Showing 1–1 of 1 results for author: Baudach, D
-
Robust Deep-Learning-Based Road-Prediction for Augmented Reality Navigation Systems
Authors:
Matthias Limmer,
Julian Forster,
Dennis Baudach,
Florian Schüle,
Roland Schweiger,
Hendrik P. A. Lensch
Abstract:
This paper proposes an approach that predicts the road course from camera sensors leveraging deep learning techniques. Road pixels are identified by training a multi-scale convolutional neural network on a large number of full-scene-labeled night-time road images including adverse weather conditions. A framework is presented that applies the proposed approach to longer distance road course estimat…
▽ More
This paper proposes an approach that predicts the road course from camera sensors leveraging deep learning techniques. Road pixels are identified by training a multi-scale convolutional neural network on a large number of full-scene-labeled night-time road images including adverse weather conditions. A framework is presented that applies the proposed approach to longer distance road course estimation, which is the basis for an augmented reality navigation application. In this framework long range sensor data (radar) and data from a map database are fused with short range sensor data (camera) to produce a precise longitudinal and lateral localization and road course estimation. The proposed approach reliably detects roads with and without lane markings and thus increases the robustness and availability of road course estimations and augmented reality navigation. Evaluations on an extensive set of high precision ground truth data taken from a differential GPS and an inertial measurement unit show that the proposed approach reaches state-of-the-art performance without the limitation of requiring existing lane markings.
△ Less
Submitted 31 May, 2016;
originally announced May 2016.
