Few-Shot Class-Incremental Learning for Classification and Object Detection: A Survey
Open accessThe DOI matches its Crossref record; 119 of the 123 references it cites were found in a registry or an index, and 2 are not checked yet; 54 of the 100 sources its claims cite were checked for support.
Abstract
Few-shot Class-Incremental Learning (FSCIL) presents a unique challenge in Machine Learning (ML), as it necessitates the Incremental Learning (IL) of new classes from sparsely labeled training samples without forgetting previous knowledge. While this field has seen recent progress, it remains an active exploration area. This paper aims to provide a comprehensive and systematic review of FSCIL. In our in-depth examination, we delve into various facets of FSCIL, encompassing the problem definition, the discussion of the primary challenges of unreliable empirical risk minimization and the stability-plasticity dilemma, general schemes, and relevant problems of IL and Few-shot Learning (FSL). Besides, we offer an overview of benchmark datasets and evaluation metrics. Furthermore, we introduce the Few-shot Class-incremental Classification (FSCIC) methods from data-based, structure-based, and optimization-based approaches and the Few-shot Class-incremental Object Detection (FSCIOD) methods from anchor-free and anchor-based approaches. Beyond these, we present several promising research directions within FSCIL that merit further investigation.
Verified 2026-11-19 · 2 references not checked yet
Overview
CoChat Summary
TL;DR
This survey systematically reviews few-shot class-incremental learning (FSCIL) for classification and object detection, covering definitions, challenges, datasets, metrics, methods, and future directions. It organizes FSCIC methods into data-, structure-, and optimization-based approaches and FSCIOD methods into anchor-free and anchor-based approaches.
Key findings
- FSCIL’s core challenges are unreliable empirical risk minimization under N-way K-shot incremental sessions and the stability-plasticity dilemma; FSCIL is a subdomain of CIL but adds scarce incremental samples and sustained unreliable risk minimization.
- For FSCIC, top base-session accuracy on mini ImageNet includes SAVC (81.12%), ALICE (80.60%), BiDistFSCIL (74.65%), TEEN (73.53%), and CLOM (73.08%); on CIFAR-100, WaRP (80.31%), BiDistFSCIL (79.45%), ALICE (79.00%), SAVC (78.77%), and TEEN (74.92%); on CUB-200, SAVC (81.85%), F2M and FSIL-GAN (81.07%), DSN (80.86%), and BiDistFSCIL (79.12%).
- Lowest performance dropping (PD) on mini ImageNet: MgSvF (17.33%), SFbFSCIL (19.14%), SPPR (19.53%), ERL++ (20.94%), and DSN (21.06%); on CIFAR-100: F2M (20.04%), SFbFSCIL (20.31%), SPPR (20.65%), TEEN (22.28%), and FACT (22.50%); on CUB-200: ALICE (17.30%), DSN (17.65%), MgSvF (17.96%), TEEN (18.13%), and BiDistFSCIL (18.19%).
- For FSCIOD, top base-class methods are Sylph, iFS-RCNN, and MCH; top novel-COCO methods are iFS-RCNN, Incremental-DETR, and iMTFA; top overall COCO methods are iFS-RCNN, Sylph, and MCH; top cross-dataset VOC methods are Incremental-DETR, BPMCH, and MCH.
Methods
- Systematic literature survey of deep FSCIL in computer vision, focused on classification and object detection; covers problem definition, core challenges, general schemes, related IL/FSL problems, benchmark datasets, evaluation metrics, method taxonomy, performance comparisons, and future directions.
- Proposes taxonomy: FSCIC methods from data-based (data replay, pseudo scenarios), structure-based (dynamic structure, attention), and optimization-based (representation learning, knowledge distillation, meta-learning, other) approaches; FSCIOD methods from anchor-free (CenterNet-, FCOS-, DETR-based) and anchor-based (Mask R-CNN) frameworks.
- Performance comparison summarizes mainstream methods on mini ImageNet, CIFAR-100, and CUB-200 for classification and COCO/VOC for detection, extracting values from original papers and noting backbones and extra factors where possible.
Limitations
- The paper states FSCIL lacks publicly available datasets and that current metrics (AA, PD; mAP, mAR, mAP50) do not fully capture detailed performance throughout continuous learning.
- It states comparisons are biased by variability in backbone networks, additional data, and implementation conditions; not all methods are open-source, and performance values are extracted from original papers.
