Potential of radar for static object classification using deep learning methods

Lombacher, Jakob; Hahn, Markus; Dickmann, Jürgen; Wöhler, Christian

doi:10.1109/icmim.2016.7533931

Cited by 69 publications

(39 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Static object classification (e.g. parked cars, traffic signs) has been shown with target-level [14] and with low-level data [15]. We will focus only on methods addressing moving road users.…”

Section: Related Workmentioning

confidence: 99%

CNN Based Road User Detection Using the 3D Radar Cube

Palffy

Dong

Kooij

et al. 2020

IEEE Robot. Autom. Lett.

132

View full text Add to dashboard Cite

This paper presents a novel radar based, singleframe, multi-class detection method for moving road users (pedestrian, cyclist, car), which utilizes low-level radar cube data. The method provides class information both on the radar targetand object-level. Radar targets are classified individually after extending the target features with a cropped block of the 3D radar cube around their positions, thereby capturing the motion of moving parts in the local velocity distribution. A Convolutional Neural Network (CNN) is proposed for this classification step. Afterwards, object proposals are generated with a clustering step, which not only considers the radar targets' positions and velocities, but their calculated class scores as well. In experiments on a real-life dataset we demonstrate that our method outperforms the state-of-the-art methods both target-and objectwise by reaching an average of 0.70 (baseline: 0.68) target-wise and 0.56 (baseline: 0.48) object-wise F1 score. Furthermore, we examine the importance of the used features in an ablation study.

show abstract

Section: Related Workmentioning

confidence: 99%

CNN Based Road User Detection Using the 3D Radar Cube

Palffy

Dong

Kooij

et al. 2020

IEEE Robot. Autom. Lett.

132

View full text Add to dashboard Cite

show abstract

“…In [26], the Multi-Layer Perceptron (MLP) neural network is used to classify vehicles, people, trees, trunks, light poles and buildings. Moreover, deep learning is used for object classification [19] and tracking [19], [20]. An intelligent surveillance system for smartphones is presented using deep learning in [22] where the system consists of a detection module and a classification module.…”

Section: State Of Art On Obstacle Classificationmentioning

confidence: 99%

Deep Multi-Layer Perceptron-Based Obstacle Classification Method From Partial Visual Information: Application to the Assistance of Visually Impaired People

2020

View full text Add to dashboard Cite

How to navigate safely, recognize encountered obstacles, and move independently from one location to another in unknown environments are some of the challenges that face visually impaired people. By proposing a solution towards overcoming these challenges, this work will be of most importance to visually impaired people. In this work, we propose a consistent, reliable and robust smartphone-based method to classify obstacles in unknown environments from partial visual information based on computer vision and machine learning techniques. Our proposed method handles high levels of noise and bad resolution in frames captured from a phone camera. In addition, our proposed method offers maximum flexibility to users and use the least expensive equipment possible. Moreover, by leveraging on deep-learning techniques, the proposed method enables semantic categorization in order to classify obstacles and increase the awareness of the explored environment. The efficiency of the work has been experimentally measured on a variety of experiments studies on different complex scenes. It records high accuracy of [90.2 % ]. INDEX TERMS Image analysis, image classification, supervised learning, mobile applications.

show abstract

“…In contrast, Lombacher et al [19] use the power spectrum alone to recognise a significant number of roadside objects with a 76 GHz radar system. There are several differences to the current paper.…”

Section: Related Workmentioning

confidence: 99%