We propose a novel method to fit and segment multistructural data via convex relaxation. Unlike greedy methodswhich maximise the number of inliers-this approach efficiently searches for a soft assignment of points to models by minimising the energy of the overall classification. Our approach is similar to state-of-the-art energy minimisation techniques which use a global energy. However, we deal with the scaling factor (as the number of models increases) of the original combinatorial problem by relaxing the solution. This relaxation brings two advantages: first, by operating in the continuous domain we can parallelize the calculations. Second, it allows for the use of different metrics which results in a more general formulation.We demonstrate the versatility of our technique on two different problems of estimating structure from images: plane extraction from RGB-D data and homography estimation from pairs of images. In both cases, we report accurate results on publicly available datasets, in most of the cases outperforming the state-of-the-art.
The classification of semantically meaningful road markings in images is an important and safety critical task for autonomous and semi-autonomous vehicles. However, beyond simple lane markings, real-time detection and interpretation of road markings is challenging as images are subject to occlusions, partial observations, lighting changes and differing weather conditions. Additionally, there is high variation in the road markings between countries and regions, which makes interpretation difficult. In this work we present a threefold approach to the semantic classification. Firstly, we employ a weakly supervised neural network to detect pixels belonging to road markings under different conditions. Subsequently, these pixels are classified into geometric primitives, from which we retrieve the semantic classes through a fast and parallel modelfitting algorithm that offers real-time performance. Unlike other methods in the literature that perform road marking classification independently, our proposed approach performs a joint classification leveraging the highly structured configurations that characterise urban traffic scenes. Consequently, we retrieve the underlying semantic classes under a variety of weather and lighting conditions as we demonstrate in our results.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.