental object detectors. To comprehensively evaluate both retention and adaptation, we introduce the Retention-Adaptability Index (RAI), which combines the Average Retention Index (Avg RI) for catastrophic forgetting and the Average Generalization Index for domain adaptability into a common ground. Extensive experiments on the Pascal Series and Diverse Weather Series demonstrate DuET's effectiveness, achieving a +13.12% RAI improvement while preserving 89.3% Avg RI on the Pascal Series (4 tasks), as well as a +11.39% RAI improvement with 88.57% Avg RI on the Diverse Weather Series (3 tasks), outperforming existing methods.

Figures

Figures reproduced from arXiv: 2506.21260 by the authors.

Figure 1
Figure 1. Comparison of Class Incremental Object Detection (CIOD), Domain Incremental Object Detection (DIOD), and our proposed Dual Incremental Object Detection (DuIOD), where ✓ : the scenario is addressed, ✗ : the scenario is NOT addressed. CIOD (left) incrementally learns new categories (\protect \mathcal {C}_1 \rightarrow \mathcal {C}_2 ) in a fixed domain but struggles in unseen domains. DIOD (middle), adapts to new doma… view at source ↗
Figure 2
Figure 2. Catastrophic forgetting of \protect \mathcal {T}_1 classes in multi-phase experiments for various methods on the DuIOD task. The left plot illustrates catastrophic forgetting in Night Sunny [1:2] classes across three incremental tasks: Night Sunny [1:2] (\protect \mathcal {T}_1 ) \rightarrow Day￾time Sunny [3:4] (\protect \mathcal {T}_2 ) \rightarrow Daytime Foggy [5:7] (\protect \mathcal {T}_3 ) on the Diverse Weat… view at source ↗
Figure 3
Figure 3. Overview of the proposed DuET framework for exemplar-free Dual Incremental Object Detection (DuIOD). Left: To setup DuIOD task, we illustrate a sequence of three tasks \{\mathcal {T}_1,\,\mathcal {T}_2,\,\mathcal {T}_3\} , each introducing new object classes (e.g. \{\textit {car, truck}\}, \{\textit {person, bus}\}, \{\textit {rider, bike, motor}\} ) under different domain shifts (Night Sunny, Daytime Sunny, Daytime… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative comparisons on two-phase experiments for different methods on the DuIOD task. The top two rows display detection results on unseen classes: Clipart [1:10] & VOC [11:20], for: VOC [1:10] → Clipart [11:20] experiment. Similarly, the bottom two rows show detec…
Figure 5
Figure 5. Figure 5: Impact of \protect \mathcal {L}_{DC} on (a) improving RAI and (b) reducing sign conflicts across different TA approaches, evaluated on VOC [1:10] → Clipart [11:20] using YOLO11n with Incremental Head and sequential fine-tuning. (first row, [PITH_FULL_IMAGE:figures/ful…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 47 canonical work pages

  1. [1]

    Rodeo: Replay for online object detection

    Manoj Acharya, Tyler L Hayes, and Christopher Kanan. Rodeo: Replay for online object detection. arXiv preprint arXiv:2008.06439, 2020. 2

    arXiv 2008

  2. [2]

    Modeling missing annotations for incremental learning in object detection

    Fabio Cermelli, Antonino Geraci, Dario Fontanel, and Bar- bara Caputo. Modeling missing annotations for incremental learning in object detection. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 3700–3710, 2022. 2

    work page 2022

  3. [3]

    Incremental detection of remote sensing ob- jects with feature pyramid and knowledge distillation

    Jingzhou Chen, Shihao Wang, Ling Chen, Haibin Cai, and Yuntao Qian. Incremental detection of remote sensing ob- jects with feature pyramid and knowledge distillation. IEEE Transactions on Geoscience and Remote Sensing , 60:1–13,

    work page

  4. [4]

    A new knowledge distillation for incremental object detection

    Li Chen, Chunyan Yu, and Lvcai Chen. A new knowledge distillation for incremental object detection. In2019 Interna- tional Joint Conference on Neural Networks (IJCNN), pages 1–7. IEEE, 2019. 2, 6, 12

    work page 2019

  5. [5]

