To enable autonomous driving, a semantic knowledge of the environment is unavoidable. We therefore introduce a multiclass classifier to determine the classes of an object relying solely on radar data. This is a challenging problem as objects of the same category have often a diverse appearance in radar data. As classification methods a random forest classifier and a deep convolutional neural network are evaluated. To get good results despite the limited training data available, we introduce a hybrid approach using an ensemble consisting of the two classifiers. Further we show that the accuracy can be improved significantly by allowing a lower detection rate.
Object classification in radar using ensemble methods
2017-03-01
4010773 byte
Conference paper
Electronic Resource
English
Buried object classification using holographic radar
Tema Archive | 2012
|Spatial object classification by radar measurements
Online Contents | 2010
|Radar object classification based on radar cross-section data
European Patent Office | 2024
|Radar object classification based on radar cross-section data
European Patent Office | 2025
|