Local Relation Networks for Image Recognition

Hu, Han; Zhang, Zheng; Xie, Zhenda; Lin, Stephen

doi:10.1109/iccv.2019.00356

Cited by 482 publications

(284 citation statements)

References 27 publications

Supporting

Mentioning

279

Contrasting

Order By: Relevance

“…Especially, the [24] is been seen as the first to use the attention mechanism to capture the global dependencies of inputs. Recently, the attention mechanism is popular and widely used in computer vision tasks [25][26][27][28][29][30]. The attention mechanism can simulate the human visual system and focus on the areas of interest.…”

Section: B Attention Mechanismmentioning

confidence: 99%

MA-Net: A Multi-Scale Attention Network for Liver and Tumor Segmentation

et al. 2020

View full text Add to dashboard Cite

Automatic assessing the location and extent of liver and liver tumor is critical for radiologists, diagnosis and the clinical process. In recent years, a large number of variants of U-Net based on Multi-scale feature fusion are proposed to improve the segmentation performance for medical image segmentation. Unlike the previous works which extract the context information of medical image via applying the multi-scale feature fusion, we propose a novel network named Multi-scale Attention Net (MA-Net) by introducing self-attention mechanism into our method to adaptively integrate local features with their global dependencies. The MA-Net can capture rich contextual dependencies based on the attention mechanism. We design two blocks: Position-wise Attention Block (PAB) and Multi-scale Fusion Attention Block (MFAB). The PAB is used to model the feature interdependencies in spatial dimensions, which capture the spatial dependencies between pixels in a global view. In addition, the MFAB is to capture the channel dependencies between any feature map by multi-scale semantic feature fusion. We evaluate our method on the dataset of MICCAI 2017 LiTS Challenge. The proposed method achieves better performance than other state-of-the-art methods. The Dice values of liver and tumors segmentation are 0.960±0.03 and 0.749±0.08 respectively.

show abstract

Section: B Attention Mechanismmentioning

confidence: 99%

MA-Net: A Multi-Scale Attention Network for Liver and Tumor Segmentation

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Recently, relation network have been successfully applied in the fields of natural language processing [35,40], few-shot learning [38], object detection [18,19,46], action recognition [36] and multiobject tracking [48]. Since there is not any excessive assumption about the data forms, relation network can be widely utilized to capture long-range, non-grid or differently distributed dependencies between data, such as word-word [40], pixel-pixel [19,46], object-pixel [13] and object-object [18,48]. However, these data is hard to be modeled by regular convolutions or sequential networks.…”

Section: Relation Networkmentioning

confidence: 99%

“…A relation network [35] is firstly used as a general solution to capture the core common properties of relational reasoning. Furthermore, RN [38] is used to learn a distance metric to compare few-numbers of images within episodes, while Han et al [18,19] proposed an object-object relation module and local relation networks to adaptively aggregate weight based on the content-aware relationship of elements. In order to capture heterogeneous dependencies, STRN [48] developed a spatial-temporal relation network to model the topological connection of objects in the spatial domain and performs aggregation over the temporal domains.…”

Section: Relation Networkmentioning

confidence: 99%

Reinforced Similarity Learning

Zhang

Zheng

et al. 2020

Proceedings of the 28th ACM International Conference on Multimedia

View full text Add to dashboard Cite

Recently, Siamese networks based tracking algorithms have shown favorable performance. Latest work focuses on better feature embedding and target state estimation, which greatly improves the accuracy. Nevertheless, the simple cross-correlation operation of the features between a fixed template and the search region limits their robustness and discrimination capability. In this paper, we pay more attention to learn an outstanding similarity measure for robust tracking. We propose a novel relation network that can be integrated on top of previous trackers without any need for further training of the siamese networks, which achieves a superior discriminative ability. During online inference, we utilize the feedback from high-confidence tracking results to obtain an additional template and update it, which improves the robustness and generalization. We implement two versions of the proposed approach with the SiamFC-based tracker and SiamRPN-based tracker to validate the strong compatibility of our algorithm. Extensive experimental results on several tracking benchmarks indicate that the proposed method can effectively improve the performance and robustness of the underlying trackers without reducing speed too much, and performs superiorly against the state-of-the-art trackers.

show abstract

“…All previous works try to find discriminative regions/patches from high-level feature (HLF) maps directly while neglect the fact that HLF maps are derived based on the spatial aggregation of convolution which is basically a pattern matching process that applies fixed filters [15,24,25]. This will inherently cause certain response differences for semantically similar image parts with spatial variations.…”

Section: Introductionmentioning

confidence: 99%

Category-specific Semantic Coherency Learning for Fine-grained Image Recognition

Wang

et al. 2020

Proceedings of the 28th ACM International Conference on Multimedia

View full text Add to dashboard Cite

Existing deep learning based weakly supervised fine-grained image recognition (WFGIR) methods usually pick out the discriminative regions from the high-level feature (HLF) maps directly. However, as HLF maps are derived based on spatial aggregation of convolution which is basically a pattern matching process that applies fixed filters, it is ineffective to model visual contents of same semantic but varying posture or perspective. We argue that this will cause the selected discriminative regions of same sub-category are not semantically corresponding and thus degrade the WFGIR performance. To address this issue, we propose an end-to-end Category-specific Semantic Coherency Network (CSC-Net) to semantically align the discriminative regions of the same subcategory. Specifically, CSC-Net consists of: 1) Local-to-Attribute Projecting Module (LPM), which automatically learns a set of latent attributes via collecting the category-specific semantic details while eliminating the varying spatial distributions from the local regions. 2) Latent Attribute Aligning (LAA), which aligns the latent attributes to specific semantic via graph convolution based on their discriminability, to achieve category-specific semantic coherency; 3) Attribute-to-Local Resuming Module (ARM), which resumes the original Euclidean space of latent attributes and construct latent attribute aligned feature maps by a location-embedding graph unpooling operation. Finally, the new feature maps are used which applies the category-specific semantic coherency implicitly for more accurate discriminative regions localization. Extensive experiments verify that CSC-Net yields the best performance under the same settings with most competitive approaches, on CUB Bird, Stanford-Cars, and FGVC Aircraft datasets.

show abstract

Local Relation Networks for Image Recognition

Cited by 482 publications

References 27 publications

MA-Net: A Multi-Scale Attention Network for Liver and Tumor Segmentation

MA-Net: A Multi-Scale Attention Network for Liver and Tumor Segmentation

Reinforced Similarity Learning

Category-specific Semantic Coherency Learning for Fine-grained Image Recognition

Contact Info

Product

Resources

About