    The cityscapes dataset for semantic urban scene understanding

    Marius Cordts, Mohamed Omran, Sebastian Ramos, Timo Rehfeld, Markus Enzweiler, Rodrigo Benenson, Uwe Franke, Stefan Roth, and Bernt Schiele. The cityscapes dataset for semantic urban scene understanding. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 3213–3223, 2016. 6, 17

    work page 2016

  6. [6]

    Improving single domain-generalized object detection: A focus on diversification and alignment

    Muhammad Sohail Danish, Muhammad Haris Khan, Muhammad Akhtar Munir, M Saquib Sarfraz, and Mohsen Ali. Improving single domain-generalized object detection: A focus on diversification and alignment. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17732–17742, 2024. 6, 17

    work page 2024

  7. [7]

    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. International journal of computer vision, 88:303–338, 2010. 6, 17

    work page 2010

  8. [8]

    Overcoming catastrophic forgetting in incremental object detection via elastic response distillation

    Tao Feng, Mang Wang, and Hangjie Yuan. Overcoming catastrophic forgetting in incremental object detection via elastic response distillation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 9427–9436, 2022. 2, 3, 6, 7, 18, 19, 20, 21, 22

    work page 2022

Show all 54 references
  1. [9]

    Ow-detr: Open-world detection transformer

    Akshita Gupta, Sanath Narayan, KJ Joseph, Salman Khan, Fahad Shahbaz Khan, and Mubarak Shah. Ow-detr: Open-world detection transformer. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9235–9244, 2022. 1, 2, 3

    2022

  2. [10]

    An end-to- end architecture for class-incremental object detection with knowledge distillation

    Yu Hao, Yanwei Fu, Yu-Gang Jiang, and Qi Tian. An end-to- end architecture for class-incremental object detection with knowledge distillation. In 2019 IEEE International Con- ference on Multimedia and Expo (ICME), pages 1–6. IEEE,

    2019

  3. [11]

    Vehicle detection and tracking in ad- verse weather using a deep learning framework

    Mahmoud Hassaballah, Mourad A Kenk, Khan Muhammad, and Shervin Minaee. Vehicle detection and tracking in ad- verse weather using a deep learning framework. IEEE trans- actions on intelligent transportation systems , 22(7):4230– 4242, 2020. 6, 17

    2020

  4. [12]

    Emr-merging: Tuning- free high-performance model merging

    Chenyu Huang, Peng Ye, Tao Chen, Tong He, Xi- angyu Yue, and Wanli Ouyang. Emr-merging: Tuning- free high-performance model merging. arXiv preprint arXiv:2405.17461, 2024. 3, 5, 14, 16

    2024 arXiv

  5. [13]

    Editing models with task arithmetic

    Gabriel Ilharco, Marco Tulio Ribeiro, Mitchell Wortsman, Suchin Gururangan, Ludwig Schmidt, Hannaneh Hajishirzi, and Ali Farhadi. Editing models with task arithmetic. arXiv preprint arXiv:2212.04089, 2022. 2, 3, 4, 5, 14, 16

    2022 arXiv

  6. [14]

    Cross-domain weakly-supervised object de- tection through progressive domain adaptation

    Naoto Inoue, Ryosuke Furuta, Toshihiko Yamasaki, and Kiy- oharu Aizawa. Cross-domain weakly-supervised object de- tection through progressive domain adaptation. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 5001–5009, 2018. 6, 17

    2018

  7. [15]

    Purified distillation: Bridging do- main shift and category gap in incremental object detection

    Shilong Jia, Tingting Wu, Yingying Fang, Tieyong Zeng, Guixu Zhang, and Zhi Li. Purified distillation: Bridging do- main shift and category gap in incremental object detection. In Proceedings of the 32nd ACM International Conference on Multimedia, pages 1197–1205, 2024. 2, 3

    2024

  8. [16]

    Dataless knowledge fusion by merging weights of language models

    Xisen Jin, Xiang Ren, Daniel Preotiuc-Pietro, and Pengxiang Cheng. Dataless knowledge fusion by merging weights of language models. arXiv preprint arXiv:2212.09849, 2022. 3

    2022 arXiv

  9. [17]

    Ultralytics yolo11, 2024

    Glenn Jocher and Jing Qiu. Ultralytics yolo11, 2024. 2, 3, 4, 6, 7, 13, 14, 16, 20

    2024

  10. [18]

    Alleviat- ing catastrophic forgetting of incremental object detection via within-class and between-class knowledge distillation

    Mengxue Kang, Jinpeng Zhang, Jinming Zhang, Xiashuang Wang, Yang Chen, Zhe Ma, and Xuhui Huang. Alleviat- ing catastrophic forgetting of incremental object detection via within-class and between-class knowledge distillation. In Proceedings of the IEEE/CVF International Confere...

