Pith. sign in

REVIEW 4 major objections 6 minor 63 references

Through-the-Wall Radar Human Activity Recognition WITHOUT Using Neural Networks

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A completely non-neural pipeline of SIFT, active-contour segmentation, and Mapper topology matching recognizes through-wall radar activities without any learned weights.

desk verdict A genuine non-neural TWR HAR pipeline with open code and honest limitations, but the headline accuracy numbers rest on an ambiguous template/validation split that could inflate them by several points. read the letter →

arxiv 2506.05169 v1 pith:HOM66XN6 submitted 2025-06-05 cs.CV eess.SP

Weicheng Gao This is my paper · ORCID
classification cs.CVeess.SP
keywords through-the-wallradarhumanactivityrecognitionmicro-DopplersignatureactivecontourmodelSIFTcornerdetectionMapperalgorithmpointcloudtopologymatchingtemplate
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that through-the-wall radar human activity recognition does not require a neural network. It assembles a complete non-learned chain: generate range-time and Doppler-time maps, locate the human micro-Doppler signature and the noise background using SIFT corner detection, segment the signature with a multiphase Chan-Vese active contour model, discretize the contour into a two-dimensional point cloud, and classify by topological similarity to pre-collected templates via the Mapper algorithm. The reported validation accuracy is 73.63% on simulated range-time maps and 65.13% on simulated Doppler-time maps, with 52.88% and 38.63% on the corresponding measured maps. The paper reads those numbers as a real but modest validity: below network-based methods, but enough to show that a physically interpretable, low-supervision signal-processing route still exists.

What carries the argument

The load-bearing mechanism is the pair of starting points for the level-set evolution: the center of gravity of SIFT keypoints, assumed to lie inside the human micro-Doppler signature, and the pixel with the largest average distance to all keypoints, assumed to lie in the noise background. Two level-set functions initialized from these points evolve under the four-phase Chan-Vese energy, splitting the image into four regions, and the region $\Omega_{+-}$ is kept as the micro-Doppler signature. The contour of that level set is then discretized into a point cloud, and Mapper converts each point cloud into a graph of overlapping grid cells; the Jaccard similarity between the input graph and each template graph, summed over template samples, yields the recognized activity.

What would settle it

Run the released implementation on a measured Doppler-time map recorded under strong wall multipath and check whether the extracted level-set region encloses the known limb micro-Doppler traces; since the paper already reports poor extraction there, a controlled experiment that injects interference before corner detection and watches range-time-map accuracy fall toward chance would show whether the center-initialization premise is the weak link.

Watch

Extended reading notes

Core claim

The central claim is that a deliberately non-neural pipeline can perform through-the-wall human activity recognition by decomposing the radar image into a foreground micro-Doppler region and then measuring the topology of that region's contour. The pipeline first uses SIFT corners to estimate a near point inside the human signature and a far point in the background, initializes two level-set functions from those points in a multiphase Chan-Vese energy minimization, and treats one of the four segmented regions as the micro-Doppler signature. That signature is converted to a point cloud, and the Mapper algorithm builds overlapping-grid graphs whose edge overlap with template graphs, scored by Jaccard similarity, selects the activity label. The author argues the method has validity on range-time maps in both simulation and measurement, while explicitly cautioning that measured Doppler-time maps are poorly segmented under system interference and are not recommended for recognition.

Load-bearing premise

The load-bearing premise is that SIFT corners place the near starting point inside the human micro-Doppler trace and the far starting point in the noise background, so the two evolving contours converge to the right region; the paper reports that this premise breaks down on measured Doppler-time maps, where system interference corrupts feature extraction.

Editorial extensions

If this is right

  • Through-the-wall activity recognition can be achieved with no training step, using only a small bank of template point clouds per activity.
  • The approach is most dependable on range-time maps, where it retains validity on measured data despite trailing neural-network baselines.
  • The pipeline degrades by no more than about 15 percentage points when 10 dB of Gaussian noise is added, a robustness comparable to most network baselines reported in the same experiments.
  • Because each stage is an explicit signal-processing operation, failures are traceable: the measured Doppler-time-map shortfall is attributed to interference corrupting segmentation before matching.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An inference beyond the paper: a clutter-suppression front end that protects the corner-detection step could raise measured Doppler-time-map accuracy substantially, since the paper identifies interference during segmentation, not the matching step, as the main failure.
  • An inference beyond the paper: the template bank could be replaced by contours synthesized from a human motion model, making the whole chain parameter-driven and removing the need for collected template data.
  • An inference beyond the paper: Mapper's grid-graph representation is a coarse shape signature, so combining it with per-scattering-center Doppler estimates might separate activities whose contours overlap.
  • An inference beyond the paper: the reported tolerance to Gaussian noise suggests structured interference, not noise, is the sharpest test; stress experiments varying wall thickness and antenna coupling would be more informative than further noise sweeps.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes a non-neural-network pipeline for through-the-wall radar human activity recognition. After generating range-time maps (RTM) and Doppler-time maps (DTM), SIFT corner detection estimates foreground and background seed points, a multiphase Chan-Vese active contour model segments the micro-Doppler signature, the contour is discretized into a point cloud, and classification is performed by summing Mapper-based Jaccard similarities to template point clouds (Eq. 49). On simulated data the method reports 73.63% accuracy on RTM and 65.13% on DTM; on measured data it reports 52.88% on RTM and 38.63% on DTM. The paper includes detailed derivations, ablations, robustness tests, and open-source code.

