Support Vector Machine based Vehicle Make and Model Recognition System

Vehicle analysis is a very useful component in various real world applications. In this paper, we have developed a Vehicle Make and Model Recognition (VMMR) system using Support Vector Machine (SVM). Scale Invariant Feature Transform (SIFT) and Speed-Up Robust Transform (SURF) are used to extract local features from an image. Bag-of-Features (BoF) model is used to create visual dictionaries and convert the local image features into global image feature representation. Multiple dictionaries of different sizes are created for both features; SIFT and SURF and the dataset is coded using these dictionaries to determine the best size for the visual dictionary. NTOU-MMR is a publicly available vehicle dataset which we have used to evaluate the performance of proposed VMMR system. 92% recognition rate is achieved by using the proposed VMMR system.