Abstract: HEVC is the latest video coding standard aimed to compress double to that its predecessor standard H.264. Motion Estimation is one of the critical parts in the encoder due to the introduction of asymmetric motion partitioning and higher size of coding tree unit. In this paper, a design for an Integer Motion Estimator of HEVC is presented over specific hardware architecture for real time implementation. The implementation shows a new IME unit supporting asymmetric partitioning mode which significantly reduce the overall motion estimation processing time. The prototyped architecture has been designed in VHDL, synthesized and implemented using the Xilinx FPGA, Zynq-7000 xc7z020 clg484-1. The proposed design is able to process 30 fps at Full- HD and 15 fps at 2K resolution.

• Publication Date: 11-Dec-2015
• DOI: 10.15224/978-1-63248-077-4-09
• Views: 0
• Downloads: 0

Abstract: Educational Data Mining (EDM) techniques play an important role in understanding hidden students’ data patterns to improve the quality of teaching-learning professions. In machine learning, feature selection usually emerges as a preprocessing step to extract necessary and sufficiently small subset of features for predictive / decision-making type of learning tasks. In this study, authors decided to work only upon external (changeable) attributes of students by assigning weights that reflect their academic efforts put in for those attributes. The attribute precedence levels extracted student-wise by current FE model due to academic efforts put up by students in their ongoing course were compared with equivalently generated precedence relations from RELIEF method and it’s variant. The favorable model accuracies of these precedence relations when compared with RELIEF have given a new meaning to EDM objectives in the direction of individual student counseling encouraging them to appraise t

• Publication Date: 11-Dec-2015
• DOI: 10.15224/978-1-63248-077-4-10
• Views: 0
• Downloads: 0