Video Processing
A. Akbari; H. Farsi; S. Mohamadzadeh
Abstract
Background and Objectives: Video processing is one of the essential concerns generally regarded over the last few years. Social group detection is one of the most necessary issues in crowd. For human-like robots, detecting groups and the relationship between members in groups are important. Moving in ...
Read More
Background and Objectives: Video processing is one of the essential concerns generally regarded over the last few years. Social group detection is one of the most necessary issues in crowd. For human-like robots, detecting groups and the relationship between members in groups are important. Moving in a group, consisting of two or more people, means moving the members of the group in the same direction and speed. Methods: Deep neural network (DNN) is applied for detecting social groups in the proposed method using the parameters including Euclidean distance, Proximity distance, Motion causality, Trajectory shape, and Heat-maps. First, features between pairs of all people in the video are extracted, and then the matrix of features is made. Next, the DNN learns social groups by the matrix of features.Results: The goal is to detect two or more individuals in social groups. The proposed method with DNN and extracted features detect social groups. Finally, the proposed method’s output is compared with different methods.Conclusion: In the latest years, the use of deep neural networks (DNNs) for learning and detecting has been increased. In this work, we used DNNs for detecting social groups with extracted features. The indexing consequences and the outputs of movies characterize the utility of DNNs with extracted features.
Artificial Intelligence
S. F. Mirmousavi; S. Kianian
Abstract
Background: The link prediction issue is one of the most widely used problems in complex network analysis. Link prediction requires knowing the background of previous link connections and combining them with available information. The link prediction local approaches with node structure objectives are ...
Read More
Background: The link prediction issue is one of the most widely used problems in complex network analysis. Link prediction requires knowing the background of previous link connections and combining them with available information. The link prediction local approaches with node structure objectives are fast in case of speed but are not accurate enough. On the other hand, the global link prediction methods identify all path structures in a network and can determine the similarity degree between graph-extracted entities with high accuracy but are time-consuming instead. Most existing algorithms are only using one type of feature (global or local) to represent data, which not well described due to the large scale and heterogeneity of complex networks.Methods: In this paper, a new method presented for Link Prediction using node embedding due to the high dimensions of real-world networks. The proposed method extracts a smaller model of the input network by getting help from the deep neural network and combining global and local nodes in a way to preserve the network's information and features to the desired extent. First, the feature vector is being extracted by an encoder-decoder for each node, which is a suitable tool for modeling complex nonlinear phenomena. Secondly, both global and local information concurrently used to improve the loss function. More obvious, the clustering similarity threshold considered as the local criterion and the transitive node similarity measure used to exploit the global features. To the end, the accuracy of the link prediction algorithm increased by designing the optimization operation accurately.Results: The proposed method applied to 4 datasets named Cora, Wikipedia, Blog catalog, Drug-drug-interaction, and the results are compared with laplacian, Node2vec, and GAE methods. Experimental results show an average accuracy achievement of 0.620, 0.723, 0.875, and 0.845 on the mentioned datasets, and confirm that the link prediction can effectively improve the prediction performance using network embedding based on global similarity.
