Fine-grained few shot learning with foreground object transformation

Wang, Chaofei; Song, Shiji; Yang, Qisen; Li, Xiang; Huang, Gao

doi:10.1016/j.neucom.2021.09.016

Cited by 16 publications

(10 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A few generic few-shot learning literatures explored the performance of their algorithms on FSFG settings [11,23,24]. However, the nature of fine-grained domain is hardly addressed except for some recently published specialized approaches [46,47,48,49,50,51].…”

Section: Few-shot Fine-grained Visual Categorizationmentioning

confidence: 99%

“…Deep data-augmentation based method have also been proposed in FSFG [50] which achieve significant improvement in generalization performance, but with exorbitant computation cost. And in theory, augmentation-based techniques can work in tandem with other techniques including ours but the engineering required to bind them together is not explored here.…”

Section: Few-shot Fine-grained Visual Categorizationmentioning

confidence: 99%

“…• Exp.B [24]: A experimental setting recently well received in the community [48,50,51]. The three datasets are split into a meta-training set, a validation-set and a meta-testing set by category and evaluated respectively.…”

Section: Datasets and Evaluation Protocolsmentioning

confidence: 99%

See 2 more Smart Citations

T2L: Trans-transfer Learning for few-shot fine-grained visual categorization with extended adaptation

Sun

Yang

2023

Knowledge-Based Systems

View full text Add to dashboard Cite

Section: Few-shot Fine-grained Visual Categorizationmentioning

confidence: 99%

Section: Few-shot Fine-grained Visual Categorizationmentioning

confidence: 99%

See 1 more Smart Citation

T2L: Trans-transfer Learning for few-shot fine-grained visual categorization with extended adaptation

Sun

Yang

2023

Knowledge-Based Systems

View full text Add to dashboard Cite

“…Then they further employed a new low-rank bilinear pool operation [14], and designed a feature alignment layer to match the support features with the query features. Meanwhile, FOT [25] is a data enhancement method, which uses the posture transformation generator to generate additional samples of new sub-categories. Zhu et al [12] proposed a multi-attention meta learning method that exploited the attention mechanism of base learner and task learner to capture the different parts of an image.…”

Section: The Related Workmentioning

confidence: 99%

Few-shot Fine-grained Image Classification via Multi-Frequency Neighborhood and Double-cross Modulation

Zhu¹,

Gao²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Traditional fine-grained image classification typically relies on large-scale training samples with annotated ground-truth. However, some sub-categories may have few available samples in real-world applications. In this paper, we propose a novel few-shot fine-grained image classification network (FicNet) using multi-frequency Neighborhood (MFN) and double-cross modulation (DCM). Module MFN is adopted to capture the information in spatial domain and frequency domain. Then, the self-similarity and multi-frequency components are extracted to produce multi-frequency structural representation. DCM employs bi-crisscross component and double 3D crossattention components to modulate the embedding process by considering global context information and subtle relationship between categories, respectively. The comprehensive experiments on three fine-grained benchmark datasets for two few-shot tasks verify that FicNet has excellent performance compared to the state-of-the-art methods. Especially, the experiments on two datasets, "Caltech-UCSD Birds" and "Stanford Cars", can obtain classification accuracy 93.17% and 95.36%, respectively. They are even higher than that the general fine-grained image classification methods can achieve.

show abstract

“…In contrast, humans can identify new classes with only few labeled examples. Recently, some studies [25,31] focus on a more challenging setting, which aims to recognize fine-grained images from few samples, and is called fine-grained few-shot learning(FG-FSL). Learning from fine-grained images with few samples brings two challenges.…”

Section: Introductionmentioning

confidence: 99%

A Task-Aware Dual Similarity Network for Fine-Grained Few-Shot Learning

Qi¹,

Sun²,

Liu³

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

The goal of fine-grained few-shot learning is to recognize sub-categories under the same super-category by learning few labeled samples. Most of the recent approaches adopt a single similarity measure, that is, global or local measure alone. However, for fine-grained images with high intra-class variance and low inter-class variance, exploring global invariant features and discriminative local details is quite essential. In this paper, we propose a Task-aware Dual Similarity Network(TDSNet), which applies global features and local patches to achieve better performance. Specifically, a local feature enhancement module is adopted to activate the features with strong discriminability. Besides, task-aware attention exploits the important patches among the entire task. Finally, both the class prototypes obtained by global features and discriminative local patches are employed for prediction. Extensive experiments on three fine-grained datasets demonstrate that the proposed TDSNet achieves competitive performance by comparing with other stateof-the-art algorithms.

show abstract

Fine-grained few shot learning with foreground object transformation

Cited by 16 publications

References 14 publications

T2L: Trans-transfer Learning for few-shot fine-grained visual categorization with extended adaptation

T2L: Trans-transfer Learning for few-shot fine-grained visual categorization with extended adaptation

Few-shot Fine-grained Image Classification via Multi-Frequency Neighborhood and Double-cross Modulation

A Task-Aware Dual Similarity Network for Fine-Grained Few-Shot Learning

Contact Info

Product

Resources

About