PROJECT TITLE :

Semi-automatic annotation samples for vehicle type classification in urban environments

ABSTRACT:

Data collection, and especially data annotation, are surprisingly time consuming and pricey tasks for vehicle classification. Annotation is employed to label examples of vehicles, manually outlining their shapes and assigning their correct classification, to be used in classifier training and performance evaluation. This study presents a semi-automatic approach for the annotation of the vehicle samples recorded from roadside CCTV video cameras. Vehicles are detected by using automatic image analysis and classified into four main classes: automobile, van, bus and motorbike/bicycle by employing a vehicle observation vector constructed from the dimensions, the form and the appearance features. Unsupervised K-suggests that clustering is used to automatically compute an initial class label for each detected vehicle. Then, in an iterative process, the output scores of a linear support vector machines classifier are used to identify the low confidence samples, for which the annotations are considered for manual correction. Experimental results are presented for both synthetic and real datasets to demonstrate the effectiveness and the potency of the authors approach, which significantly reduces the time required to generate an annotated dataset. The tactic is general enough that it can be employed in other classification problems and domains that use a manually-created ground-truth.