- It states the current FSCIC setting is idealistic versus real-world needs, and FSCIOD research is insufficient, with performance far from practical application and non-unified evaluation metrics. As a survey, it presents no new experiments; this is evident from the text.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
39 Valid · 16 Not valid · 27 Not checked · 134 Own results and interpretations
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Despite existing surveys on FSL [27, 28, 29, 30] and IL [5, 6, 7, 11, 31, 32, 33, 34, 35], there is a clear lack of systematic and comprehensive surveys specifically on FSCIL.”
What the sources say
- Not checked — no full text of the sourcecites [27] [27] Y. Wang, Q. Yao, J. T. Kwok, and L. M. Ni, 'Generalizing from a few examples: A survey on few-shot learning,' ACM Computing Surveys , vol. 53, no. 3, pp. 1-34, 2020. - source
- Not checked — no full text of the sourcecites [28] [28] J. Lu, P. Gong, J. Ye, J. Zhang, and C. Zhang, 'A survey on machine learning from few samples,' Pattern Recognition , vol. 139, p. 109480, 2023. - source
- Not checked — no full text of the sourcecites [29] [29] S. Antonelli, D. Avola, L. Cinque, D. Crisostomi, G. L. Foresti, F. Galasso, M. R. Marini, A. Mecca, and D. Pannone, 'Few-shot object detection: A survey,' ACM Computing Surveys , vol. 54, no. 11s, pp. 1-37, 2022. - source
- Supportedcites [30] [30] G. Huang, I. Laradji, D. Vazquez, S. Lacoste-Julien, and P. Rodriguez, 'A survey of self-supervised and few-shot object detection,' IEEE TPAMI , 2022. - source
The citations in the sentence are attached to the existence of earlier FSL and IL surveys, not to the citing authors’ conclusion about a lack of FSCIL surveys. This paper identifies itself as a survey of few-shot object detection in the Introduction and reviews that field in Section 4, so it supports its use as an example of an existing survey in a few-shot learning area. It does not establish the claimed FSCIL survey gap. (LLM, conf=0.86)
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“This approach doesn't ensure strong generalization on test data, especially with limited training samples.”
What the sources say
- Not checked — no full text of the sourcecites [27] [27] Y. Wang, Q. Yao, J. T. Kwok, and L. M. Ni, 'Generalizing from a few examples: A survey on few-shot learning,' ACM Computing Surveys , vol. 53, no. 3, pp. 1-34, 2020. - source
- Partly supportedcites [39] [39] Y. Song, T. Wang, P. Cai, S. K. Mondal, and J. P. Sahoo, 'A comprehensive survey of few-shot learning: Evolution, applications, challenges, and opportunities,' ACM Computing Surveys , 2023. - sourceExperimental — accuracy not yet measured
Section 8.1 says excellent generalization is difficult with very few samples, supporting the limited-data concern. The Introduction discusses minimizing empirical risk, but our copy does not specifically say that this approach fails to ensure generalization on test data. No page numbers are available in our copy. (LLM, conf=0.78)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Every source the paper cites for this claim was checked and supports it.
In the paper
“However, none of these works systematically review FSCIL.”
What the sources say
- Supportedcites [7] [7] M. Masana, X. Liu, B. Twardowski, M. Menta, A. D. Bagdanov, and J. van de Weijer, 'Class-incremental learning: Survey and performance evaluation on image classification,' IEEE TPAMI , pp. 1-20, 2022. - source
The paper identifies class-incremental learning as the subject of its survey (p. 2), and §2.3 (p. 3) narrows its evaluation to specified class-IL methods rather than few-shot class-incremental learning (FSCIL). It therefore supports the claim as applied to this cited survey, though it cannot establish the statement about every other work the citing sentence may refer to. (LLM, conf=0.84)
- Supportedcites [34] [34] D.-W. Zhou, Q.-W. Wang, Z.-H. Qi, H.-J. Ye, D.-C. Zhan, and Z. Liu, 'Class-incremental learning: A survey,' IEEE TPAMI , 2024. - source
The Introduction describes this paper as a comprehensive survey of class-incremental learning methods, while §2.1 defines FSCIL as a distinct few-shot setting. §5 mentions a few FSCIL works as a future-direction topic rather than systematically reviewing them. The cited paper therefore supports the claim as applied to this survey; it cannot establish the scope of the other works referred to as “these works.” (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“To mitigate catastrophic forgetting, some studies [44, 45, 46] use KD to maintain competent classification capabilities on previous classes while learning new ones.”
