Automatic segmentation of retinal vessels in fundus images plays an important role in the diagnosis of some diseases such as diabetes and hypertension. In this paper, we propose Deformable U-Net ...(DUNet), which exploits the retinal vessels’ local features with a U-shape architecture, in an end to end manner for retinal vessel segmentation. Inspired by the recently introduced deformable convolutional networks, we integrate the deformable convolution into the proposed network. The DUNet, with upsampling operators to increase the output resolution, is designed to extract context information and enable precise localization by combining low-level features with high-level ones. Furthermore, DUNet captures the retinal vessels at various shapes and scales by adaptively adjusting the receptive fields according to vessels’ scales and shapes. Public datasets: DRIVE, STARE, CHASE_DB1 and HRF are used to test our models. Detailed comparisons between the proposed network and the deformable neural network, U-Net are provided in our study. Results show that more detailed vessels can be extracted by DUNet and it exhibits state-of-the-art performance for retinal vessel segmentation with a global accuracy of 0.9566/0.9641/0.9610/0.9651 and AUC of 0.9802/0.9832/0.9804/0.9831 on DRIVE, STARE, CHASE_DB1 and HRF respectively. Moreover, to show the generalization ability of the DUNet, we use another two retinal vessel data sets, i.e., WIDE and SYNTHE, to qualitatively and quantitatively analyze and compare with other methods. Extensive cross-training evaluations are used to further assess the extendibility of DUNet. The proposed method has the potential to be applied to the early diagnosis of diseases.
•A deep neural network (DUNet) for automatic segmentation of retinal vessel is built.•DUNet exploits the retinal vessels’ features with a U-shape architecture.•DUNet captures retinal vessels adaptively, according to vessels’ scales and shapes.•DUNet extracts more detailed vessels than deformable neural network and U-Net.•Comparisons between many methods show competitive performance and generalization.
It is a challenging task to learn discriminative representation from images and videos, due to large local redundancy and complex global dependency in these visual data. Convolution neural networks ...(CNNs) and vision transformers (ViTs) have been two dominant frameworks in the past few years. Though CNNs can efficiently decrease local redundancy by convolution within a small neighborhood, the limited receptive field makes it hard to capture global dependency. Alternatively, ViTs can effectively capture long-range dependency via self-attention, while blind similarity comparisons among all the tokens lead to high redundancy. To resolve these problems, we propose a novel Unified transFormer (UniFormer), which can seamlessly integrate the merits of convolution and self-attention in a concise transformer format. Different from the typical transformer blocks, the relation aggregators in our UniFormer block are equipped with local and global token affinity respectively in shallow and deep layers, allowing tackling both redundancy and dependency for efficient and effective representation learning. Finally, we flexibly stack our blocks into a new powerful backbone, and adopt it for various vision tasks from image to video domain, from classification to dense prediction. Without any extra training data, our UniFormer achieves 86.3 top-1 accuracy on ImageNet-1 K classification task. With only ImageNet-1 K pre-training, it can simply achieve state-of-the-art performance in a broad range of downstream tasks. It obtains 82.9/84.8 top-1 accuracy on Kinetics-400/600, 60.9/71.2 top-1 accuracy on Something-Something V1/V2 video classification tasks, 53.8 box AP and 46.4 mask AP on COCO object detection task, 50.8 mIoU on ADE20 K semantic segmentation task, and 77.4 AP on COCO pose estimation task. Moreover, we build an efficient UniFormer with a concise hourglass design of token shrinking and recovering, which achieves 2-4 <inline-formula><tex-math notation="LaTeX">\bm {\times }</tex-math> <mml:math><mml:mo>×</mml:mo></mml:math><inline-graphic xlink:href="li-ieq1-3282631.gif"/> </inline-formula> higher throughput than the recent lightweight models.
One essential problem in skeleton-based action recognition is how to extract discriminative features over all skeleton joints. However, the complexity of the recent State-Of-The-Art (SOTA) models for ...this task tends to be exceedingly sophisticated and over-parameterized. The low efficiency in model training and inference has increased the validation costs of model architectures in large-scale datasets. To address the above issue, recent advanced separable convolutional layers are embedded into an early fused Multiple Input Branches (MIB) network, constructing an efficient Graph Convolutional Network (GCN) baseline for skeleton-based action recognition. In addition, based on such the baseline, we design a compound scaling strategy to expand the model's width and depth synchronously, and eventually obtain a family of efficient GCN baselines with high accuracies and small amounts of trainable parameters, termed EfficientGCN-Bx, where "x" denotes the scaling coefficient. On two large-scale datasets, i.e. , NTU RGB+D 60 and 120, the proposed EfficientGCN-B4 baseline outperforms other SOTA methods, e.g. , achieving 92.1% accuracy on the cross-subject benchmark of NTU 60 dataset, while being 5.82× smaller and 5.85× faster than MS-G3D, which is one of the SOTA methods. The source code in PyTorch version and the pretrained models are available at https://github.com/yfsong0709/EfficientGCNv1 .
Due to their spatial and spectral information, hyperspectral images are frequently used in various scientific and industrial fields. Recent developments in hyperspectral image classification have ...revolved around the use of convolutional neural networks (CNNs) and transformers, which are capable of modeling local and global data. However, most of the backbone networks of existing methods are based on 3-D convolution, which have high complexity in network structure. Moreover, local information and global information are extracted through different modules, and the coupling relationship between the two types of information is weak. To address the above issues, we propose a method named main-sub transformer network with spectral-spatial separable convolution method (MST-SSSNet), which includes two key modules: the spectral-spatial separable convolution (SSSC) module and the main-sub transformer encoder (MST) module. The SSSC module uses the proposed spectral-spatial separable convolution, reducing network parameters and efficiently extracting local features. The MST module adds the designed sub-transformer in front of the conventional transformer encoder (main-transformer). It assists the main-transformer encoder to establish global correlation by learning local information. The WHU-Hi dataset can be used as a benchmark dataset for precise crop classification and hyperspectral image classification research. MST-SSSNet is shown to deliver better classification performance than current state-of-the-art methods on the datasets. The code will be downloaded to https://github.com/fengqinshou/MST-SSSNet .
