Slanted Stixels: A Way to Represent Steep Streets
Open Access
- 14 September 2019
- journal article
- research article
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 127 (11-12) , 1643-1658
Abstract
This work presents and evaluates a novel compact scene representation based on Stixels that infers geometric and semantic information. Our approach overcomes the previous rather restrictive geometric assumptions for Stixels by introducing a novel depth model to account for non-flat roads and slanted objects. Both semantic and depth cues are used jointly to infer the scene representation in a sound global energy minimization formulation. Furthermore, a novel approximation scheme is introduced in order to significantly reduce the computational complexity of the Stixel algorithm, and then achieve real-time computation capabilities. The idea is to first perform an over-segmentation of the image, discarding the unlikely Stixel cuts, and apply the algorithm only on the remaining Stixel cuts. This work presents a novel over-segmentation strategy based on a fully convolutional network, which outperforms an approach based on using local extrema of the disparity map. We evaluate the proposed methods in terms of semantic and geometric accuracy as well as run-time on four publicly available benchmark datasets. Our approach maintains accuracy on flat road scene datasets while improving substantially on a novel non-flat road dataset.Keywords
All Related Versions
Funding Information
- Ministerio de Economía y Competitividad (TIN2017-84553-C2-1-R, TIN2017-88709-R)
- Dirección General de Tráfico (SPIP2017-02237)
- Generalitat de Catalunya (2017-SGR-313)
- Sociedad Económica Barcelonesa de Amigos del País
- Generalitat de Catalunya (http://dx.doi.org/10.13039/501100002809)
This publication has 34 references indexed in Scilit:
- Fusion of laser and radar sensor data with a sequential Monte Carlo Bayesian occupancy filter2015
- Sensor fusion for semantic segmentation of urban scenes2015
- Object-Level Priors for Stixel Generation2014
- The Pascal Visual Object Classes Challenge: A RetrospectiveInternational Journal of Computer Vision, 2014
- Making Bertha Drive—An Autonomous Journey on a Historic RouteIEEE Intelligent Transportation Systems Magazine, 2014
- Exploiting the Power of Stereo Confidences2013
- Fast Stixel Computation for Fast Pedestrian Detection2012
- Towards a Global Optimal Multi-Layer Stixel Representation of Dense 3D Data2011
- Compression of time series by extracting major extremaJournal of Experimental & Theoretical Artificial Intelligence, 2010
- Model compression2006