Evaluation of Bag of Visual Words for Category Level Object Recognition

Abstract

Object recognition in a large scale collection of images has become an important application in machine vision. The recent advances in the object or image recognition for classification of objects shows that Bag-of-visual words approach is a better method for image classification problems. In this work, the effect of different possible parameters and performance evaluation of Bag of visual words approach in terms of their recognition performance such as Accuracy rate, Precision and F1 measure using 8 different classes of real world datasets that are commonly used in restaurant applications is explored. The system presented here is based on visual vocabulary. Features are extracted, clustered, trained and evaluated on an image database of 1600 images of different categories. To validate the obtained results,a performance evaluation on vehicle datasetsunder SURF and SIFT descriptors with Kmeans and K-medoid clustering and KNN classifier has been made. Among these SURF K-means performs better.