Robust object representation by boosting-like deep learning architecture

Abstract

Northumbria University has developed Northumbria Research Link (NRL) to enable users to access the University's research output. Copyright © and moral rights for items on NRL are retained by the individual author(s) and/or other copyright owners. Single copies of full items can be reproduced, displayed or performed, and given to third parties in any format or medium for personal research or study, educational, or not-for-profit purposes without prior permission or charge, provided the authors, title and full bibliographic details are given, as well as a hyperlink and/or URL to the original metadata page. The content must not be changed in any way. Full items must not be sold commercially in any format or medium without formal permission of the copyright holder. The full policy is available online: http://nrl.northumbria.ac.uk/policies.html This document may differ from the final, published version of the research and has been made available online in accordance with publisher policies. To read and/or cite from the published version of the research, please visit the publisher's website (a subscription may be required.) Abstract. This paper presents a new deep learning architecture for robust object representation, aiming at efficiently combining the proposed synchronized multi-stage feature (SMF) and a boosting-like algorithm. The SMF structure can capture a variety of characteristics from the inputting object based on the fusion of the handcraft features and deep learned features. With the proposed boosting-like algorithm, we can obtain more convergence stability on training multi-layer network by using the boosted samples. We show the generalization of our object representation architecture by applying it to undertake various tasks, i.e. pedestrian detection and action recognition. Our approach achieves 15.89% and 3.85% reduction in the average miss rate compared with ACF and JointDeep on the largest Caltech dataset, and acquires competitive results on the MSRAction3D dataset.

Point-to-Set Distance Metric Learning on Deep Representations for Visual Tracking

“…With the great success of CNNs on tasks such as image classification [22]- [25], natural language processing [26], [27], object detection [28]- [30], and image segmentation [31], [32], several trackers based on CNNs have been proposed. Fan et al [33] pre-train a network using the location and the appearance information of the object of interest to extract both spatial and temporal features.…”

New Keypoint Matching Method Using Local Convolutional Features for Power Transmission Line Icing Monitoring

“…Assume there are L layers in a CNN model, and k feature maps in the l th layer, where l = 1, 2, ……, L . At a certain layer, the previous layer’s feature maps are convolved with the learning filter kernel and put through the activation function F (·) to output the feature maps [ 25 ]. Thus, the k th feature map of the l th layer can be computed as: where is n th feature map in ( l − 1 )th layer, * represents the convolutional operation, is an additive bias of n th feature map in l th layer, I k denotes all the input convolved images of the k th feature map.…”

Scene text detection via extremal region based double threshold convolutional network classification

“…Gupta et al [ 23 ] developed a Fully-Convolutional Regression Network (FCRN) trained with synthetic images which performs both text detection and bounding box regression. A robust object representation which is a fusion of handcraft features and deep learned features is proposed in [ 44 ].…”