Supervision : Maaike de Boer

Location : The Machine learning department of Radboud University in Nijmegen, the Netherlands

Author : Camille Escher

Start date : 15/02/16

Main goal : Build a model and implement a prototype to better recognize events in videos by using the temporal relations of the sub-events sequence of events.

Progress : The basis (3 keyframes, 5 possible sub-events, one sub-event of interest per keyframe) of the method is working on Ubuntu for the 2 classes case. Need to evaluate the complexity and to generalize the method.

Tools and language :

• Ubuntu 15.04 (OS)
• Python 2.7.9 (Language)
• Pdb (Debugger)
• VIM (Editor)
• Scikit-learn (Library)