What the sources say
- Not checked — no full text of the sourcecites [44] [44] J. Zhu, G. Yao, W. Zhou, G. Zhang, W. Ping, and W. Zhang, 'Feature distribution distillation-based few shot class incremental learning,' in PRAI , 2022, pp. 108-113. - source
- Not checked — no full text of the sourcecites [45] [45] Y. Cui, W. Deng, X. Xu, Z. Liu, Z. Liu, M. Pietik¨ ainen, and L. Liu, 'Uncertainty-guided semi-supervised few-shot classincremental learning with knowledge distillation,' IEEE TMM , 2022. - source
- Supportedcites [46] [46] Y. Cui, W. Deng, H. Chen, and L. Liu, 'Uncertainty-aware distillation for semi-supervised few-shot class-incremental learning,' IEEE TNNLS , 2023. - source
The Introduction describes UaD-CE as preserving discriminative ability on base categories while obtaining knowledge of novel categories. The Ablation Studies section explicitly says the framework uses knowledge distillation for class-incremental learning and that it helps memorize old knowledge. (LLM, conf=0.93)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Although some surveys briefly mention FSCIL [32, 35], they only provide a short introduction to the concept and a few studies without offering a thorough or systematic analysis.”
What the sources say
- Could not verifycites [32] [32] E. Belouadah, A. Popescu, and I. Kanellos, 'A comprehensive study of class incremental learning algorithms for visual tasks,' Neural Networks , vol. 135, pp. 38-54, 2021. - sourceExperimental — accuracy not yet measured
The Introduction and Conclusion describe a thorough comparison of class-incremental learning algorithms, but do not characterize the paper's coverage of FSCIL. A search for “few-shot class-incremental learning” found a cited study in the References, not a passage introducing FSCIL or assessing FSCIL studies. Our copy therefore does not establish the specific claim that this survey gives FSCIL only a short introduction and a few studies. (LLM, conf=0.72)
- Supportedcites [35] [35] L. Wang, X. Zhang, H. Su, and J. Zhu, 'A comprehensive survey of continual learning: theory, method and application,' IEEE TPAMI , 2024. - source
Section 5.2, “Scarcity of Labeled Data” (page not provided), introduces FSCIL in one paragraph and briefly names representative methods. That passage supports the characterization of this survey’s FSCIL coverage as brief rather than a thorough, systematic analysis. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
56 more claims · 10 with their cited sources checked
Create a free account to see all 216 claims and to verify the 10 references still open.
- Probably not a substantial claim. 127
- It may state two separate claims as one. 74
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2.1 Problem Definition · p. 3
“| FSCIL | 2023 | Neural Networks | A Survey on Few-Shot Class-Incremental Learning [36]”
§ REFERENCES · p. 19
“- [36] S. Tian, L. Li, W. Li, H. Ran, X. Ning, and P. Tiwari, 'A survey on fewshot class-incremental learning,' Neural Networks , vol. 169, pp. 307324, 2024.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 3.2.1 Evaluation Metrics for Classification · p. 7
“PD = A B - A 0”
§ 4.4.1 Performance Comparison · p. 14
“| SAVC [96] | CVPR 23 | 18 | 81.12 | 57.11 | 67.05 | 24.01”
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 1×
“as demonstrated by AlexNet [1] and ResNet [2], to the combined strategy of supervised pre-training and fine-tuning, with Transformer-based BERT [3] being a prime example.”
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404 · cited 1×
“Therefore, this paper proposed a correction model based on Transformer [75].”
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Exploiting Generative AI to Scale up Intelligent Tutoring Systems
Jakubův, J., Chvalovský, K., Goertzel, Z., Kaliszyk, C., Olšák, M., Piotrowski, B., Schulz, S., Suda, M., & Urban, J. (2023) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itp.2023.19 · Cited by 76,658
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Microsoft COCO: Common Objects in Context
Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., & Zitnick, C. L. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10602-1_48 · Cited by 45,460 · cited 1×
“In FSCIL, the prevalent data partitioning method by Tao et al. [13] 2) Datasets for Object Detection: COCO: The Microsoft Common Objects in Context (COCO) dataset, widely used for object detection tasks, comprises 80 object classes including people, animals, vehicles, furniture, and food [61].”
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380 · cited 2×
“The core structure of this framework included the followingcomponents:First,featureextractorsobtainedthrough supervisedtrainingonbase-classdataandself-supervisedtasks onImageNet[58]orOpenImages-v6[87]usingmethodssuch aspretexttasks,contrastiveloss,orclustering.”
He, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.322 · Cited by 29,137
Research works printed but not linked: 3
- [80] D.-Y. Kim, D.-J. Han, J. Seo, and J. Moon, 'Warping the space: Weight space rotation for class-incremental few-shot learning,' in ICLR , 2023. -
- [84] N. Sankaran, 'Feature fusion for deep representations,' Ph.D. dissertation, State University of New York at Buffalo, 2021. -
- [88] R. M. Neal, Bayesian learning for neural networks . Springer Science & Business Media, 2012, vol. 118. -
Cited by
Raza, A., Hanif, F., & Mohammed, H. A. (2025) · Scientific Reports · doi:10.1038/s41598-025-26645-2 · Cited by 57
“Transformer-based detectors (Deformable DETR, RT-DETR, PVT backbones) demonstrably improve small-object recall and occlusion robustness and can reach competitive speed–accuracy trade-offs when engineered for throughput (RT-DETR reports AP ∼ 53.1% on COCO and high FPS on T4). tasks They are particularly valuable for de…”
InfluentialYang, Y., Qu, J., Dong, W., Huang, L., & Li, Y. (2026) · IEEE Transactions on Image Processing · doi:10.1109/tip.2025.3650395 · Cited by 22
“Existing FSCIL methods can be classified into five categories: 1) feature-based methods; 2) meta-learning methods; 3) replay-based methods; 4) dynamic and prompt-based methods; and 5) regularization-based methods [30].”
Liu, Y., Jiang, T., Li, R., Yuan, L., Grzegorzek, M., Li, C., & Li, X. (2025) · Frontiers in Medicine · doi:10.3389/fmed.2025.1551894 · Cited by 16
Scale-Dependent Performance Analysis of YOLO26 and YOLOv11 for PPE Detection
Yavuz, B. Ç. (2026) · Electronics · doi:10.3390/electronics15061146 · Cited by 15
WP-FSCIL: A Well-Prepared Few-Shot Class-Incremental Learning Framework for Pill Recognition
Zhang, J., Li, C., Cristani, M., Sun, H., Grzegorzek, M., & Chen, H. (2025) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2025.3548691 · Cited by 10
“…which measures the absolute decline in accuracy from the first session to the last, calculated as PD = A 0 − A N , where A 0 represents the classification accuracy in the base session, and A N represents the accuracy in the final session [41]; and 3) The Average Accuracy (AA) across all sessions.”
Self-citationLiu, B., Chang, H., Li, P., Chang, H., Wang, X., & He, W. (2025) · Frontiers in Physiology · doi:10.3389/fphys.2025.1729997 · Cited by 5
“Furthermore, exploring the integration of our framework with an FSCIL paradigm could be a promising step towards building a truly forward-compatible system capable of seamlessly incorporating new diagnostic criteria or patient cohorts without catastrophic forgetting (Zhang et al., 2025c).”
A Few-Shot Domain Class Incremental Motor Fault Diagnosis Method With Dynamic Anchor Prototyping
He, Z., Gao, Y., Shao, H., et al. (2025) · IEEE Sensors Journal · doi:10.1109/jsen.2025.3563850 · Cited by 5
“The core task of domain-incremental few-shot learning focuses on achieving efficient generalization of few-shot samples under new working conditions through a series of strategies [17].”
A review of speaker verification: Methods, network architectures, tasks and challenges
Wang, W., Zhao, H., Yang, Y., & Yang, Y. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.112351 · Cited by 8
A Few-Shot Class Incremental Learning Method Using Graph Neural Networks
Ma, Y., Liu, Y., & Du, B. (2026) · IEEE Transactions on Image Processing · doi:10.1109/tip.2026.3657170 · Cited by 7
“Few-Shot Class Incremental Learning (FSCIL), as an important research direction in this field, endeavors to enable models to learn new categories with limited training samples and avoid forgetting previously learned categories, that is, to solve the stability-plasticity dilemma [3], [4], [5], [6].”
