Pedestrian Detection Using Visual Saliency and Deep Learning.

This paper explores the use of Visual Saliency to detect pedestrian for images. Our method computes visual saliency map from the image, then the input image is multiplied by the multiplied visual salient map and product is fed to the Convolutional Neural Network. Using deeply supervised network with short connections (DSS) to compute visual saliency map and pedestrian detection is carried out using Faster R-CNN. DSS Network is trained on MSRA and ECSSD datasets, Faster R-CNN is trained on three challenging databases - Penn-Fudan, INRIA and Daimler datasets. Experimental results demonstrate that the proposed achieves state-of-the-art performance on Penn-Fudan Dataset with 91% detection accuracy and it achieves average miss-rate of 15% on the INRIA Dataset and achieves average miss-rate of 28% on the Daimler Dataset. [ABSTRACT FROM AUTHOR]