Vehicle Make and Model Recognition using Auto Extracted Parts

After vehicle detection and vehicle type recognition, it is vehicle make and model recognition (VMMR) that has attracted researchers attention in the last decade. This problem is known as a hard classification problem due to the large number of classes and small inter-class distance. The present paper proposes a new approach for VMMR. The proposed approach includes a new part-based VMMR viewpoint and a new method for auto extraction of parts. This viewpoint concentrates on discriminative parts of vehicle like lights, grilles and logo for classification of different make and models. The Histogram of Oriented Gradients (HOG) and Support Vector Machine (SVM) have been used for feature extraction and classification tasks respectively. HOG is robust to lighting changes and small location variability of vehicle parts. For evaluation purposes, a novel dataset (BVMMR) including 676 images from frontal and rear view of 19 different classes of vehicles have been prepared and fully annotated. The experimental results showed the effectiveness of the part-based approach in compare to the traditional approaches. Moreover, the high accuracy gained from auto extracted parts convinced us that we are on a right track. The proposed method achieved 100% accuracy on both frontal and rear view

کلیدواژه ها:

Fine-grained Recognition ، Object Classification ، Vehicle Make and Model Recognition (VMMR) ، Deformable Part-based Model ، Auto Extraction of Parts