Cross-Attention-Driven Propose-and-Select Generative Data Augmentation for Few-Shot Image Classification
Sensors · 20.07.2026 · 26 (14) · 4590
DOI: 10.3390/s26144590 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 20.07.2026. Verfasst von Ying Liu und 2 weiteren Autor:innen. Beteiligte Einrichtung: Shenyang Institute of Computing Technology, Chinese Academy of Sciences, Shenyang 110168, China. Die Arbeit nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Development of Embodied Intelligence Human-Robot Collaborative System for Industrial Multi-Task Scenarios Program, Central Government Guidance Fund for Local Science and Technology Development.
Abstract (Originalsprache)
Generative data augmentation based on diffusion models has emerged as a promising approach for few-shot image classification. Existing methods, such as DA-Fusion, typically follow a “generate-once, use-directly” paradigm, which often suffers from uncontrollable generation quality, unstable semantic consistency, insufficient global diversity, and high sample redundancy. To address these limitations, we propose a two-stage Propose-and-Select framework for controllable data augmentation. This framework curates high-quality synthetic data offline, ensuring that no additional training overhead is introduced to downstream models. For selector optimization, our method eliminates the need for additional human annotations by leveraging the zero-shot prior knowledge of a vision–language model (CLIP) to construct relative-quality pseudo-labels. Furthermore, we develop an adaptive-temperature listwise ranking distillation objective to transfer quality-aware supervision effectively. We also introduce a multi-objective consistency regularization strategy to stabilize training and improve convergence. Under a strictly controlled augmentation budget, where all methods are provided with the same number of synthetic samples, the proposed approach consistently outperforms existing diffusion-based augmentation baselines across both few-shot classification benchmarks, achieving an accuracy of 79.58% on PASCAL VOC and 80.74% on the fine-grained Oxford 102 Flowers dataset. These results demonstrate the effectiveness of the proposed generation-selection paradigm in improving the quality, diversity, and semantic relevance of synthetic samples, thereby enhancing downstream few-shot classification performance.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Shenyang Institute of Computing Technology, Chinese Academy of Sciences, Shenyang 110168, China.
- University of Chinese Academy of Sciences, Beijing 100049, China.
- Shenyang CASNC Technology Co., Ltd., Shenyang 110168, China.
Förderung
- Development of Embodied Intelligence Human-Robot Collaborative System for Industrial Multi-Task Scenarios Program
- Central Government Guidance Fund for Local Science and Technology Development
Zitieren
APA
Liu, Y., Zheng, L., & Wang, S. (2026). Cross-Attention-Driven Propose-and-Select Generative Data Augmentation for Few-Shot Image Classification. Sensors, 26(14), 4590. https://doi.org/10.3390/s26144590
Vancouver
Liu Y, Zheng L, Wang S. Cross-Attention-Driven Propose-and-Select Generative Data Augmentation for Few-Shot Image Classification. Sensors. 2026;26(14):4590. doi:10.3390/s26144590
BibTeX
@article{Liu2026,
author = {Liu, Ying and Zheng, Liaomo and Wang, Shiyu},
title = {Cross-Attention-Driven Propose-and-Select Generative Data Augmentation for Few-Shot Image Classification},
journal = {Sensors},
year = {2026},
volume = {26},
number = {14},
pages = {4590},
doi = {10.3390/s26144590}
}
EUVIMED