Publication

Abstract : pObject Proposal methods are pre-processing steps for detecting objects and vehicles in motion in intelligent transportation systems. Most of the object detection algorithms in literature are developed on daytime datasets and do not produce satisfactory results for nighttime images. This paper presents an approach to generate object proposal for detecting vehicles at night time. Different feature weights, proposals, sliding window are combined to generate a set of proposals to locate probable regions for vehicle detection at night-time. Sun Yat-sen University benchmark dataset is used for validation. This dataset contains images of night time vehicles in two different conditions. One being the vehicles on urban road with lamps and the other on urban road without lamps. © 2017 IEEE./p