摘要:
Object co-segmentation is a challenging task, which aims to segment common objects in multiple images at the same time. Generally, common information of the same object needs to be found to solve this problem. For various scenarios, common objects in different images only have the same semantic information. In this paper, we propose a deep object co-segmentation method based on channel and spatial attention, which combines the attention mechanism with a deep neural network to enhance the common semantic information. Siamese encoder and decoder structure are used for this task. Firstly, the encoder network is employed to extract low-level and high-level features of image pairs. Secondly, we introduce an improved attention mechanism in the channel and spatial domain to enhance the multi-level semantic features of common objects. Then, the decoder module accepts the enhanced feature maps and generates the masks of both images. Finally, we evaluate our approach on the commonly used datasets for the co-segmentation task. And the experimental results show that our approach achieves competitive performance. (C) 2020 Elsevier B.V. All rights reserved.
期刊:
International Journal of Wavelets, Multiresolution and Information Processing,2019年17(1):1950001 ISSN:0219-6913
通讯作者:
Wei, Yantao
作者机构:
[Yao, Huang; Shi, Yafei; Wei, Yantao; Tong, Mingwen; Liu, Qingtang; Chen, Tiantian; Deng, Wei; Zhao, Gang] Cent China Normal Univ, Sch Educ Informat Technol, Wuhan 430079, Hubei, Peoples R China.;[Pan, Donghui] Anhui Univ, Sch Math Sci, Hefei 230601, Anhui, Peoples R China.
通讯机构:
[Wei, Yantao] C;Cent China Normal Univ, Sch Educ Informat Technol, Wuhan 430079, Hubei, Peoples R China.
关键词:
Student body gesture recognition;fisher broad learning system;learning analytics
摘要:
Observing student body gesture has been widely used to assess teaching effectiveness over the past few decades. However, manual observation is not suitable for the automatic data analysis in the field of learning analytics. Consequently, a student body gesture recognition method based on Fisher Broad Learning System (FBLS) and Local Log-Euclidean Multivariate Gaussian (L2EMG) is proposed in this paper. FBLS is designed by introducing the discriminative information into the hidden layer of Broad Learning System (BLS) and reducing the dimensionality of hidden-layer representations. FBLS has superiorities in accuracy and speed. In addition, L2EMG, which is a highly distinctive descriptor, characterizes the local image with a multivariate Gaussian distribution. So L2EMG features are fed into the FBLS for recognition in this paper. Extensive experimental results on self-built dataset show that the proposed student body gesture recognition method obtains better results than other benchmarking methods.
期刊:
ICBDT '18: Proceedings of the 1st International Conference on Big Data Technologies,2018年:63-68
作者机构:
[Xia, Ying; Chen, Min; Tong, Mingwen; Zhou, Chuang] School of Educational Information Technology, Central China Normal University, Wuhan, Hubei, China
会议名称:
2018 International Conference on Big Data Technologies, ICBDT 2018 and its Workshop, 2018 2nd International Conference on Business Information Systems, ICBIS 2018
会议时间:
May 18, 2018 - May 20, 2018
会议地点:
Hangzhou, China
会议论文集名称:
ICBDT '18: Proceedings of 2018 International Conference on Big Data Technologies
摘要:
This research started with the topic of educational technology of Zhihu. In order to explore other hot topics associated with it and the hotspots and trends in the parent topics of the annual questions, this research used the method of co-word analysis, social network analysis. Through digging topics network structure to understand the relationship between topics topics location and roles in the network. This research found that there were two major hot topics under educational technology topic. One was a hot spot of other professional topics closely linked to educational technology. The second was hot topics of interest to users under this topic. Through social network analysis, it was found that the structure of high-frequency topics network was relatively close and the structure of the parent topics network was loose, forming a higher tendency of centripetalism with educational technology as the core, and the topic with the core position in the network also had strong intermediary. The hotspot for future parent topics will be those at the edge of the high-frequency parent topics network, but in line with the current research hotspots in the field of educational technology.
摘要:
Intelligent methods are needed to organize the large amount of teaching and learning resources, one important aspect is to plan the learning path. According to the existing research, ant colony algorithm showed great advantages in learning path planning. Different from the traditional ant colony algorithm, Mahalanobis distance was adopted to calculate the distance between the data in the improved ACO. This paper proposed a method to recommend learning path using an improved ant colony algorithm based on a novel coordinate system. Also, In order to transform the unmeasurable concept map and information in syllabus into measurable data, a novel coordinate system was built to draw points which represent the teaching or learning units in it. The experimental results showed that this method can recommend an efficient learning path.
作者机构:
[Han, Meimei] Cent China Normal Univ, Educ Informat Technol Collaborat Innovat Ctr, Wuhan, Hubei, Peoples R China.;[Liu, ChunMiao; Tong, Mingwen; Chen, Mengyuan] Cent China Normal Univ, Sch Educ Informat Technol, Wuhan, Hubei, Peoples R China.;[Liu, JiaMin] Cent China Normal Univ, Natl Engn Res Ctr E Learning, Wuhan, Hubei, Peoples R China.
会议名称:
6th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI)
会议时间:
JUL 09-13, 2017
会议地点:
Hamamatsu, JAPAN
会议主办单位:
[Han, Meimei] Cent China Normal Univ, Educ Informat Technol Collaborat Innovat Ctr, Wuhan, Hubei, Peoples R China.^[Tong, Mingwen;Chen, Mengyuan;Liu, ChunMiao] Cent China Normal Univ, Sch Educ Informat Technol, Wuhan, Hubei, Peoples R China.^[Liu, JiaMin] Cent China Normal Univ, Natl Engn Res Ctr E Learning, Wuhan, Hubei, Peoples R China.
摘要:
With data accumulating rapidly in education, it is possible for researchers to predict students' academic performance. Among the prediction models, the machine learning model is outstanding. In this study, the model based on ensemble algorithm, AdaBoost was proposed to predict the classes of students. Experiments were carried out to compare the model to other models such as decision tree, neural network, random forests and SVM. The results showed the model presented the best prediction performance.