    2023

  11. [19]

    Sddgr: Stable diffusion-based deep generative replay for class incremental object detection

    Junsu Kim, Hoseong Cho, Jihyeon Kim, Yihalem Yimolal Tiruneh, and Seungryul Baek. Sddgr: Stable diffusion-based deep generative replay for class incremental object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 28772–2878...

    2024

  12. [20]

    Vlm-pl: Advanced pseudo labeling ap- proach for class incremental object detection via vision- language model

    Junsu Kim, Yunhoe Ku, Jihyeon Kim, Junuk Cha, and Se- ungryul Baek. Vlm-pl: Advanced pseudo labeling ap- proach for class incremental object detection via vision- language model. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition , pages 417...

    2024

  13. [21]

    Incremental multi-target domain adaptation for object detection with efficient domain transfer

    Madhu Kiran, Marco Pedersoli, Jose Dolz, Louis-Antoine Blais-Morin, Eric Granger, et al. Incremental multi-target domain adaptation for object detection with efficient domain transfer. Pattern Recognition, 129:108771, 2022. 6, 17

    2022

  14. [22]

    Rilod: Near real-time incre- mental learning for object detection at the edge

    Dawei Li, Serafettin Tasci, Shalini Ghosh, Jingwen Zhu, Junting Zhang, and Larry Heck. Rilod: Near real-time incre- mental learning for object detection at the edge. In Proceed- ings of the 4th ACM/IEEE Symposium on Edge Computing , pages 113–126, 2019. 2

    2019

  15. [23]

    Incre- mental learning of single-stage detectors with mining mem- ory neurons

    Wei Li, Qingbo Wu, Linfeng Xu, and Chao Shang. Incre- mental learning of single-stage detectors with mining mem- ory neurons. In 2018 IEEE 4th International Conference on Computer and Communications (ICCC) , pages 1981–1985. IEEE, 2018. 2

    2018

  16. [24]

    Exploring plain vision transformer backbones for object de- tection

    Yanghao Li, Hanzi Mao, Ross Girshick, and Kaiming He. Exploring plain vision transformer backbones for object de- tection. In European conference on computer vision , pages 280–296. Springer, 2022. 6, 7, 16, 20 9

    2022

  17. [25]

    Learning without forgetting

    Zhizhong Li and Derek Hoiem. Learning without forgetting. IEEE transactions on pattern analysis and machine intelli- gence, 40(12):2935–2947, 2017. 1, 2, 3, 6, 7, 13, 18, 19, 20, 21, 22

    2017

  18. [26]

    Multi-task incremental learn- ing for object detection

    Xialei Liu, Hao Yang, Avinash Ravichandran, Rahul Bhotika, and Stefano Soatto. Multi-task incremental learn- ing for object detection. arXiv preprint arXiv:2002.05347 ,

    2002 arXiv

  19. [27]

    Augmented box replay: Overcom- ing foreground shift for incremental object detection

    Yuyang Liu, Yang Cong, Dipam Goswami, Xialei Liu, and Joost van de Weijer. Augmented box replay: Overcom- ing foreground shift for incremental object detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 11367–11377, 2023. 2, 3, 17

    2023

  20. [28]

    Continual detection transformer for incremen- tal object detection

    Yaoyao Liu, Bernt Schiele, Andrea Vedaldi, and Christian Rupprecht. Continual detection transformer for incremen- tal object detection. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 23799–23808, 2023. 1, 2, 3, 6, 7, 16, 18, 19, 2...