Krassnig, P. J., & Gruber, D. P. (2026) · Journal of Intelligent Manufacturing · doi:10.1007/s10845-025-02778-z · Cited by 14
Xiang, H., Gu, J., Xiang, J., et al. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.115769 · Cited by 3
An effective global structure-aware feature aggregation network for multi-modal medical clustering
Guan, R., Quan, H., Li, D., & Hu, D. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.126835 · Cited by 6
Self-citationAdaptive exploration for few-shot incremental learning
Han, C., Gu, Z., Xu, C., & Cui, Z. (2025) · Knowledge-Based Systems · doi:10.1016/j.knosys.2025.114496 · Cited by 2
Qin, F., Yu, C., Qin, Y., Qiu, N., & Meng, F. (2025) · Advanced Engineering Informatics · doi:10.1016/j.aei.2025.104137 · Cited by 3
FS2-DETR: Transformer-Based Few-Shot Sonar Object Detection with Enhanced Feature Perception
Yang, S., Zhang, X., & Tan, P. (2026) · Journal of Marine Science and Engineering · doi:10.3390/jmse14030304 · Cited by 3
“In response, numerous few-shot object detection frameworks have been proposed in recent years [10,11].”
Zhou, Y., Yuan, M., Shi, D., et al. (2026) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2026.3662865 · Cited by 2
“The strategy of reducing the dependence of the model on data is a few-shot learning (FSL) method represented by metalearning and metric learning methods [6], which are used for few-shot classification (FSC) in the image field.”
Self-citationZhu, S., Ji, L., Chen, S., Duan, W., Ye, M., & Ren, H. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3664876 · Cited by 6
“The emergence of incremental few-shot learning (IFSL) [14], which addresses the challenge of learning new tasks with limited data while alleviating catastrophic forgetting of previously acquired knowledge, has motivated researchers to explore its application in object detection [15].”
Semi-Supervised Open-Set Through-Wall User Identification via MIMO Radar Respiratory Signals
Yang, K. W., Han, C., Xia, Y., et al. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3676024 · Cited by 2
“This problem relates to the emerging field of class incremental learning [63].”
Yang, Y., Li, M., & Huang, F. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.128927 · Cited by 2
Li, Z., Xia, S., Bao, W., Wang, S., Yue, J., & Fang, L. (2025) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2025.3601529 · Cited by 6
“These approaches can be broadly categorized into data replay-based methods, distillation-based methods, dynamic network structure-based methods, metalearning-based methods, foundation model-based methods, and feature space-based methods [31], [35].”
Mi, K., Kang, G., Li, G., Zhao, L., Zhou, T., & Gong, C. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3755184 · Cited by 1
“The model needs to classify all seen classes during testing without access to the task ID. CIL methods generally rely on labeled data, which is often limited due to the high cost of data annotation in real-world scenarios [29, 31, 44, 45].”
Chen, R., Zheng, X., Cao, H., Niu, T., Yu, S., Zhang, F., & Ji, S. (2025) · Electric Power Systems Research · doi:10.1016/j.epsr.2025.112679 · Cited by 1
Cui, S., Xiong, X., Yang, X., He, J., & Shen, T. (2026) · Frontiers in Oncology · doi:10.3389/fonc.2025.1699430 · Cited by 1
“Recently, the advancement of deep learning (DL) has led to major advances in computer-aided diagnosis (CAD) in the fi eld of medical imaging (11), with extensive applications across diverse tasks including few-shot learning, histopathology analysis, and biometric recognition (12 – 15).”
Personalized Federated Learning with Hierarchical Two-Branch Aggregation for Few-Shot Scenarios
Miao, Y., Zhang, W., Wang, Y., Liu, Y., Zhang, Z., Meng, L., & Zhang, B. (2026) · Sensors · doi:10.3390/s26031037 · Cited by 2
“Moreover, comprehensive surveys on few-shot class-incremental learning provide critical insights into adapting to new classes with limited data [19].”
RICO: Two Realistic Benchmarks and an In-Depth Analysis for Incremental Learning in Object Detection
Neuwirth-Trapp, M., Bieshaar, M., Paudel, D. P., & Goo, L. V. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00543 · Cited by 6
“For class-incremental learning (CIL) in object detection, where each task introduces a new class, most studies split the Pascal VOC 2007 [23] or MS COCO [53] datasets by classes [19– 21, 56, 61, 78, 80, 88, 94, 100, 106].”
Lu, Y., Lu, Y., Wang, Z., Wu, W., Ke, C., & Liu, S. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3674767 · Cited by 2
“More recently, generative and memory-augmented approaches have been explored to further mitigate performance degradation caused by extreme data scarcity [11].”
Liu, H., Chen, C. L. P., & Zhang, T. (2026) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2026.3674085 · Cited by 1
“This perspective allows the problem to be interpreted as a form of incremental FSL [46], in which a new subject can effectively leverage semantically similar instances corresponding to the same emotion from previously observed subjects.”
