Project: Comparison of Feature Extraction Approaches For Image Classification

Team: Dr. Harald Sack, Christian Hentschel(M.Sc.), Timur Pratama Wiradarma

Research institution: Hasso-Plattner-Institute Potsdam

Abstract: This project will compare two competing approaches for image classification Bag-of-Visual-Words (BoVW) and Convolutional Neural Networks (CNN). While CNNs have proven to provide better results recently, BoVW is assumed to outperform CNNs when the number of training data is below a certain threshold. As one result, we will determine this threshold.

Last modified 7 years ago Last modified on Nov 4, 2014 2:58:25 PM