Weakly Supervised Deep Learning Method for Vulnerable Road User Detection in FMCW Radar

Dimitrievski, Martin; Shopovska, Ivana; Hamme, David Van; Veelaert, Peter; Philips, Wilfried

doi:10.1109/itsc45102.2020.9294399

Cited by 18 publications

(31 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It also provides a signal processing SDK [23] to process and annotate radar data. The Ghent VRU dataset [49] collects radar data specifically for VRU detection. The sensor suite includes a TI AWR1243 radar, a camera, and a 16-beam LiDAR.…”

Section: Radar Datasetsmentioning

confidence: 99%

“…The CARRADA dataset [45] and RADDet dataset [46] apply logarithmic scaling to their radar data. The normalisation can be applied in different ways, including local power normalising in the Ghent VRU dataset [49], min-max scaling in the CRUW dataset [47], and Z-score standardisation in the RADDet dataset [46]. Here, we only summarize some operations that are explicitly mentioned.…”

Section: Radar Datasetsmentioning

confidence: 99%

See 1 more Smart Citation

Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges

Zhou

Liu

Zhao

et al. 2022

Sensors

View full text Add to dashboard Cite

With recent developments, the performance of automotive radar has improved significantly. The next generation of 4D radar can achieve imaging capability in the form of high-resolution point clouds. In this context, we believe that the era of deep learning for radar perception has arrived. However, studies on radar deep learning are spread across different tasks, and a holistic overview is lacking. This review paper attempts to provide a big picture of the deep radar perception stack, including signal processing, datasets, labelling, data augmentation, and downstream tasks such as depth and velocity estimation, object detection, and sensor fusion. For these tasks, we focus on explaining how the network structure is adapted to radar domain knowledge. In particular, we summarise three overlooked challenges in deep radar perception, including multi-path effects, uncertainty problems, and adverse weather effects, and present some attempts to solve them.

show abstract

Section: Radar Datasetsmentioning

confidence: 99%

Section: Radar Datasetsmentioning

confidence: 99%

Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges

Zhou

Liu

Zhao

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…Nevertheless, this functionality is well explored in automotive settings. For example, in [10], a deep convolutional neural network model (U-Net) was weakly trained on radar cubes data. The radar cube is a 3D data structure containing range-azimuth-Doppler maps.…”

Section: Obstacle Detection and Trackingmentioning

confidence: 99%

Radar Perception for Autonomous Unmanned Aerial Vehicles: a Survey

Corradi

Fioranelli

2022

System Engineering for Constrained Embedded Systems

View full text Add to dashboard Cite

The advent of consumer and industrial Unmanned Aerial Vehicles (UAVs), commonly referred to as drones, has opened business opportunities in many fields, including logistics, smart agriculture, inspection, surveillance, and construction. In addition, the autonomous operations of UAVs reduce risks by minimizing the time spent by human workers in harsh environments and lowering costs by automating tasks. For reliability and safety, the drones must sense and avoid potential obstacles and must be capable of safely navigating in unknown environments. UAVs' perception requires reliability in various settings, such as high dust levels, humidity, intense sun glare, dark, and fog that can severely obstruct many conventional sensing methods. Radar systems have unique strengths; they can reliably estimate how far an object is and measure its relative speed via the Doppler effect. In addition, because radars exploit radio waves to sense, they perform well in rain, fog, snow, or smoky environments. This stands in contrast to optical technologies, such as cameras or LIght Detection And Ranging (Lidars), which are more susceptible to the same challenges as the human eye. This survey paper aims to address the signal processing challenges for the exploitation of radar systems in unmanned aerial vehicles for advanced perception, considering recent integration trends and technology capabilities. The focus is on signal processing techniques for low-cost and power-efficient radar sensors, which operate onboard the UAVs in real-time to ensure their needs in terms of perception, situational awareness, and navigation. Additionally, we highlight the challenges that remain to be tackled and the opportunities that lie ahead in the search for a more efficient, safe, and autonomous way for UAVs to perceive and interact with the world. CCS CONCEPTS• Computer systems organization → System on a chip; Embedded hardware;This work is licensed under a Creative Commons Attribution International 4.0 License.

show abstract

“…Moreover, convolutional neural networks (CNNs) [ 7 , 8 ] or U-shaped neural networks (i.e., U-nets) [ 9 , 10 ] were used to detect targets on the range–velocity plane. Recently, deep learning techniques to replace the CFAR algorithm in the automotive MIMO FMCW radar system were also introduced in [ 11 , 12 ]. A U-net-based target detector was proposed in [ 11 ] for detecting a vulnerable road user on the range-angle (RA) map.…”

Section: Introductionmentioning

confidence: 99%

“…Recently, deep learning techniques to replace the CFAR algorithm in the automotive MIMO FMCW radar system were also introduced in [ 11 , 12 ]. A U-net-based target detector was proposed in [ 11 ] for detecting a vulnerable road user on the range-angle (RA) map. In addition, the authors in [ 12 ] compensated for the disadvantages of the conventional CFAR algorithm by replacing the peak detection step of the CFAR algorithm with the deep neural network.…”

Section: Introductionmentioning

confidence: 99%

Autoencoder-Based Target Detection in Automotive MIMO FMCW Radar System

Kang

Jang

Lee

2022

Sensors

View full text Add to dashboard Cite

In general, a constant false alarm rate algorithm (CFAR) is widely used to automatically detect targets in an automotive frequency-modulated continuous wave (FMCW) radar system. However, if the number of guard cells, the number of training cells, and the probability of false alarm are set improperly in the conventional CFAR algorithm, the target detection performance is severely degraded. Therefore, we propose a method using a convolutional neural network-based autoencoder (AE) to replace the CFAR algorithm in the multiple-input and multiple-output FMCW radar system. In the AE, the entire detection result is compressed at the encoder side, and only significant signal components are recovered on the decoder side. In this work, by changing the number of hidden layers and the number of filters in each layer, the structure of the AE showing a high signal-to-noise ratio in the target detection result is determined. To evaluate the performance of the proposed method, the AE-based target detection result is compared with the target detection results of conventional CFAR algorithms. As a result of calculating the correlation coefficient with the data marked with the actual target position, the proposed AE-based target detection shows the highest similarity with a correlation of 0.73 or higher.

show abstract

Weakly Supervised Deep Learning Method for Vulnerable Road User Detection in FMCW Radar

Cited by 18 publications

References 14 publications

Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges

Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges

Radar Perception for Autonomous Unmanned Aerial Vehicles: a Survey

Autoencoder-Based Target Detection in Automotive MIMO FMCW Radar System

Contact Info

Product

Resources

About