Lin, P., Ma, Z., Zhang, J. M., et al. (2026) · Solar Energy · doi:10.1016/j.solener.2026.114524 · Cited by 1
Self-citationLightweight Semantic-Guided FCOS for In-Line Micro-Defect Inspection in Semiconductor Manufacturing
Zhang, T., Yan, S., & Qin, G. (2026) · Micromachines · doi:10.3390/mi17040473 · Cited by 1
Wang, Y., Liu, L., & Feng, G. (2026) · Displays · doi:10.1016/j.displa.2026.103484 · Cited by 1
Self-citationShi, W., Yan, X., Yuan, J., et al. (2026) · Neural Networks · doi:10.1016/j.neunet.2026.109070 · Cited by 1
Confusion distillation for Continual Self-Supervised Learning
Liu, G., Zhou, Q., Fu, Z., & Wang, Z. (2026) · Pattern Recognition · doi:10.1016/j.patcog.2026.114071 · Cited by 1
Hierarchical context enhancement for long-tail entity retrieval augmented generation
Peng, Y., & Pan, K. (2026) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2026.1861536 · Cited by 1
Qu, C., Chen, D., Zhang, Y., et al. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5348781
YOLO-CM: Class-Aware Instance Segmentation Using Combine-Mask
Wu, R., Wen, X.-B., Liao, S., Li, J., Liu, L., & Kui, X. (2025) · Lecture notes in computer science · doi:10.1007/978-981-96-9891-2_35
Self-citationAskri, N., Attal, F., Chibani, A., et al. (2025) · 2022 IEEE 18th International Conference on Automation Science and Engineering (CASE) · doi:10.1109/case58245.2025.11163742
“Incremental learning techniques have recently emerged to handle concept drift and annotation scarcity [26].”
Yang, L., Tan, Y., Liang, L., Xu, H., Wu, T., Huang, Z., Li, X., & Tang, Y. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.130911 · Cited by 2
A Fair Federated Graph Learning Algorithm Based on Gradient and Prototype Contributions
Li, Y., Li, Z., Zhang, Y., et al. (2025) · IEEE International Conference on Information Systems and Computer Aided Education (ICISCAE) · doi:10.1109/iciscae66104.2025.11307285
“In computer vision, prototypes have been used to form compact class representations for few-shot learning and object detection [23].”
Novel Jamming Pattern Class Discovery via Meta-Metric in Few-Shot Scenarios
Liang, F., & Gao, Y. L. (2026) · IEEE Communications Letters · doi:10.1109/lcomm.2026.3650880
“When adversaries integrate AI-driven strategies with conventional countermeasures [1], they can generate adaptive and previously unseen jamming patterns [2].”
Wang, J., Wang, R., Hu, L., & Chen, M. (2025) · International Conference on Mobility, Sensing and Networking (MSN) · doi:10.1109/msn69125.2025.00063
“This approach enables models to incrementally acquire new knowledge while preserving existing information, leveraging resource-efficient techniques such as knowledge distillation and experience replay [8].”
Zhou, J., Zhang, Y., Xu, S., Hu, Y., Zeng, J., & Yin, Y. (2026) · Frontiers in Microbiology · doi:10.3389/fmicb.2026.1770528
Incremental Object Detection with Prompt-Based Methods
Neuwirth-Trapp, M., Bieshaar, M., Paudel, D. P., & Gool, L. V. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00544 · Cited by 4
“We study the problem of domain incremental object detection, where a model is exposed to a sequence of tasks, i.e . domains [46, 47, 58]”
Kim, S., Hwang, D., Woo, S., & Singh, R. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00546 · Cited by 5
“Specifically, the FSCIL task consists of a base session with sufficient training data, followed by multiple incremental sessions where an extremely limited number of samples are provided [42].”
Liu, D., Zhang, Y., Shi, Z., Zhang, Y., & Ling, F. (2026) · Optics & Laser Technology · doi:10.1016/j.optlastec.2026.115040 · Cited by 1
Hu, K., Wen, L., Zhang, T., & Zhang, H. (2026) · Scientific Reports · doi:10.1038/s41598-026-42970-6 · Cited by 1
“Meanwhile, several works investigate continual learning stability from complementary perspectives, such as unlearning accumulated data bias 55 , improving robustness to class arrival order 56 , and addressing few-shot incremental scenarios 57 .”
