Scale channel attention network for image segmentation

Chen, Jianjun; Tian, Youliang; Ma, Wei; Mao, Zhengdong; Hu, Yue

doi:10.1007/s11042-020-08921-7

Cited by 4 publications

(2 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Nonetheless, the AP [23] produces the average, reducing the greatest values, where the essential features sit. In contrast, MP keeps the most significant value [24] but worsens the overfitting hurdle.…”

Section: Figure 5 Diagram Of Block-wise Poolingmentioning

confidence: 97%

Fingerspelling Recognition by 12-Layer CNN with Stochastic Pooling

2022

View full text Add to dashboard Cite

Aim) Fingerspelling is a method of spelling words via hand movements. This study aims to propose a novel fingerspelling recognition system. (Method) We use 1320 fingerspelling images in our dataset. Our method is based on the convolutional neural network (CNN) model. We propose a 12-layer CNN as the backbone. Particularly, stochastic pooling (SP) is used to help solve the problems caused by max pooling or average pooling. In addition, an improved 20-way data augmentation method is proposed to circumvent overfitting. Our method is dubbed CNNSP.(Results) The results show that our CNNSP method achieved a micro-averaged F1 (MAF) score of 90.04±0.82%. In contrast, the MAFs of l2-pooling, average pooling, and max pooling are 86.21±1.12%, 87.54±1.39%, and 89.07±0.78%, respectively. (Conclusion) Our CNNSP attains better results than eight state-of-the-art fingerspelling recognition methods. Besides, the SP is better than l2-pooling, average pooling, and max pooling.

show abstract

Section: Figure 5 Diagram Of Block-wise Poolingmentioning

confidence: 97%

Fingerspelling Recognition by 12-Layer CNN with Stochastic Pooling

2022

View full text Add to dashboard Cite

show abstract

“…However, the existing congestion control protocols do not fully consider the quality of service and weighted fairness. Xiong et al [21,22] mainly consider how to improve the efficiency of data transmission, that is, quality of service, without involving weighted fairness. Wu et al [23,24] consider not only the quality of service, but also fairness, but only simple fairness, that is, ensure that sink nodes receive as many packets as each node in a certain period.…”

Section: Related Workmentioning

confidence: 99%

Improved Routing Protocol Based on Multiobjective Optimization in Industrial Robot Networks

Liu

Alias

Liu

et al. 2022

Advances in Multimedia

View full text Add to dashboard Cite

This paper decomposes the routing process of industrial robot network using the application of analytic hierarchy process in decision-making. The influence of four factors, such as path length, data integrity, energy consumption, and receiving delay, on routing effect is analyzed. Simultaneous interpreting routes are selected to achieve the purpose of routing. Simulation results show that this method can more comprehensively consider the factors affecting routing and is superior to the existing methods in terms of energy consumption, data integrity, and transmission delay.

show abstract

Road extraction in vague images on gray scale consistency and improved MSR and D-S evidence

Wang

Zhou

2022

Multimed Tools Appl

View full text Add to dashboard Cite

Road detection on aerial and remote sensing vague images is a hard task. In this paper, an automatic road detection method for the vague images is proposed. The method firstly uses an improved MSR algorithm to enhance image, and it automatically takes different scales in different image regions, based on the image depths obtained by the dark channel prior algorithm. Then the enhanced image is roughly segmented on the principle of the local gray scale consistency, in that, an eight-neighborhood template is considered as a processing unit in which a threshold is utilized for all the neighboring pixels of the detecting pixel. Finally, the Dempster-Shafer (D-S) evidence for road features is applied to finalize road tracing in the binary image, where, the road features include length, width, aspect ratio and fullness rate, all the parameters are obtained in the least external rectangle of a road segment, and then the detected roads are regulated. In experiments, 300 vague road images were selected for testing, by comparing to several traditional algorithms and recent semantic methods, the testing results show that the new method is satisfactory, and the detection accuracy is up to 89%.

show abstract

Scale channel attention network for image segmentation

Cited by 4 publications

References 33 publications

Fingerspelling Recognition by 12-Layer CNN with Stochastic Pooling

Fingerspelling Recognition by 12-Layer CNN with Stochastic Pooling

Improved Routing Protocol Based on Multiobjective Optimization in Industrial Robot Networks

Road extraction in vague images on gray scale consistency and improved MSR and D-S evidence

Contact Info

Product

Resources

About