Combining Deep Learning and Model-Based Methods for Robust Real-Time Semantic Landmark Detection

TL;DR

This paper introduces a hybrid approach combining deep learning and model-based methods for real-time semantic landmark detection, improving robustness and speed in challenging outdoor environments with limited training data.

Contribution

The paper presents a novel method integrating CNN-based classification with model-based techniques, reducing data requirements and enhancing real-time performance for landmark detection.

Findings

Outperforms pure learning-based 3D detectors in challenging scenarios

Achieves faster processing speeds while maintaining accuracy

Demonstrates robustness to environmental changes like lighting and vegetation

Abstract

Compared to abstract features, significant objects, so-called landmarks, are a more natural means for vehicle localization and navigation, especially in challenging unstructured environments. The major challenge is to recognize landmarks in various lighting conditions and changing environment (growing vegetation) while only having few training samples available. We propose a new method which leverages Deep Learning as well as model-based methods to overcome the need of a large data set. Using RGB images and light detection and ranging (LiDAR) point clouds, our approach combines state-of-the-art classification results of Convolutional Neural Networks (CNN), with robust model-based methods by taking prior knowledge of previous time steps into account. Evaluations on a challenging real-wold scenario, with trees and bushes as landmarks, show promising results over pure learning-based state-of-the-art 3D detectors, while being significant faster.