A Prototype-Based Personalized and Fair Federated Graph Learning Algorithm
Li, Y., Li, Z., Zhang, Y., et al. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11401582
“In computer vision, prototypes have been introduced to construct compact class representations, supporting applications such as few-shot learning and object detection [19].”
The effect of speech representations on EEG-based auditory attention detection
Hu, H., Liao, Y., Cai, S., & Li, H. (2026) · Pattern Recognition Letters · doi:10.1016/j.patrec.2026.03.005
Dual-graph knowledge distillation for few-shot class-incremental microorganism recognition
Xu, S., Hu, Y., Zhang, Y., Chen, L., & Yin, Y. (2026) · Frontiers in Microbiology · doi:10.3389/fmicb.2026.1791871
Li, M., Zhang, J., Zhang, Y., Shan, Z., Yang, J., Beheshti, A., & Qi, Y. (2026) · Frontiers in Microbiology · doi:10.3389/fmicb.2026.1783160
“In the Tribonema (G008) and Diversicornis (G039) categories, DMM and MDPMFN achieved slight leads, yet the performance gap remained narrow, with our model maintaining high detection accuracy within 3.1% and 0.1% of the top values, respectively.”
InfluentialSelf-citationBai, X., Zhang, J., Jiang, Z., Jiang, Z., & Liu, H. (2026) · Frontiers in Microbiology · doi:10.3389/fmicb.2026.1795951
“Early rule-based and classical machine-learning approaches relied on hand-crafted rules or TF-IDF features combined with SVMs; however, these methods failed to scale effectively to the extensive ICD label space (Zhang et al., 2025a; Perotte et al., 2014).”
Self-citationDSG: Dual-Semantic Guidance from LLM to Token Distillation for Few-Shot Incremental Learning
Tao, Z., Shen, T., Liu, J., Wang, K., & Wang, K. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11465116
Huang, X., Wang, S., Li, P., Zhang, R., & Yuan, Z. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11464832
Xiao, Y., Xiang, H., Xu, X., et al. (2026) · ACM Web Conference (WWW) · doi:10.1145/3774904.3792442
“Despite this progress, current methods still face two major limitations that affect their real-world applicability [37].”
Class-incremental few-shot underwater object detection framework
Zhao, X., Zhou, B., & Wang, Y. (2026) · Scientific Reports · doi:10.1038/s41598-026-53216-w
“By leveraging a small number of annotated samples from novel classes, it enables the detection and localization of novel classes while maintaining identification ability on previously learned classes, without relying on the full set of base class data 14 .”
Xu, H., Li, H., & Zhao, J. (2026) · Frontiers in Forests and Global Change · doi:10.3389/ffgc.2026.1846683
SA-LoRA: Shared-A decoupled low-rank adaptation for class-incremental learning
Bing, X., Xi, Y., & Yahyapour, R. (2026) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-026-00925-x
Efficient and practical purchase recognition system for unmanned vending machines
Kim, G.-H., Kim, E.-S., Ha, J.-H., Kim, C.-G., & Ko, K. (2026) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2026.104871
UAD-ICL: Uncertainty-aware semantic control for trustworthy latent context in-context learning
Zhang, J., Hu, Y., Liu, T., Liu, B., Zhang, Z., & Liu, H. (2026) · Pattern Recognition · doi:10.1016/j.patcog.2026.114241
Self-citationChen, S., Song, X., Mao, J., Huang, Y., Yang, A., Shan, R., Gao, H., & Su, D. (2026) · Remote Sensing · doi:10.3390/rs18132128
Improving Alignment for Few-Shot Class-Incremental Learning via Prompt-Tuned Vision Transformer
Xu, L., Qian, Y., Hou, X., Geng, K., & Huang, G. (2026) · Displays · doi:10.1016/j.displa.2026.103604
Efficient multi-scale feature fusion for lightweight remote sensing image recognition
Chen, H.-Z., Ni, X., Fan, B., Guo, B.-S., Ma, X.-X., & Sun, W.-C. (2026) · The Visual Computer · doi:10.1007/s00371-026-04635-3 · Cited by 1
“Existing few-shot learning methods [26, 55, 56] can provide valuable references for the subsequent improvement of LRRN. Future work will focus on enhancing the model’s recognition capability in few-shot scenarios, further improving remote sensing image recognition accuracy, and boosting the…”
FbRA: Frequency-band Reconstruction and Adaptation for Few-Shot Class-Incremental Learning