    2023

  21. [29]

    Fixing weight decay regularization in adam

    Ilya Loshchilov, Frank Hutter, et al. Fixing weight decay regularization in adam. arXiv preprint arXiv:1711.05101, 5: 5, 2017. 16

    2017 arXiv

  22. [30]

    Detrs beat yolos on real-time object detection, 2023

    Wenyu Lv, Shangliang Xu, Yian Zhao, Guanzhong Wang, Jinman Wei, Cheng Cui, Yuning Du, Qingqing Dang, and Yi Liu. Detrs beat yolos on real-time object detection, 2023. 2, 3, 4, 7, 13, 14, 16

    2023

  23. [31]

    Magmax: Leveraging model merging for seamless continual learning

    Daniel Marczak, Bartłomiej Twardowski, Tomasz Trzci ´nski, and Sebastian Cygert. Magmax: Leveraging model merging for seamless continual learning. In European Conference on Computer Vision, pages 379–395. Springer, 2025. 3, 4, 8, 14, 16

    2025

  24. [32]

    Merging models with fisher-weighted averaging

    Michael S Matena and Colin A Raffel. Merging models with fisher-weighted averaging. Advances in Neural Information Processing Systems, 35:17703–17716, 2022. 3, 5, 14, 16

    2022

  25. [33]

    Continual object detection: a re- view of definitions, strategies, and challenges

    Angelo G Menezes, Gustavo de Moura, C ´ezanne Alves, and Andr´e CPLF de Carvalho. Continual object detection: a re- view of definitions, strategies, and challenges. Neural net- works, 161:476–493, 2023. 2, 3, 6, 12, 17

    2023

  26. [34]

    An efficient domain-incremental learning ap- proach to drive in all weather conditions

    M Jehanzeb Mirza, Marc Masana, Horst Possegger, and Horst Bischof. An efficient domain-incremental learning ap- proach to drive in all weather conditions. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3001–3011, 2022. 2

    2022

  27. [35]

    Libra r-cnn: Towards balanced learning for object detection

    Jiangmiao Pang, Kai Chen, Jianping Shi, Huajun Feng, Wanli Ouyang, and Dahua Lin. Libra r-cnn: Towards balanced learning for object detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 821–830, 2019. 2

    2019

  28. [36]

    Ver- satile incremental learning: Towards class and domain- agnostic incremental learning

    Min-Yeong Park, Jae-Ho Lee, and Gyeong-Moon Park. Ver- satile incremental learning: Towards class and domain- agnostic incremental learning. In European Conference on Computer Vision, pages 271–288. Springer, 2025. 1, 3

    2025

  29. [37]

    Faster ilod: In- cremental learning for object detectors based on faster rcnn

    Can Peng, Kun Zhao, and Brian C Lovell. Faster ilod: In- cremental learning for object detectors based on faster rcnn. Pattern recognition letters, 140:109–115, 2020. 2

    2020

  30. [38]

    Sid: incremental learning for anchor-free object de- tection via selective and inter-related distillation

    Can Peng, Kun Zhao, Sam Maksoud, Meng Li, and Brian C Lovell. Sid: incremental learning for anchor-free object de- tection via selective and inter-related distillation. Computer vision and image understanding, 210:103229, 2021. 2

    2021

  31. [39]

    Catastrophic forgetting, rehearsal and pseudorehearsal

    Anthony Robins. Catastrophic forgetting, rehearsal and pseudorehearsal. Connection Science, 7(2):123–146, 1995. 1

    1995

  32. [40]

    Continual learning strategy in one-stage ob- ject detection framework based on experience replay for au- tonomous driving vehicle

    Jeng-Lun Shieh, Qazi Mazhar ul Haq, Muhamad Amirul Haq, Said Karam, Peter Chondro, De-Qin Gao, and Shanq- Jang Ruan. Continual learning strategy in one-stage ob- ject detection framework based on experience replay for au- tonomous driving vehicle. Sensors, 20(23):6777, 2020. 2

    2020

  33. [41]

    Incremental learning of object detectors without catas- trophic forgetting

    Konstantin Shmelkov, Cordelia Schmid, and Karteek Ala- hari. Incremental learning of object detectors without catas- trophic forgetting. In Proceedings of the IEEE international conference on computer vision, pages 3400–3409, 2017. 1, 2, 3

    2017

  34. [42]

