Data-driven Radar Processing Using a Parametric Convolutional Neural Network for Human Activity Classification

Stadelmayer, Thomas; Santra, Avik

doi:10.36227/techrxiv.12896108.v2

Cited by 3 publications

(11 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are also some works that use real-valued input. Stadelmayer and Santra proposed using a parametric convolutional neural network (2D Sinc Filter and 2D Wavelet Filter) [28] to extract Range and Doppler information from raw radar data. Ye et al proposed using two real-valued convolutional layers with Fourier initialization for human-activity classification [37,38] on Continuous-wave radar data.…”

Section: End-to-end Mmwave Radar Gesture and Activity Recognition Met...mentioning

confidence: 99%

“…We also use the D-T pre-processing + 2DCNN classifier model on HGR task and R-D-T pre-processing + 3DCNN classifier model on HAR task here, as they are the most commonly seen models in previous works. Besides, learnable D-T pre-processing can be directly compared to RadarNet [38] and learnable R-D-T pre-processing can be directly compared to 2D Sinc/Wavelet Filters [28]. As mentioned in Section 2, RadarNet and 2D Sinc/Wavelet Filters are two previously proposed methods for end-to-end radar signal processing.…”

Section: Impact Of Network Structurementioning

confidence: 99%

“…(5) Fully-Complex Network (FC), where we do not take modulus after pre-processing, and use a fully complex version of the downstream classifier; (6) RadarNet [38]; and (7) 2D Sinc Filters and 2D Wavelet Filters [28].…”

Section: Impact Of Network Structurementioning

confidence: 99%

“…Besides, we use log operation instead of tanh which is suggested in the original paper, as the network does not converge when using tanh in our case. For Sinc and Wavelet Filters [28], we apply it on each frame and use the frame number as an additional dimension, since the radar configuration is different in this work. Besides, we find that the hyper-parameters greatly impact the Sinc/Wavelet Filter performance, so we try different combinations and select the best one.…”

Section: Impact Of Network Structurementioning

confidence: 99%

“…The difficulty here lies in that the raw radar ADC samples are complex values corresponding to the downmixed baseband, and both the magnitude and the phase are important. In previous works on end-to-end learning for radar applications, either only the real part is used [28,38], which loses some information; or the real part and the imaginary part are treated as two channels [41], which loses the physical meaning, and using a specific network structure could make it less universal.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

CubeLearn: End-to-end Learning for Human Motion Recognition from Raw mmWave Radar Signals

Zhao¹,

Lu²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

mmWave FMCW radar has attracted huge amount of research interest for human-centered applications in recent years, such as human gesture/activity recognition. Most existing pipelines are built upon conventional Discrete Fourier Transform (DFT) pre-processing and deep neural network classifier hybrid methods, with a majority of previous works focusing on designing the downstream classifier to improve overall accuracy. In this work, we take a step back and look at the pre-processing module. To avoid the drawbacks of conventional DFT pre-processing, we propose a learnable preprocessing module, named CubeLearn, to directly extract features from raw radar signal and build an end-to-end deep neural network for mmWave FMCW radar motion recognition applications. Extensive experiments show that our CubeLearn module consistently improves the classification accuracies of different pipelines, especially benefiting those previously weaker models. We provide ablation studies on initialization methods and structure of the proposed module, as well as an evaluation of the running time on PC and edge devices. This work also serves as a comparison of different approaches towards data cube slicing. Through our task agnostic design, we propose a first step towards a generic end-to-end solution for radar recognition problems. CCS CONCEPTS• Computing methodologies → Machine learning.

show abstract

Section: End-to-end Mmwave Radar Gesture and Activity Recognition Met...mentioning

confidence: 99%

Section: Impact Of Network Structurementioning

confidence: 99%

Section: Impact Of Network Structurementioning

confidence: 99%

Section: Impact Of Network Structurementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

CubeLearn: End-to-end Learning for Human Motion Recognition from Raw mmWave Radar Signals

