LEADER |
01828nam a22002297a 4500 |
001 |
2146184 |
041 |
|
|
|a eng
|
044 |
|
|
|b العراق
|
100 |
|
|
|9 738289
|a Ahmed, Wafaa Shihab
|e Author
|
245 |
|
|
|a Human Motion Generation Using Variational Autoencoder and Mixture Density Network
|
260 |
|
|
|b الجامعة العراقية - مركز البحوث والدراسات الإسلامية
|c 2023
|g يونيو
|m 1444
|
300 |
|
|
|a 632 - 643
|
336 |
|
|
|a بحوث ومقالات
|b Article
|
520 |
|
|
|b The modelling of Human motion is critical in several fields such as computer graphics, vision and virtual reality, with applications of interaction between human and computer, synthesizing motions, and motion generation for virtual reality. In this work the CNN-VAE and RNN-MDN models have been used for modelling the human motions by learning time-dependent representations to obtain short-term motion prediction and long-term human motion synthesis. The new generated human motion video achieved good quality. This proposed system has been implemented on the KTH dataset and Weizmann dataset to generate the boxing and waving motions.
|
653 |
|
|
|a الواقع الافتراضي
|a الحركات البشرية
|a توليف الحركات
|a التشفير التلقائي
|a الشبكات العصبية
|
692 |
|
|
|b Variational Autoencoder "VAE"
|b Convolution Neural Network "CNN"
|b Mixture Density Network "MDN"
|b Wavelet Transform "WT"
|
700 |
|
|
|9 660529
|a Karim, Abdulamir Abbdullah
|e Co-Author
|
773 |
|
|
|4 الدراسات الإسلامية
|6 Islamic Studies
|c 051
|e Journal of The Iraqi University
|f Maǧallat al-ğāmiʻaẗ al-islāmiyyaẗ
|l 003
|m ع60, ج3
|o 2164
|s مجلة الجامعة العراقية
|v 060
|x 1813-4521
|
856 |
|
|
|u 2164-060-003-051.pdf
|
930 |
|
|
|d y
|p y
|q n
|
995 |
|
|
|a IslamicInfo
|
999 |
|
|
|c 1395877
|d 1395877
|