近年来，视觉变换器（ViTs）主导了计算机视觉领域。然而，ViTs计算开销大，不适合移动设备；这导致了卷积神经网络（CNN）和基于ViT的混合模型在移动视觉应用中的普及。最近，视觉图神经网络（ViG）和CNN混合模型也被提出用于移动视觉任务。然而，所有这些方法在速度上仍比纯CNN模型慢。在这项工作中，我们提出多级膨胀卷积来设计一个纯CNN的移动端骨干网络。使用多级膨胀卷积可以获得比标准卷积更大的理论感受野。不同级别的膨胀还允许图像中短程和长程特征之间的交互。实验表明，我们提出的模型在图像分类、目标检测、实例分割和语义分割上，在准确性和/或速度方面优于最先进的（SOTA）移动CNN、ViT、ViG和混合架构。我们最快的模型RapidNet-Ti在ImageNet-1K上达到76.3%的top-1准确率，在iPhone 13 mini NPU上推理延迟为0.9毫秒，比MobileNetV2x1.4（74.7% top-1准确率，1.0毫秒延迟）更快且更准确。我们的工作表明，纯CNN架构在正确设计时，可以在准确性和速度上击败SOTA混合模型和ViT模型。
英文摘要：
Vision transformers (ViTs) have dominated computer vision in recent years. However, ViTs are computationally expensive and not well suited for mobile devices; this led to the prevalence of convolutional neural network (CNN) and ViT-based hybrid models for mobile vision applications. Recently, Vision GNN (ViG) and CNN hybrid models have also been proposed for mobile vision tasks. However, all of these methods remain slower compared to pure CNN-based models. In this work, we propose Multi-Level Dilated Convolutions to devise a purely CNN-based mobile backbone. Using Multi-Level Dilated Convolutions allows for a larger theoretical receptive field than standard convolutions. Different levels of dilation also allow for interactions between the short-range and long-range features in an image. Experiments show that our proposed model outperforms state-of-the-art (SOTA) mobile CNN, ViT, ViG, and hybrid architectures in terms of accuracy and/or speed on image classification, object detection, instance segmentation, and semantic segmentation. Our fastest model, RapidNet-Ti, achieves 76.3\% top-1 accuracy on ImageNet-1K with 0.9 ms inference latency on an iPhone 13 mini NPU, which is faster and more accurate than MobileNetV2x1.4 (74.7\% top-1 with 1.0 ms latency). Our work shows that pure CNN architectures can beat SOTA hybrid and ViT models in terms of accuracy and speed when designed properly.
Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；Deepwise AI Lab（深度智作AI实验室 (Deepwise AI Lab)）
Modern approaches for semantic segmentation usually employ dilated convolutions in the backbone to extract high-resolution feature maps, which brings heavy computation complexity and memory footprint. To replace the time and memory consuming dilated convolutions, we propose a novel joint upsampling module named Joint Pyramid Upsampling (JPU) by formulating the task of extracting high-resolution feature maps into a joint upsampling problem. With the proposed JPU, our method reduces the computation complexity by more than three times without performance loss. Experiments show that JPU is superio…
CUHK-SenseTime Joint Laboratory, The Chinese University of Hong Kong（香港中文大学-商汤科技联合实验室）；Shenzhen Key Lab of Computer Vision and Pattern Recognition, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（中国科学院深圳先进技术研究院计算机视觉与模式识别深圳市重点实验室）；SenseTime Research（商汤科技研究院）
Both performance and efficiency are important to semantic segmentation. State-of-the-art semantic segmentation algorithms are mostly based on dilated Fully Convolutional Networks (dilatedFCN), which adopt dilated convolutions in the backbone networks to extract high-resolution feature maps for achieving high-performance segmentation performance. However, due to many convolution operations are conducted on the high-resolution feature maps, such dilatedFCN-based methods result in large computational complexity and memory consumption. To balance the performance and efficiency, there also exist en…
Beyond Mamba: Enhancing State-space Models with Deformable Dilated Convolutions for Multi-scale Traffic Object Detection
Department of Physics, J.K. Institute of Science（J.K.科学研究所物理系）；World Scientific University（世界科学出版社大学）；University of Intelligent Studies（智能研究大学）；School of Computer and Electronic Information, Nanjing Normal University（南京师范大学计算机与电子信息学院）
Beihang University（北京航空航天大学）；SenseTime Research（商汤科技研究院）；University of Oxford（牛津大学）；The University of Sydney（悉尼大学）
As a variant of standard convolution, a dilated convolution can control effective receptive fields and handle large scale variance of objects without introducing additional computational costs. To fully explore the potential of dilated convolution, we proposed a new type of dilated convolution (referred to as inception convolution), where the convolution operations have independent dilation patterns among different axes, channels and layers. To develop a practical method for learning complex inception convolution based on the data, a simple but effective search algorithm, referred to as effici…
