Learning-To-Detect-Unseen-Object-Classes-by-Between-Class-Attribute-Transfer
In this project, we study the paper entitled "Learning to detect unseen object classes by between-class attribute transfer". The aim is to understand the out- put prediction methods developed by the authors. The study copes with the problem of object classification for computer vision tasks when training and test classes are disjoint. Attribute-based classification technique is introduced to handle such tasks. The object detection relies on human-specified high-level description of target object. The description is based upon semantic attributes which allow to infer new classes without any additional training phase because the inner structure of the description has as property knowledge transfer.
Original paper : http://ftp.idiap.ch/pub/courses/EE-700/material/28-11-2012/lampert-cvpr2009.pdf
Class taught by : Florence d’Alché-Buc http://perso.telecom-paristech.fr/~fdalche/Site/index.html