ResNeSt: Split-Attention Networks

Zhang, Hang; Wu, Chongruo; Zhang, Zhongyue; Zhu, Yi; Zhang, Zhi; Lin, Haibin; Sun, Yue; He, Tong; Mueller, Jonas; Manmatha, R.; Li, Mu; Smola, Alexander J.

doi:10.1109/cvprw56347.2022.00309

Cited by 750 publications

(308 citation statements)

References 38 publications

Supporting

Mentioning

303

Contrasting

Unclassified

Order By: Relevance

“…e residual block is very efficient to build a deeper network. More recently, ResNeXt [22] and ResNest [23] have been designed to improve feature extraction process. Although these above deep CNN architectures extract high-quality object features, they require intensive computations due to large numbers of parameters.…”

Section: Lightweight Network For Extracting High-level Semanticmentioning

confidence: 99%

An Efficient License Plate Detection Approach Using Lightweight Deep Convolutional Neural Networks

Nguyen

2022

Advances in Multimedia

View full text Add to dashboard Cite

Benefited from deep convolutional neural networks, various license plate detection methods based on deep networks have been proposed and achieved significant improvements compared with traditional methods. However, the high computational cost due to complex structures prevents these methods from being deployed in real-world applications. This paper proposes an efficient license plate detection method based on lightweight deep convolutional neural networks for improving the detection speed. To extract high-level features from input images, this paper designs a lightweight feature pyramid generation module based on a lightweight architecture and depth-wise convolutions. To further enhance feature pyramid, an efficient feature enhancement module is designed to fuse features generated by the region proposal network with backbone features. In the detection network, a light head structure based on fully connected layers is employed to further reduce the computational cost of the model. In experiments, floating point operations and detection ratio are used to evaluate the efficient of the proposed method. Experimental results on public datasets show that the proposed method achieves the best trade-off between speed and accuracy.

show abstract

Section: Lightweight Network For Extracting High-level Semanticmentioning

confidence: 99%

An Efficient License Plate Detection Approach Using Lightweight Deep Convolutional Neural Networks

Nguyen

2022

Advances in Multimedia

View full text Add to dashboard Cite

show abstract

“…The approach can be applied in principle to any CNN. In the following experiments we have tested VGG16, as it was commonly used in many competing methods and comparisons, ResNet, as it has been used in the most recent state-of-the-art methods, and the novel ResNeSt architecture [58] that has recently improved results over previous ResNet in image classification, object detection, instance segmentation and semantic segmentation.…”

Section: Pooling Of Local Cnn Features and Descriptor Splittingmentioning

confidence: 99%

“…On these premises, and after selecting our more prominent setup (4-Split + ReLU + Max-Poolings), we proceed for a more time-expensive experiment training the networks at resolution 512 × 512. Finally, we test a very recent ResNet variant, the ResNeSt101 and ResNeSt50 [58], that have shown improved results in different tasks, from segmentation to classification; our experiments show that this architecture heavily outperforms its more traditional version also for the image retrieval task. In our experimental setup, the ResNeSt101 and ResNeSt50, respectively, completed the features extraction from the test datasets in about 120% and 85% of the total time employed by the original ResNet101.…”

Section: Trainings At Higher Resolution With Modern Architecturesmentioning

confidence: 99%

“…The proposed method is evaluated on two standard image retrieval datasets: Revisited Oxford5K and Revisited Paris6K, as well as the previous versions Oxford5K and Paris6K, outperforming previous state-of-the-art results. Experiments show that the proposed approach can be applied to different network architectures, from VGG [46] and ResNet [17], used to provide a fair comparison with competing approaches, to the very recent ResNeSt [58], used to show the applicability to more modern computationally efficient backbones.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Effective triplet mining improves training of multi-scale pooled CNN for image retrieval

Vaccaro

Bertini

Uricchio

et al. 2022

Machine Vision and Applications

View full text Add to dashboard Cite

In this paper, we address the problem of content-based image retrieval (CBIR) by learning images representations based on the activations of a Convolutional Neural Network. We propose an end-to-end trainable network architecture that exploits a novel multi-scale local pooling based on the trainable aggregation layer NetVLAD (Arandjelovic et al in Proceedings of the IEEE conference on computer vision and pattern recognition CVPR, NetVLAD, 2016) and bags of local features obtained by splitting the activations, allowing to reduce the dimensionality of the descriptor and to increase the performance of retrieval. Training is performed using an improved triplet mining procedure that selects samples based on their difficulty to obtain an effective image representation, reducing the risk of overfitting and loss of generalization. Extensive experiments show that our approach, that can be effectively used with different CNN architectures, obtains state-of-the-art results on standard and challenging CBIR datasets.

show abstract

“…The backbone network of the semantic feature enhancement module is the ResNest [27] model. The Split‐Attention module in the network groups the channel dimensions of the feature map and calculates the attention weights corresponding to each group of features, which effectively improves the quality of features extracted from the model without significantly increasing the model computation and parameters.…”

Section: Approachmentioning

confidence: 99%

Occluded pedestrian detection combined with semantic features

Ruan

Zhang

2021

IET Image Processing

View full text Add to dashboard Cite

The task of pedestrian detection is to identify the location and size of pedestrians in images or videos. However, occlusions are very common in real-life scenarios, which make pedestrian detection more difficult. In order to solve the occlusion problem in pedestrian detection, a semantic feature enhancement module to acquire more informative and richer semantic features is proposed. The detector enhances semantic features by fusing feature maps of different layers, and detects pedestrians based on their locations and scales. The Experiments performed on Caltech and CityPersons datasets show that the algorithm achieves superior performance for detecting occluded pedestrians, especially heavily occluded ones. 30.6% and 47.9% log-average missing rates are achieved in the heavily occluded subsets of Caltech and Cityperons, respectively. Moreover, this method is robust to the detection of heavily occluded pedestrians, and the module can be easily used by other detection frameworks.How to cite this article: Ruan, B., Zhang, C.: Occluded pedestrian detection combined with semantic features.

show abstract

ResNeSt: Split-Attention Networks

Cited by 750 publications

References 38 publications

An Efficient License Plate Detection Approach Using Lightweight Deep Convolutional Neural Networks

An Efficient License Plate Detection Approach Using Lightweight Deep Convolutional Neural Networks

Effective triplet mining improves training of multi-scale pooled CNN for image retrieval

Occluded pedestrian detection combined with semantic features

Contact Info

Product

Resources

About