Zhao¹,

Lu²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Universal Adversarial Attacks on the Raw Data From a Frequency Modulated Continuous Wave Radar

Valtl

Issakov

2022

IEEE Access

View full text Add to dashboard Cite

This work is a result of a collaboration between the projects "KI-Flex" (project number 16ES1027), funded by the German Federal Ministry of Education and Research (BMBF) within the founding program Microelectronic from Germany innovation driver, the "KoSi" project, (project number 01MM20011C) funded by the German Federal Ministry for Digital and Transport (BMDV) within the funding program "Automatisiertes, Vernetztes Fahren" and the project "TEACHING" (project number 871385) founded by the Horizon 2020 program."ABSTRACT As more and more applications rely on Artificial Intelligence (AI), it is inevitable to explore the associated safety and security risks, especially for sensitive applications where physical integrity is at risk. One of the most interesting challenges that come with AI are adversarial attacks being a wellresearched problem in the visual domain, where a small change in the input data can cause the Neural Network (NN) to make an incorrect prediction. In the radar domain, AI is not that widespread yet but the results that AI applications produce are very promising, which is why more and more applications based on it are being used. This work presents three possible attack methods that are particularly suitable for the radar domain. The developed algorithms generate universal adversarial attack patches for all sorts of radar applications based on NN. The main goal of the algorithms, apart from the computation of universal patches, is the identification of sensitive areas in the raw radar data input which than can be examined more closely. To the best of our knowledge, this is the first work that deals with calculating universal patches on raw radar data, which is of great importance especially for interference analysis. The developed algorithms have been verified on two data sets. One in the field of autonomous driving where the attacks lead to a steering misprediction of up to 0.3 for the steering value which is within [-1,1], with the results also being successfully tested on a demonstrator. The other data set originated from a gesture recognition task, where the attacks decreased the accuracy, originally at 97.0% up to a minimum of 16.5%, which is slightly above 12.5% being the accuracy for a purely random prediction.INDEX TERMS adversarial attacks, artificial neural networks, autonomous vehicles, edge computing, object recognition, radar applications, real-time systems

show abstract

CubeLearn: End-to-End Learning for Human Motion Recognition From Raw mmWave Radar Signals

Zhao

Wang

et al. 2023

IEEE Internet Things J.

View full text Add to dashboard Cite

mmWave FMCW radar has attracted huge amount of research interest for human-centered applications in recent years, such as human gesture and activity recognition. Most existing pipelines are built upon conventional Discrete Fourier Transform (DFT) pre-processing and deep neural network classifier hybrid methods, with a majority of previous works focusing on designing the downstream classifier to improve overall accuracy. In this work, we take a step back and look at the pre-processing module. To avoid the drawbacks of conventional DFT pre-processing, we propose a complex-weighted learnable pre-processing module, named CubeLearn, to directly extract features from raw radar signal and build an end-to-end deep neural network for mmWave FMCW radar motion recognition applications. Extensive experiments show that our CubeLearn module consistently improves the classification accuracies of different pipelines, especially benefiting those simpler models, which are more likely to be used on edge devices due to their computational efficiency. We provide ablation studies on initialization methods and structure of the proposed module, as well as an evaluation of the running time on PC and edge devices. This work also serves as a comparison of different approaches towards data cube slicing. Through our task agnostic design, we propose a first step towards a generic end-to-end solution for radar recognition problems.

show abstract

Data-driven Radar Processing Using a Parametric Convolutional Neural Network for Human Activity Classification

Cited by 3 publications

References 16 publications

CubeLearn: End-to-end Learning for Human Motion Recognition from Raw mmWave Radar Signals

CubeLearn: End-to-end Learning for Human Motion Recognition from Raw mmWave Radar Signals

Universal Adversarial Attacks on the Raw Data From a Frequency Modulated Continuous Wave Radar

CubeLearn: End-to-End Learning for Human Motion Recognition From Raw mmWave Radar Signals

Contact Info

Product

Resources

About