The Chinese University of Hong Kong（香港中文大学）；SenseTime Research（商汤科技研究院）；Shandong Normal University（山东师范大学）
Both high-level and high-resolution feature representations are of great importance in various visual understanding tasks. To acquire high-resolution feature maps with high-level semantic information, one common strategy is to adopt dilated convolutions in the backbone networks to extract high-resolution feature maps, such as the dilatedFCN-based methods for semantic segmentation. However, due to many convolution operations are conducted on the high-resolution feature maps, such methods have large computational complexity and memory consumption. In this paper, we propose one novel holistically…
Google Research（谷歌研究院）；University of Texas at Austin（德克萨斯大学奥斯汀分校）
Transformers have recently gained significant attention in the computer vision community. However, the lack of scalability of self-attention mechanisms with respect to image size has limited their wide adoption in state-of-the-art vision backbones. In this paper we introduce an efficient and scalable attention model we call multi-axis attention, which consists of two aspects: blocked local and dilated global attention. These design choices allow global-local spatial interactions on arbitrary input resolutions with only linear complexity. We also present a new architectural element by effective…
Beijing Jiaotong University（北京交通大学）；National University of Singapore（新加坡国立大学）
To detect salient objects accurately, existing methods usually design complex backbone network architectures to learn and fuse powerful features. However, the saliency inference module that performs saliency prediction from the fused features receives much less attention on its architecture design and typically adopts only a few fully convolutional layers. In this paper, we find the limited capacity of the saliency inference module indeed makes a fundamental performance bottleneck, and enhancing its capacity is critical for obtaining better saliency prediction. Correspondingly, we propose a de…
Though the state-of-the architectures for semantic segmentation, such as HRNet, demonstrate impressive accuracy, the complexity arising from their salient design choices hinders a range of model acceleration tools, and further they make use of operations that are inefficient on current hardware. This paper demonstrates that a simple encoder-decoder architecture with a ResNet-like backbone and a small multi-scale head, performs on-par or better than complex semantic segmentation architectures such as HRNet, FANet and DDRNets. Naively applying deep backbones designed for Image Classification to …
Accurate lesion detection in computer tomography (CT) slices benefits pathologic organ analysis in the medical diagnosis process. More recently, it has been tackled as an object detection problem using the Convolutional Neural Networks (CNNs). Despite the achievements from off-the-shelf CNN models, the current detection accuracy is limited by the inability of CNNs on lesions at vastly different scales. In this paper, we propose a Multi-Scale Booster (MSB) with channel and spatial attention integrated into the backbone Feature Pyramid Network (FPN). In each pyramid level, the proposed MSB captu…
University of Michigan – Shanghai Jiao Tong University Joint Institute, Shanghai Jiao Tong University（上海交通大学密歇根学院）
Cardiac structure segmentation plays an important role in medical analysis procedures. Images' blurred boundaries issue always limits the segmentation performance. To address this difficult problem, we presented a novel network structure which embedded dilated pyramid pooling block in the skip connections between networks' encoding and decoding stage. A dilated pyramid pooling block is made up of convolutions and pooling operations with different vision scopes. Equipped the model with such module, it could be endowed with multi-scales vision ability. Together combining with other techniques, i…
