DP-LinkNet: A convolutional network for historical document image binarization

doi:10.3837/tiis.2021.05.011

Cited by 14 publications

(3 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The dilation rate depends on the stride of the input feature map. The dilated convolution is commonly available in two connection of types called parallel type and cascade type (Xiong et al, 2021 ). The HAC has parallel mode and cascade mode.…”

Section: The Proposed Methodsmentioning

confidence: 99%

A Multi-Scale Densely Connected Convolutional Neural Network for Automated Thyroid Nodule Classification

et al. 2022

View full text Add to dashboard Cite

Automated thyroid nodule classification in ultrasound images is an important way to detect thyroid nodules and to make a more accurate diagnosis. In this paper, we propose a novel deep convolutional neural network (CNN) model, called n-ClsNet, for thyroid nodule classification. Our model consists of a multi-scale classification layer, multiple skip blocks, and a hybrid atrous convolution (HAC) block. The multi-scale classification layer first obtains multi-scale feature maps in order to make full use of image features. After that, each skip-block propagates information at different scales to learn multi-scale features for image classification. Finally, the HAC block is used to replace the downpooling layer so that the spatial information can be fully learned. We have evaluated our n-ClsNet model on the TNUI-2021 dataset. The proposed n-ClsNet achieves an average accuracy (ACC) score of 93.8% in the thyroid nodule classification task, which outperforms several representative state-of-the-art classification methods.

show abstract

Section: The Proposed Methodsmentioning

confidence: 99%

A Multi-Scale Densely Connected Convolutional Neural Network for Automated Thyroid Nodule Classification

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Use-cases needing precision-based analytics for QoS parameters in the CCPS network infrastructure must be powered by Machine learning models. Deep learning and convolutional network security will be investigated considering efforts in [65]- [68] for GB/s optical DCNs in Smart grid ecosystems in future. Time-space complexity analysis will be developed and validated.…”

Section: Future Research Directionsmentioning

confidence: 99%

Integrating Resilient Tier N+1 Networks with Distributed Non-Recursive Cloud Model for Cyber-Physical Applications

2022

KSII TIIS

View full text Add to dashboard Cite

Cyber-physical systems (CPS) have been growing exponentially due to improved clouddatacenter infrastructure-as-a-service (CDIaaS). Incremental expandability (scalability), Quality of Service (QoS) performance, and reliability are currently the automation focus on healthy Tier 4 CDIaaS. However, stable QoS is yet to be fully addressed in Cyber-physical data centers (CP-DCS). Also, balanced agility and flexibility for the application workloads need urgent attention. There is a need for a resilient and fault-tolerance scheme in terms of CPS routing service including Pod cluster reliability analytics that meets QoS requirements. Motivated by these concerns, our contributions are fourfold. First, a Distributed Non-Recursive Cloud Model (DNRCM) is proposed to support cyber-physical workloads for remote lab activities. Second, an efficient QoS stability model with Routh-Hurwitz criteria is established. Third, an evaluation of the CDIaaS DCN topology is validated for handling largescale, traffic workloads. Network Function Virtualization (NFV) with Floodlight SDN controllers was adopted for the implementation of DNRCM with embedded rule-base in Open vSwitch engines. Fourth, QoS evaluation is carried out experimentally. Considering the nonrecursive queuing delays with SDN isolation (logical), a lower queuing delay (19.65%) is observed. Without logical isolation, the average queuing delay is 80.34%. Without logical resource isolation, the fault tolerance yields 33.55%, while with logical isolation, it yields 66.44%. In terms of throughput, DNRCM, recursive BCube, and DCell offered 38.30%, 36.37%, and 25.53% respectively. Similarly, the DNRCM had an improved incremental scalability profile of 40.00%, while BCube and Recursive DCell had 33.33%, and 26.67% respectively. In terms of service availability, the DNRCM offered 52.10% compared with recursive BCube and DCell which yielded 34.72% and 13.18% respectively. The average delays obtained for DNRCM, recursive BCube, and DCell are 32.81%, 33.44%, and 33.75% respectively. Finally, workload utilization for DNRCM, recursive BCube, and DCell yielded 50.28%, 27.93%, and 21.79% respectively.

show abstract

“…With the success of deep learning in visual perception and recognition, a number of deep learning-based object detection methods have been proposed in the past decade [1][2][3] and have been applied to many tasks, such as vehicle detection [4,5], pedestrian detection [6][7][8], and so on. Object detection based on supervised deep learning requires a large number of labeled samples for training [9,10]. Then, we can define these methods as data-driven object detectors.…”

Section: Introductionmentioning

confidence: 99%

Analysis of Scale Sensitivity of Ship Detection in an Anchor-Free Deep Learning Framework

Jiang¹,

Huang

Zhang³

et al. 2022

Electronics

View full text Add to dashboard Cite

Ship detection is an important task in sea surveillance. In the past decade, deep learning-based methods have been proposed for ship detection from images and videos. Convolutional features are observed to be very effective in representing ship objects. However, the scales of convolution often lead to different capacities of feature representation. It is unclear how the scale influences the performance of deep learning methods in ship detection. To this end, this paper studies the scale sensitivity of ship detection in an anchor-free deep learning framework. Specifically, we employ the classical CenterNet as the base and analyze the influence of the size, the depth, and the fusion strategy of convolution features on multi-scale ship target detection. Experiments show that, for small targets, the features obtained from the top-down path fusion can improve the detection performance more significantly than that from the bottom-up path fusion; on the contrary, the bottom-up path fusion achieves better detection performance on larger targets.

show abstract

DP-LinkNet: A convolutional network for historical document image binarization

Cited by 14 publications

References 45 publications

A Multi-Scale Densely Connected Convolutional Neural Network for Automated Thyroid Nodule Classification

A Multi-Scale Densely Connected Convolutional Neural Network for Automated Thyroid Nodule Classification

Integrating Resilient Tier N+1 Networks with Distributed Non-Recursive Cloud Model for Cyber-Physical Applications

Analysis of Scale Sensitivity of Ship Detection in an Anchor-Free Deep Learning Framework

Contact Info

Product

Resources

About