This project aims to tackle the problem of egocentric activity recognition based on the information available from two modalities which are video and eye tracking data. It achieves this by the fusion of multi stream convnet architecture to learn the spatial and temporal features from video data. It also makes use of the object saliency detection obtained from eye tracking data to further improve the identification of activity based on its surrounding.

Organization

Student

Gyanesh Malhotra

Mentors

  • Mehul Bhatt
  • Rajesh Kasturirangan
  • Jakob Suchan
close

2018