Skip to content

Implementation of novel Convolutional neural network architecture for Action Recognition

Notifications You must be signed in to change notification settings

nelaturuharsha/Two-Way-Convolutional-Network

Repository files navigation

Action Recognition Convolutional Neural Network

Contributors: Sree Harsha Nelaturu, Aparna Krishnakumar, Rohan Pooniwala and Sourav Sharan.

Implementation of the research paper "Two-Stream Convolutional Networks for Action Recognition in Videos " https://arxiv.org/pdf/1406.2199.pdf

Approach: The approach involves building and training 2 convolutional neural networks simultaneously- one spatial network which captures action recognition on still images ina video frame by frame and a temproal network which computes motion between frames in a video by stacking optical flow field vectors on top of one another. The output of both the convolutional networks are then passed through a SVM to detect action in videos.

About

Implementation of novel Convolutional neural network architecture for Action Recognition

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •  

Languages