Yang, S., & Yin, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3420-2_3
IEM-DETR: Integrating and Enhancing Multi-Level Features for Limited-Data Object Detection
Jin, Y. Z., Zheng, B., Wu, M., & Zhu, S. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3410-3_31
Zhang, S., Tian, F., Yao, H., Shi, J., & Yan, Y. (2026) · Journal of Marine Science and Engineering · doi:10.3390/jmse14141324
Chen, P., Hui, X., Guo, L., Li, S., Wang, J., Liu, F., & Jiao, L. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7163011
A Novel Remote Sensing Benchmark for Few-Shot Class-Incremental Learning Based on NWPU-RESISC45
Zhou, J., Sun, S., Zhang, R., Xu, S., & Yin, Y. (2026) · Smart innovation, systems and technologies · doi:10.1007/978-981-92-0607-0_15
Brain-inspired synergistic-memory prompt tuning for few-shot class-incremental learning
Chen, J., Xu, F., Lyu, X., Li, X., Chen, S., Huang, Y., & Tang, H. (2026) · Information Fusion · doi:10.1016/j.inffus.2026.104680
HiCPS: Hierarchical complementary prompt synergy for few-shot class-incremental learning
Huang, Q., Wu, S., Wan, S., Lv, G., Wang, X., & Lu, H. (2026) · Neurocomputing · doi:10.1016/j.neucom.2026.134790
Wang, K., Chen, J., Cai, S., et al. (2026) · IEEE Annual Computers, Software, and Applications Conference (COMPSAC) · doi:10.1109/compsac69091.2026.00477
“Class-incremental learning (CIL) provides a key solution by learning new categories sequentially while retaining old knowledge [7].”
Discriminative Prototype Refinement with Semantic-Aware Margin for Few-Shot Object DetectionPreprint
Chen, Y., Mao, S., & Yang, Y. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7349259
Xu, X., Li, L., Jiang, S., Li, Y., & Li, B. (2026) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2026.1911412
“Combined with the incremental learning paradigm of small sample classes, the diagnostic model can effectively solidify the memory of the basic features of the old domain while absorbing the knowledge flow of the new domain (Zhang J. et al., 2025).”
Jiang, Y., Xie, Z., Liu, W., Huang, L., Yang, F., & Liu, X. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37196-6_10
MetaMalDet: A few-shot class-incremental meta-learning framework for malware detection
Zhang, Y., Zhang, X., Ma, M., Bai, H., & Zhang, Y. (2026) · Journal of Network and Computer Applications · doi:10.1016/j.jnca.2026.104579
Li, S., Guo, D., Yang, Y. X., & Zhao, H. (2026) · International Journal of Fatigue · doi:10.1016/j.ijfatigue.2026.110015
A prototype-aligned multi-scale attention network for few-shot industrial surface defect detection
Huang, R., & Zhao, L. (2026) · PLoS ONE · doi:10.1371/journal.pone.0359400
Hu, G., Song, Y., & Mohsin, M. F. M. (2026) · Earth Science Informatics · doi:10.1007/s12145-026-02247-y
Qin, F., Yu, C., Qin, Y., Qiu, N., & Meng, F. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.134595
Zhu, L., Dong, H., Zhang, Y., Zhang, Y., Xinyi, 黄. H., & Wang, H. (2026) · Frontiers in Physiology · doi:10.3389/fphys.2026.1967219
Metadata
- Authors
- Zhang, J., Liu, L., Silvén, O., Pietikäinen, M., & Hu, D.
- Affiliations
- National University of Defense Technology (CN)
- CMVS, University of Oulu, Oulu, Finland
- University of Oulu (FI)
- Venue
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2025-04
- Volume
- 47
- Issue
- 4
- Pages
- 2924-2945
- ISSN
- 0162-8828 (ISSN-L)
- 2160-9292
- 1939-3539
- OpenAlex
- W4406354592
- Semantic Scholar
- 7dbb2b74102b7bf25c2a616e68340129d6775d3a
- Cited by
- 84
- References
- 123
- Citations by year
- 2026: 58 · 2025: 24
- FWCI
- 32.63
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Natural Science Foundation of China (62036013, 62376283)
- Science and Technology Innovation Program of Hunan Province (2024QK2006)
- Key Stone (JS2023-03)
- National University of Defense Technology
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Keywords
- few-shot class-incremental learning, class-incremental learning, few-shot learning, object detection, classification, catastrophic forgetting, stability-plasticity dilemma, benchmark dataset, evaluation metrics, anchor-free detection, anchor-based detection
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.