In this work we address the task of semantic image segmentation with Deep Learning and make three main contributions that are experimentally shown to have substantial practical merit. First, we ...highlight convolution with upsampled filters, or `atrous convolution', as a powerful tool in dense prediction tasks. Atrous convolution allows us to explicitly control the resolution at which feature responses are computed within Deep Convolutional Neural Networks. It also allows us to effectively enlarge the field of view of filters to incorporate larger context without increasing the number of parameters or the amount of computation. Second, we propose atrous spatial pyramid pooling (ASPP) to robustly segment objects at multiple scales. ASPP probes an incoming convolutional feature layer with filters at multiple sampling rates and effective fields-of-views, thus capturing objects as well as image context at multiple scales. Third, we improve the localization of object boundaries by combining methods from DCNNs and probabilistic graphical models. The commonly deployed combination of max-pooling and downsampling in DCNNs achieves invariance but has a toll on localization accuracy. We overcome this by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF), which is shown both qualitatively and quantitatively to improve localization performance. Our proposed "DeepLab" system sets the new state-of-art at the PASCAL VOC-2012 semantic image segmentation task, reaching 79.7 percent mIOU in the test set, and advances the results on three other datasets: PASCAL-Context, PASCAL-Person-Part, and Cityscapes. All of our code is made publicly available online.
Direct convolution methods are now drawing increasing attention as they eliminate the additional storage demand required by indirect convolution algorithms (i.e., the transformed matrix generated by ...the im2col convolution algorithm). Nevertheless, the direct methods require special input-output tensor formatting, leading to extra time and memory consumption to get the desired data layout. In this article, we show that indirect convolution, if implemented properly, is able to achieve high computation performance with the help of highly optimized subroutines in matrix multiplication while avoid incurring substantial memory overhead. The proposed algorithm is called efficient convolution via blocked columnizing (ECBC). Inspired by the im2col convolution algorithm and the block algorithm of general matrix-to-matrix multiplication, we propose to conduct the convolution computation blockwisely. As a result, the tensor-to-matrix transformation process (e.g., the im2col operation) can also be done in a blockwise manner so that it only requires a small block of memory as small as the data block. Extensive experiments on various platforms and networks validate the effectiveness of ECBC, as well as the superiority of our proposed method against a set of widely used industrial-level convolution algorithms.
An energy-efficient, unified convolutional neural network (CNN) accelerator is proposed with a lightweight RGB-D network to achieve real-time, multi-object semantic segmentation in autonomous ...electric vehicle system. First, a lightweight Depth-fused Trilateral Network (DTN) is proposed to achieve high accuracy and real-time operation for road and multi-object segmentation at the same time. Optimized with various types of convolution layers and limited hardware resources, the DTN achieves 94.73% accuracy on KITTI Road dataset. Second, the unified CNN processor is designed with dual-mode shift-register-based input reconfiguration units and layer fusion architecture with 2-types of processing elements for depth-wise separable convolution (DSC) to support 5 different types of convolution layers including standard convolution, dilated convolution, transposed convolution, point-wise convolution, and DSC. With flexible architecture, it achieves <inline-formula> <tex-math notation="LaTeX">17.97\times </tex-math></inline-formula> higher throughput with DTN and DSC layer fusion architecture reduces 34.7% of overall external memory access. Implemented with 28nm CMOS technology, the unified CNN processor shows 43.6 mW power consumption and 4.94 TOPS/W energy efficiency. As a result, the proposed system with DTN realizes 40.07 frames-per-second (fps) throughputs in multi-object semantic segmentation application with high resolution driving scenes dataset.
Traffic prediction is a core problem in the intelligent transportation system and has broad applications in the transportation management and planning, and the main challenge of this field is how to ...efficiently explore the spatial and temporal information of traffic data. Recently, various deep learning methods, such as convolution neural network (CNN), have shown promising performance in traffic prediction. However, it samples traffic data in regular grids as the input of CNN, thus it destroys the spatial structure of the road network. In this paper, we introduce a graph network and propose an optimized graph convolution recurrent neural network for traffic prediction, in which the spatial information of the road network is represented as a graph. Additionally, distinguishing with most current methods using a simple and empirical spatial graph, the proposed method learns an optimized graph through a data-driven way in the training phase, which reveals the latent relationship among the road segments from the traffic data. Lastly, the proposed method is evaluated on three real-world case studies, and the experimental results show that the proposed method outperforms state-of-the-art traffic prediction methods.
Abstract
Despite the fact that deep learning has achieved remarkable success in various domains over the past decade, its application in molecular informatics and drug discovery is still limited. ...Recent advances in adapting deep architectures to structured data have opened a new paradigm for pharmaceutical research. In this survey, we provide a systematic review on the emerging field of graph convolutional networks and their applications in drug discovery and molecular informatics. Typically we are interested in why and how graph convolution networks can help in drug-related tasks. We elaborate the existing applications through four perspectives: molecular property and activity prediction, interaction prediction, synthesis prediction and de novo drug design. We briefly introduce the theoretical foundations behind graph convolutional networks and illustrate various architectures based on different formulations. Then we summarize the representative applications in drug-related problems. We also discuss the current challenges and future possibilities of applying graph convolutional networks to drug discovery.