Video Classification using Slow Feature Analysis and Neural Network

International Journal of Electrical and Electronics Engineering

© 2017 by SSRG - IJEEE Journal

Volume 4 Issue 7

Year of Publication : 2017

Authors : Bilkis .A. Inamdar, Prof. Ujwal. Harode

10.14445/23488379/IJEEE-V4I7P104

How to Cite?

Bilkis .A. Inamdar, Prof. Ujwal. Harode, "Video Classification using Slow Feature Analysis and Neural Network," SSRG International Journal of Electrical and Electronics Engineering, vol. 4,  no. 7, pp. 1-5, 2017. Crossref, https://doi.org/10.14445/23488379/IJEEE-V4I7P104

Abstract:

Videos concepts based feature classification is a process, where the classification of videos will be classified on the basis of video classes. For classification initially frames will be extracted from the dataset then these frames will be processes for feature extraction. In our context we made an approach towards the slowness principle which has a fundamental ability to perform as close to the function of human brain to visualize the change in the surrounding and has the tendency to create slow motion output from the fast varying input signal due to the slowness it provides with a stable output as well. Using the frames from the input videos from the datasets features are extracted. Initially V1-like features is obtained as a result these features are then processed to slow feature analysis method which determines different category of video sections composed of natural scenes from the dataset input. Hence the system will be trained and motion components are computed thus using the knowledge based classification algorithm this videos will be classified according to their classes using a classifier i.e. feed forward back propagation neural network.

Keywords:

Unsupervised learning, motion analysis, neural network.

References:

[1] Christian Thériault, Nicolas Thome, Matthieu Cord, and Patrick Pérez, “Perceptual Principles for Video Classification With Slow Feature Analysis,” IEEE journal of selected topics in signal processing, VOL. 8, NO. 3, JUNE 2014
[2] K. G. Derpanis, M. Lecce, K. Daniilidis, and R. P. Wildes, “Dynamic scene understanding: the role of orientation features in space and time in scene classification,” in Proc. CVPR, 2012
[3] S. Klampfl andW.Maass, “A theoretical basis for emergent pattern discrimination in neural systems through slow feature extraction,” Neural Comput., vol. 22, no. 12, pp. 2979–3035, Dec. 2010.
[4] A.A. Efros,A. C. Berg, G. Mori, and J. Malik, “Recognizing action at a distance,” in Proc. ICCV, 2003.
[5] S. Soatto, G. Doretto, and Y. N. Wu, “Dynamic textures,” in ICCV, 2001.
[6] S. S. Beauchemin and J. L. Barron, The Computation of Optical Flow. New York, NY, USA: ACM, 1995.
[7] L. Wiskott and T Sejnowski, “Slow feature analysis: unsupervised learning of invariances,” Neural Comput., vol. 14, pp. 715–770, 2002.