Weakly Supervised Fine-Grained Image Classification via Guassian Mixture Model Oriented Discriminative Learning

Wang, Zhihui; Wang, Shijie; Shu-hui, Yang; Li, Haojie; Liu, Jianjun; Li, Zezhou

doi:10.1109/cvpr42600.2020.00977

Cited by 72 publications

(33 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using the same backbone, our method gains a clear margin on Aircraft (0.8%), Flowers (1.0%) and Pets (1.1%). We also achieve the SotA results compared to the others based on the attention mechanism [4,8,42,62], GCN [55,60,84], and low-rank discriminative bases [58]. Moreover, our approach gives superior performance over these existing methods (marked * in Table 1) without leveraging secondary datasets.…”

Section: Fine-grained Image Classificationmentioning

confidence: 79%

“…For a higher number of regions, it fails to learn spatial contextual information resulting in lower accuracy. Likewise, DFG [58] also suffers from the same scalability problem in their graph-structure. Our model is scalable to any number of regions without increasing the model parameters since the graph node's parameters are shared.…”

Section: Fine-grained Image Classificationmentioning

confidence: 99%

See 1 more Smart Citation

An attention-driven hierarchical multi-scale representation for visual recognition

Wharton¹,

Behera²,

Bera³

2021

Preprint

View full text Add to dashboard Cite

Convolutional Neural Networks (CNNs) have revolutionized the understanding of visual content. This is mainly due to their ability to break down an image into smaller pieces, extract multi-scale localized features and compose them to construct highly expressive representations for decision making. However, the convolution operation is unable to capture long-range dependencies such as arbitrary relations between pixels since it operates on a fixed-size window. Therefore, it may not be suitable for discriminating subtle changes (e.g. fine-grained visual recognition). To this end, our proposed method captures the high-level long-range dependencies by exploring Graph Convolutional Networks (GCNs), which aggregate information by establishing relationships among multiscale hierarchical regions. These regions consist of smaller (closer look) to larger (far look), and the dependency between regions is modeled by an innovative attention-driven message propagation, guided by the graph structure to emphasize the neighborhoods of a given region. Our approach is simple yet extremely effective in solving both the finegrained and generic visual classification problems. It outperforms the state-of-the-arts with a significant margin on three and is very competitive on other two datasets.

show abstract

Section: Fine-grained Image Classificationmentioning

confidence: 79%

Section: Fine-grained Image Classificationmentioning

confidence: 99%

An attention-driven hierarchical multi-scale representation for visual recognition

Wharton¹,

Behera²,

Bera³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We determine the merged local region L o as the detected object localization. It is shown that the larger δ h is, the more local regions are needed to cover the object, and the easier it is to obtain better lo- ResNet-50 87.3 --DFL-CNN [31] ResNet-50 × 87.4 92.0 93.8 NTS-Net [15] ResNet-50 × 87.5 91.4 93.9 TASN [17] ResNet-50 × 87.9 -93.8 DF-GMM [32] ResNet-50 calization performance. Under δ h = 0.6, the mIoU performance of n = 3 is lower than that of n = 2, which is consistent with the retrieval (in Table 4) and classification performances.…”

Section: Localizationmentioning

confidence: 99%

Searching and Learning Discriminative Regions for Fine-Grained Image Retrieval and Classification

Sun

Zhu

2022

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

Local discriminative regions play important roles in finegrained image analysis tasks. How to locate local discriminative regions with only category label and learn discriminative representation from these regions have been hot spots. In our work, we propose Searching Discriminative Regions (SDR) and Learning Discriminative Regions (LDR) method to search and learn local discriminative regions in images. The SDR method adopts attention mechanism to iteratively search for highresponse regions in images, and uses this as a clue to locate local discriminative regions. Moreover, the LDR method is proposed to learn compact within category and sparse between categories representation from the raw image and local images. Experimental results show that our proposed approach achieves excellent performance in both fine-grained image retrieval and classification tasks, which demonstrates its effectiveness.

show abstract

“…In real-world applications, it is infeasible to annotate a large amount of training data. Therefore, unsupervised representation learning has been widely studied in many computer vision tasks like image classification [1,13,14,20,38,26,48], image retrieval [21], and object detection [19,31,4,40]. Recently, self-supervised learning methods [2,15,18,30,36,41,54] were in favor for unsupervised pre-training tasks, where a contrastive loss was adopted to learn instance discriminative representations.…”

Section: Related Workmentioning

confidence: 99%

Refining Pseudo Labels with Clustering Consensus over Generations for Unsupervised Object Re-identification

Zhang¹,

Ge²,

Qiao³

et al. 2021

Preprint

View full text Add to dashboard Cite

Unsupervised object re-identification targets at learning discriminative representations for object retrieval without any annotations. 46,10] conduct training with the generated pseudo labels and currently dominate this research direction. However, they still suffer from the issue of pseudo label noise. To tackle the challenge, we propose to properly estimate pseudo label similarities between consecutive training generations with clustering consensus and refine pseudo labels with temporally propagated and ensembled pseudo labels. To the best of our knowledge, this is the first attempt to leverage the spirit of temporal ensembling [25] to improve classification with dynamically changing classes over generations. The proposed pseudo label refinery strategy is simple yet effective and can be seamlessly integrated into existing clustering-based unsupervised re-identification methods. With our proposed approach, state-of-the-art method [10] can be further boosted with up to 8.8% mAP improvements on the challenging MSMT17 [39] dataset. The code is released on https : / / github . com / 2han9x1a0release/RLCC.

show abstract

Weakly Supervised Fine-Grained Image Classification via Guassian Mixture Model Oriented Discriminative Learning

Cited by 72 publications

References 27 publications

An attention-driven hierarchical multi-scale representation for visual recognition

An attention-driven hierarchical multi-scale representation for visual recognition

Searching and Learning Discriminative Regions for Fine-Grained Image Retrieval and Classification

Refining Pseudo Labels with Clustering Consensus over Generations for Unsupervised Object Re-identification

Contact Info

Product

Resources

About