    Non-exemplar domain incremental object detection via learning domain bias

    Xiang Song, Yuhang He, Songlin Dong, and Yihong Gong. Non-exemplar domain incremental object detection via learning domain bias. In Proceedings of the AAAI Con- ference on Artificial Intelligence, pages 15056–15065, 2024. 1, 2, 3, 6, 7, 16, 17, 18, 19, 20, 21, 22

    2024

  35. [43]

    Non-exemplar domain incremental learning via cross-domain concept integration

    Qiang Wang, Yuhang He, Songlin Dong, Xinyuan Gao, Shaokun Wang, and Yihong Gong. Non-exemplar domain incremental learning via cross-domain concept integration. In European Conference on Computer Vision , pages 144–

  36. [44]

    S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning

    Yabin Wang, Zhiwu Huang, and Xiaopeng Hong. S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning. Advances in Neural Informa- tion Processing Systems, 35:5682–5695, 2022. 2

    2022

  37. [45]

    Single-domain generalized ob- ject detection in urban scene via cyclic-disentangled self- distillation

    Aming Wu and Cheng Deng. Single-domain generalized ob- ject detection in urban scene via cyclic-disentangled self- distillation. In Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition, pages 847–856,

  38. [46]

    Ties-merging: Resolving interference when merging models

    Prateek Yadav, Derek Tam, Leshem Choshen, Colin A Raf- fel, and Mohit Bansal. Ties-merging: Resolving interference when merging models. Advances in Neural Information Pro- cessing Systems, 36, 2024. 2, 3, 5

    2024

  39. [47]

    Rd-iod: Two-level residual-distillation-based triple- network for incremental object detection

    Dongbao Yang, Yu Zhou, Wei Shi, Dayan Wu, and Weiping Wang. Rd-iod: Two-level residual-distillation-based triple- network for incremental object detection. ACM Transactions on Multimedia Computing, Communications, and Applica- tions (TOMM), 18(1):1–23, 2022. 2

    2022

  40. [48]

    Multi-view correlation distillation for incremental object detection

    Dongbao Yang, Yu Zhou, Aoting Zhang, Xurui Sun, Dayan Wu, Weiping Wang, and Qixiang Ye. Multi-view correlation distillation for incremental object detection. Pattern Recog- nition, 131:108863, 2022. 6, 12

    2022

  41. [49]

    Bdd100k: A diverse driving dataset for heterogeneous multitask learning

    Fisher Yu, Haofeng Chen, Xin Wang, Wenqi Xian, Yingying Chen, Fangchen Liu, Vashisht Madhavan, and Trevor Dar- rell. Bdd100k: A diverse driving dataset for heterogeneous multitask learning. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition ...

    2020

  42. [50]

    Language models are super mario: Absorbing abilities from homologous models as a free lunch

    Le Yu, Bowen Yu, Haiyang Yu, Fei Huang, and Yongbin Li. Language models are super mario: Absorbing abilities from homologous models as a free lunch. In Forty-first Interna- tional Conference on Machine Learning, 2024. 2, 3 10

    2024

  43. [51]

    Incre- mental learning of object detection with output merging of compact expert detectors

    Nan Zhang, Zhigang Sun, Kai Zhang, and Li Xiao. Incre- mental learning of object detection with output merging of compact expert detectors. In 2021 4th international confer- ence on intelligent autonomous systems (ICoIAS), pages 1–7. IEEE, 2021. 2

    2021

  44. [52]

    Replay-based incremental object detection with local response exploration

    Jian Zhong, Yifan Jiao, and Bing-Kun Bao. Replay-based incremental object detection with local response exploration. IEEE Transactions on Multimedia, 2025. 2

    2025

  45. [53]

    Deformable detr: Deformable trans- formers for end-to-end object detection

    Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, and Jifeng Dai. Deformable detr: Deformable trans- formers for end-to-end object detection. arXiv preprint arXiv:2010.04159, 2020. 2, 3, 6, 7, 16, 20 11 DuET: Dual Incremental Object Detection via Exemplar-Free Task Arith...

    2010 arXiv

  46. [54]

    and RT-DETR [30] object detectors with L∗ Distill and LDC on the Ultralytics [17] pipeline. In the case of YOLO11, the detection loss consists of classification loss, bounding box regression loss, and Distribution Focal Loss 13 [17], while in the case of RT-DETR, the detection...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.