Research and Empirical Analysis of Zero-Shot Image Classification Methods

Wei Chen

doi:10.47297/taposatWSP2633-456906.20260704

PDF
导出
分享
收藏
专辑

Research and Empirical Analysis of Zero-Shot Image Classification Methods

7卷第4期

浏览量: 0 Downloads: 2

Research and Empirical Analysis of Zero-Shot Image Classification Methods
Research and Empirical Analysis of Zero-Shot Image Classification Methods
Digital and Information Technology Research Institute, Chinalco Zhengzhou Nonferrous Metals Research Institute Co., Ltd., Zhengzhou, Henan, 450000, China
2026年7卷第4期页码：32-36
纸质出版：2026-04-30
DOI：10.47297/taposatWSP2633-456906.20260704
Accepted：

Scan QR Code

Wei Chen. (2026). Research and Empirical Analysis of Zero-Shot Image Classification Methods. 科技理论与实践（英文版）, 7(4), 32-36.

Wei Chen.Research and Empirical Analysis of Zero-Shot Image Classification Methods[J].Theory and Practice of Science and Technology,2026,07(04):32-36.
Wei Chen. (2026). Research and Empirical Analysis of Zero-Shot Image Classification Methods. 科技理论与实践（英文版）, 7(4), 32-36. DOI： 10.47297/taposatWSP2633-456906.20260704.

Wei Chen.Research and Empirical Analysis of Zero-Shot Image Classification Methods[J].Theory and Practice of Science and Technology,2026,07(04):32-36. DOI： 10.47297/taposatWSP2633-456906.20260704.

摘要

Abstract

With the widespread application of artificial intelligence in the field of image recognition

traditional supervised learning faces the bottleneck of data scarcity when processing tasks involving unseen categories. Zero-Shot Learning (ZSL)

as an emerging learning paradigm

has become a research hotspot by introducing semantic information to enable models to recognize unknown categories. This paper focuses on the problem of zero-shot image classification and proposes a deep model that integrates visual feature extraction

semantic representation construction

and cross-modal alignment mechanisms. The model not only combines attribute vectors and text embeddings but also incorporates knowledge graph information to construct a richer semantic space. Meanwhile

contrastive learning and generative mechanisms are introduced to improve the model's generalization ability to unseen categories. Experiments were conducted on several standard datasets (AwA2

CUB

SUN)

and the results indicate that the proposed model achieves excellent performance in both conventional ZSL and generalized ZSL (GZSL) tasks. Finally

this paper analyzes the limitations of the model and looks forward to future research directions such as multimodal fusion

graph-based modeling

and large-scale model transfer.

关键词

Keywords

references

ANWAAR M U, PAN Z, KLEINSTEUBER M. On leveraging variational graph embeddings for open world compositional zero-shot learning[J]. Proceedings of the 30th ACM International Conference on Multimedia, 2022, 2022(1): 4645-4654.

MISRA I, GUPTA A, HEBERT M. From red wine to red tomato: Composition with context[J]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, 2017(1): 1792-1801.

XU G Y, KORDJAMSHIDI P, CHAI J. Prompting large pre-trained vision-language models for compositional concept learning[J ] . arXiv preprint arXiv: 2211.05077 , 2022.

RADFORD A, KIM J W, HALLACY C, et al. Learning transferable visual models from natural language supervision[J]. International Conference on Machine Learning, 2021, 2021(1): 8748-8763.

ATZMON Y, KREUK F, SHALIT U, et al. A causal view of compositional zero-shot recognition[J]. Advances in Neural Information Processing Systems, 2020, 33(1): 1462-1473.

BAO W T, CHEN L C, HUANG H, et al. Prompting language-informed distribution for compositional zero-shot learning[J]. Proceedings of the European Conference on Computer Vision, 2024, 2024(1).

CHEN C F R, FAN Q F, PANDA R. Crossvit: Cross-attention multi-scale vision transformer for image classification[J]. Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, 2021(1): 357-366.

CHEN S F, GE C J, TONG Z, et al. Adaptformer: Adapting vision transformers for scalable visual recognition[J]. Advances in Neural Information Processing Systems, 2022, 35(1): 16664-16678.

FU J L, ZHENG H L, MEI T. Look closer to see better: Recurrent attention convolutional neural network for fine-grained image recognition[J]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, 2017(1): 4438-4446.

ISOLA P, LIM J J, ADELSON E H. Discovering states and transformations in image collections[J]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015, 2015(1): 1383-1391.

HU X M, WANG Z L. A dynamic learning method towards realistic compositional zero-shot learning[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2024, 38(3): 2265-2273.