A framework using matlab and support vector machine to recoginse landmarks (building) in an area
A dataset of 5000 images was taken from standford university and put through SURF algorithm to extract features in the image and the features were clustered bag of words (K- means clustering) and the clustered features were trained using Support Vector Machine
Also a dataset of famous landmarks in New Delhi, India was collected through google and images were preprocessed and ran through the classifier and an accuracy of 95% was achieved.
PS: For this system my dataset is in the form one major folder containing seperate folders for each building and dataset is automatically divided into training set(80%), testset(10%) and validation set(10%).
Dataset Link:https://purl.stanford.edu/vn158kj2087