Vehicle Silhouettes Classification
PA196 Pattern Recognition Project Topic
Ján Bella
Masaryk University in Brno
October 27, 2015
Ján Bella Masaryk University in Brno
Vehicle Silhouettes Classification
Problem
Dataset: Statlog (Vehicle Silhouettes) Data Set 1
∙ 4 classes of vehicles: double-decker, van, 2 types of cars
∙ originally images, but extracted features available
∙ 846 examples, each of 18 attributes
Original approach: Siebert, J.P. Vehicle Recognition Using Rule
Based Methods. Turing Institute, Glasgow, 1987.
∙ the rule tree classification, compared to k-nearest neighbour
and minimum distance classifier
∙ not available anymore
1
http://archive.ics.uci.edu/ml/datasets/Statlog+%28Vehicle+Silhouettes%29
Ján Bella Masaryk University in Brno
Vehicle Silhouettes Classification
Goals
∙ pair-wise classification on classes and evaluating the results
∙ compare performance of different methods:
∙ LDA
∙ k-nearest neighbour
∙ possible extensions (if I’ll feel confident enough)
∙ SVM
∙ multi-class classification
Ján Bella Masaryk University in Brno
Vehicle Silhouettes Classification