Fast vehicle detection based on evolving convolutional neural network

Abstract

With the rapid development of pattern recognition, computer vision and artificial intelligence technology, vehicle detection, traffic, and public safety, are core areas to extract direct benefits. To identify and locate a fast-moving vehicle, with high positional accuracy an evolving algorithm, based on convolutional neural networks, is proposed. Enhanced feature extraction is achieved by embedding our framework with a computationally cost-efficient proposal network to generate initial anchor boxes as well as to discard unlikely regions; feature fusion technology was used to extract hyper features, refine the identification and locate the vehicle, as well as improve the quality and accuracy of vehicle detection. Finally, we evaluated our network performance against the recent DETRAC benchmark [1] as well as using vehicle data sets collected by ourselves. The outcome of this study indicates a significant improvement over the state-of-the-art Faster RCNN by 9.61%, which fully highlights the effectiveness of the proposed algorithm.