Fine-Grained Visual Categorization by Localizing Object Parts With Single Image

Zheng, Xiangtao; Qi, Lei; Ren, Yutao; Lu, Xiaoqiang

doi:10.1109/tmm.2020.2993960

Cited by 20 publications

(7 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In contrast, Cross-X [45] proposed a cross-layer fusion method based on the information correlation of different convolutional layers, and used a one-squeeze multi-excitation (OSME) module to generate multi-attention feature and a cross-semantic regularizer to group maps and fuse attention block features with similar semantics. Among them, the OSME module is a module proposed by [46] to capture different positions or different attention feature maps and is used in the network of multiline feature processing. Similarly, considering the diversity of feature map information in different convolutional layers, localizing object parts (LOP) [47] performed spectral clustering on the information of multiple convolutional layers to enhance the features and obtains more accurate and effective objects.…”

Section: Feature Fusion and Enhancement Methodsmentioning

confidence: 99%

A survey of fine-grained visual categorization based on deep learning

Xie

Gong

Luan

et al. 2023

J. of Syst. Eng. Electron.

View full text Add to dashboard Cite

Deep learning has achieved excellent results in various tasks in the field of computer vision, especially in fine-grained visual categorization. It aims to distinguish the subordinate categories of the label-level categories. Due to high intra-class variances and high inter-class similarity, the fine-grained visual categorization is extremely challenging. This paper first briefly introduces and analyzes the related public datasets. After that, some of the latest methods are reviewed. Based on the feature types, the feature processing methods, and the overall structure used in the model, we divide them into three types of methods: methods based on general convolutional neural network (CNN) and strong supervision of parts, methods based on single feature processing, and methods based on multiple feature processing. Most methods of the first type have a relatively simple structure, which is the result of the initial research. The methods of the other two types include models that have special structures and training processes,which are helpful to obtain discriminative features. We conduct a specific analysis on several methods with high accuracy on public datasets. In addition, we support that the focus of the future research is to solve the demand of existing methods for the large amount of the data and the computing power. In terms of technology, the extraction of the subtle feature information with the burgeoning vision transformer (ViT) network is also an important research direction.

show abstract

Section: Feature Fusion and Enhancement Methodsmentioning

confidence: 99%

A survey of fine-grained visual categorization based on deep learning

Xie

Gong

Luan

et al. 2023

J. of Syst. Eng. Electron.

View full text Add to dashboard Cite

show abstract

“…• LOP [57]: Localizing the key object parts within each image only depending on a single image to avoid the influence of diversity between parts in different images.…”

Section: Methodsmentioning

confidence: 99%

“…Backbone Resolution Accuracy(%) MaxEnt [44] DenseNet-161 -84.9 WARN [45] WRN 224 × 224 85.6 DVAN [4] VGG-16 224 × 224 87.1 NTS-NET [18] ResNet-50 448 × 448 87.5 Cross-X [46] ResNet-50 448 × 448 87.7 CIN [51] ResNet-101 448 × 448 88.1 LAFE [48] ResNet-101 448 × 448 88.1 CDL [49] ResNet-50 448 × 448 88.4 AP-CNN [55] ResNet-50 448 × 448 88.4 DB [26] ResNet-50 448 × 448 88.6 LOP [57] ResNet-50 224 × 224 88.9 FDL [20] DenseNet-161 448 × 448 89.1 CSC-Net [47] ResNet-50 224 × 224 89.2 SCAPNet [56] ResNet-50 224 × 224 89.5 BAEM [58] DenseNet-161 448 × 448 89.5 PMG [52] ResNet-50 550 × 550 89.6 GaRD [53] ResNet-50 448 × 448 89.6 SnapMix [54] ResNet-101 448 × 448 89.6 API-NET [27] DenseNet-161 512 × 512 90.0 CPM [16] GoogLeNet over 800 90. 2) Stanford Dog: As can be seen from Table IV, our method shows more performance improvement on the Stanford Dog dataset, which is 2.2% higher than the current state-ofthe-art method API-NET [27] without using the contrastive learning mechanism and the high-resolution input.…”

Section: Methodsmentioning

confidence: 99%

Transformer with Peak Suppression and Knowledge Guidance for Fine-grained Image Recognition

Liu,

Wang,

Han

2021

Preprint

View full text Add to dashboard Cite

Fine-grained image recognition is challenging because discriminative clues are usually fragmented, whether from a single image or multiple images. Despite their significant improvements, the majority of existing methods still focus on the most discriminative parts from a single image, ignoring informative details in other regions and lacking consideration of clues from other associated images. In this paper, we analyze the difficulties of fine-grained image recognition from a new perspective and propose a transformer architecture with the peak suppression module and knowledge guidance module, which respects the diversification of discriminative features in a single image and the aggregation of discriminative clues among multiple images. Specifically, the peak suppression module first utilizes a linear projection to convert the input image into sequential tokens. It then blocks the token based on the attention response generated by the transformer encoder. This module penalizes the attention to the most discriminative parts in the feature learning process, therefore, enhancing the information exploitation of the neglected regions. The knowledge guidance module compares the image-based representation generated from the peak suppression module with the learnable knowledge embedding set to obtain the knowledge response coefficients. Afterwards, it formalizes the knowledge learning as a classification problem using response coefficients as the classification scores. Knowledge embeddings and image-based representations are updated during training simultaneously so that the knowledge embedding includes a large number of discriminative clues for different images of the same category. Finally, we incorporate the acquired knowledge embeddings into the image-based representations as comprehensive representations, leading to significantly higher recognition performance. Extensive evaluations on the six popular datasets demonstrate the advantage of the proposed method in performance.

show abstract

“…The image feature learning part aims to learn the image semantic feature ϕ I . First, following the previous works [51,52], the input training image sample I tr is preprocessed as a standard normalization I tr norm . Second, the standardized training image sample I tr norm is input into five stacked residual blocks with two fully connected (FC) layers to obtain the initial image semantic feature.…”

Section: Image Feature Learningmentioning

confidence: 99%

Audio description from image by modal translation network

et al. 2021

Self Cite

View full text Add to dashboard Cite

Audio is the main form for the visually impaired to obtain information. In reality, all kinds of visual data always exist, but audio data does not exist in many cases. In order to help the visually impaired people to better perceive the information around them, an image-to-audio-description (I2AD) task is proposed to generate audio descriptions from images in this paper. To complete this totally new task, a modal translation network (MT-Net) from visual to auditory sense is proposed. The proposed MT-Net includes three progressive sub-networks: 1) feature learning, 2) cross-modal mapping, and 3) audio generation. First, the feature learning sub-network aims to learn seman-

show abstract

Fine-Grained Visual Categorization by Localizing Object Parts With Single Image

Cited by 20 publications

References 51 publications

A survey of fine-grained visual categorization based on deep learning

A survey of fine-grained visual categorization based on deep learning

Transformer with Peak Suppression and Knowledge Guidance for Fine-grained Image Recognition

Audio description from image by modal translation network

Contact Info

Product

Resources

About