<> "The repository administrator has not yet configured an RDF license."^^ . <> . . . "Selecting active frames for action recognition with 3D convolutional network"^^ . "Recent applications of Convolutional Neural Networks, especially 3-Dimensional Convoltutional Neural Networks (3DCNNs) for human action recognition\r\n(HAR) in videos have widely used. In this paper, we\r\nuse a multi-stream framework which is a combination\r\nfrom separated networks with different kind of input\r\ngenerated from unique video dataset. To achieve the\r\nhigh results, firstly, we proposed a method to extract\r\nthe active frames (called Selected Active Frames -\r\nSAF) from a videos to build datasets for 3DCNNs in\r\nvideo classifying problem. Second, we deploy a new\r\napproach called Vote fusion which considered as an\r\neffective fusion method for ensembling multi-stream\r\nnetworks. From the various datasets generated from\r\nvideos, we extract frames by our method and feed\r\ninto 3DCNNs for feature extraction, then we carry out\r\ntraining and then fuse the results of softmax layers\r\nof these streams. We evaluate the proposed methods\r\non solving action recognition problem. These method\r\nare carried on three well-known datasets (HMFB51,\r\nUCF101, and KTH). The results are also compared to\r\nthe state-of-the-art results to illustrate the efficiency\r\nand effectiveness in our approach"^^ . "2018-09" . . . . . . . . . . . . . . . . . "Sugimoto"^^ . "Akihiro"^^ . "Sugimoto Akihiro"^^ . . "Thi Chau"^^ . "Ma"^^ . "Thi Chau Ma"^^ . . "The Duy"^^ . "Bui"^^ . "The Duy Bui"^^ . . "Tieu Binh"^^ . "Hoang"^^ . "Tieu Binh Hoang"^^ . . . . "2018 the 7th International Conference on Computer and Communication Engineering (ICCCE)"^^ . . . . . "Malaysia"^^ . . . . . . "Selecting active frames for action recognition with 3D convolutional network (PDF)"^^ . . . "ICCCE2018_BinhHT.pdf"^^ . . "HTML Summary of #3123 \n\nSelecting active frames for action recognition with 3D convolutional network\n\n" . "text/html" . . . "Information Technology (IT)"@en . .