Montreal Institute for Learning Algorithms（蒙特利尔学习算法研究所）；Computer Vision Center, Barcelona（巴塞罗那计算机视觉中心）；Facebook AI Research（Facebook人工智能研究）
State-of-the-art semantic segmentation approaches increase the receptive field of their models by using either a downsampling path composed of poolings/strided convolutions or successive dilated convolutions. However, it is not clear which operation leads to best results. In this paper, we systematically study the differences introduced by distinct receptive field enlargement methods and their impact on the performance of a novel architecture, called Fully Convolutional DenseResNet (FC-DRN). FC-DRN has a densely connected backbone composed of residual networks. Following standard image segment…
University of Electronic Science and Technology of China（电子科技大学）；Southwest Jiaotong University（西南交通大学）
Multi-scale detection plays an important role in object detection models. However, researchers usually feel blank on how to reasonably configure detection heads combining multi-scale features at different input resolutions. We find that there are different matching relationships between the object distribution and the detection head at different input resolutions. Based on the instructive findings, we propose a lightweight traffic object detection network based on matching between detection head and object distribution, termed as MHD-Net. It consists of three main parts. The first is the detec…
Multi-Scale Fruit Capsules: Dilated Convolutions and Dynamic Routing for In-the-Wild Explainable Fruit Recognition
University of Maryland Baltimore County（马里兰大学巴尔的摩分校）；Indian Institute of Technology Kharagpur（印度理工学院克勒格布尔分校）；Indian Institute of Science Education and Research Bhopal（印度科学教育与研究学院博帕尔分校）；UJM-Saint-Etienne（圣埃蒂安大学）；CNRS（法国国家科学研究中心）；Laboratoire Hubert Curien（于贝尔·居里安实验室）
MambaDSF: Multi-Scale SSM with Dilated Feature Fusion for Sonar Small Target Detection
School of Information Science and Engineering, Ocean University of China（中国海洋大学信息科学与工程学院）；School of Information and Communication Technology, Griffith University（格里菲斯大学信息与通信技术学院）
HBFormer: A Hybrid-Bridge Transformer for Microtumor and Miniature Organ Segmentation
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（深圳先进技术研究所，中国科学院）；Guangdong University of Technology（广东技术大学）；University of Macau（澳门大学）；Huizhou University（惠州大学）
Towards Accurate Camouflaged Object Detection with Mixture Convolution and Interactive Fusion
National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology, School of Computer Science and Engineering, Northwestern Polytechnical University（西北工业大学计算机学院）；Center for Brain Imaging Science and Technology, Zhejiang University（浙江大学）；School of Computer Science, China University of Geosciences（中国地质大学计算机学院）；Optical Electrical and Computer Engineering, University of Shanghai for Science and Technology（上海理工大学）；School of Electrical Engineering & Information Department, Northeast Petroleum University（东北石油大学）；School of Electronic, Electrical Engineering and Physics, Fujian University of Technology（福建工程学院）
Accurate segmentation of retinal vessels is a basic step in Diabetic retinopathy(DR) detection. Most methods based on deep convolutional neural network (DCNN) have small receptive fields, and hence they are unable to capture global context information of larger regions, with difficult to identify lesions. The final segmented retina vessels contain more noise with low classification accuracy. Therefore, in this paper, we propose a DCNN structure named as D-Net. In the proposed D-Net, the dilation convolution is used in the backbone network to obtain a larger receptive field without losing spati…
Shaanxi University of Science & Technology（陕西科技大学）；Shenzhen University（深圳大学）
Fingerprint is an important biological feature of human body, which contains abundant gender information. At present, the academic research of fingerprint gender characteristics is generally at the level of understanding, while the standardization research is quite limited. In this work, we propose a more robust method, Dense Dilated Convolution ResNet (DDC-ResNet) to extract valid gender information from fingerprints. By replacing the normal convolution operations with the atrous convolution in the backbone, prior knowledge is provided to keep the edge details and the global reception field c…