Significance. If the evaluation is sound, the paper shows that a fully interpretable, training-free pipeline can approach, though not match, the accuracy of deep networks on TWR HAR, and it provides a physically grounded baseline. The manuscript is transparent about its main weaknesses: Section IV-B admits poor feature extraction on measured DTMs due to system interference, and Section IV-E lists limitations of the overall logic and segmentation. Strengths include the open-source implementation and the honest reporting of failure cases. The main concern is whether the reported accuracy numbers are inflated by the unclear template/validation partition; this issue is checkable and should be resolved.

major comments (4)
  1. [Section IV-A/IV-C] The evaluation protocol does not establish that the template set and validation set are disjoint. The text says "1/5 of the data is used for performance verification" and "20 sets of data are randomly selected from each type of activity for template matching," both from the same 4000-set pool. Since classification sums Jaccard similarities to template point clouds (Eq. 49), any validation sample that is also a template will match its own class almost perfectly. Under the per-class counts in Table II, the expected overlap is roughly 48/800 if both selections are made independently from the full pool, which could materially inflate the headline measured RTM accuracy of 52.88%. Please state explicitly whether the template and validation selections were made disjointly, and if so, describe the split procedure.
  2. [Eq. (39) and Algorithm 1] The sign pattern in Eq. (39) for ∂E/∂φ2 is inconsistent with the derivation from Eq. (27)-(34) and with Algorithm 1. Eq. (39) has +λ+-(1-H(φ1))|I-c+-|² - λ-+H(φ1)|I-c-+|², whereas Algorithm 1 has -λ+-H(φ1)|I-c+-|² + λ-+(1-H(φ1))|I-c-+|². The latter follows from the stated energy functional. Please correct Eq. (39) or clarify which expression is used in the implementation.
  3. [Section IV-C, Tables IV and V] The robustness conclusion is not supported by the tables. The text states that "the validation accuracy of the proposed method decreases by no more than 15% when the SNR decreases by no more than 10 dB" and concludes "consistent with or even better than the robustness of the vast majority of existing network methods." At ΔSNR = -10 dB, the proposed method drops 13.38 points on simulated RTM and 14.88 points on simulated DTM, while CapsuleNet, GCN, AEN-BiGRU, and WSN-CRF drop 8.13, 7.50, 8.62, and 8.12 points respectively. At -12 dB the proposed method's drop is roughly twice that of the best network methods. The claim should be reworded to compare relative degradation or error-rate ratios, or the comparison should be removed.
  4. [Section IV-A and IV-C] The reported accuracies are single runs with one random template selection. Because the template set is a random subsample of 20 per activity, the results depend on that draw, and no error bars or repeated-selection statistics are provided. For a claim of "some validity" at 52.88%, at least a small number of repeated template selections with mean and standard deviation should be reported, or the single-run nature should be explicitly acknowledged in the experimental design.
minor comments (6)
  1. [Eq. (46)] The vertical-edge overlap region uses the same notation O_{(i,j),(i+1,j)} as the horizontal-edge formula; it should be O_{(i,j),(i,j+1)}.
  2. [Eq. (37)] The last variational term is labeled δF2/δφ1 but should be δF4/δφ1.
  3. [Table VI] ECFRNet outperforms the proposed SIFT corner detection (76.13 vs 73.63 on simulated RTM), so the sentence that "the final validation accuracy does not vary much" is imprecise.
  4. [Section IV-C, Fig. 8 paragraph] The phrase "The simulated validation accuracy of the proposed method on RTM and DTM is 52.88%..." should read "measured validation accuracy."
  5. [Eq. (49) and Table II] The notation "Class" and "Cla" is used inconsistently; please define the relationship between Class, Cla, and ClaNum.
  6. [Algorithm 1] The line "λ++, λ+−−, λ−+, λ−−" contains a typo; it should be "λ++, λ+−, λ−+, λ−−."

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the non-neural pipeline is an explicit composition of SIFT, Chan-Vese, and Mapper; the only flagged concern is a possible template/validation overlap that is a data-partition issue, not a circular derivation.

full rationale

Walking the derivation chain from Eq. (1) through Eq. (49), every stage is explicit rather than self-referential. The echo model and RTM/DTM formation are standard radar processing; the SIFT corner detection (Eqs. 17-20), the two initial points (Eqs. 21-25), the multiphase Chan-Vese energy (Eq. 27) with its Euler-Lagrange updates (Eqs. 38-39), and the Mapper/Jaccard matching (Eqs. 43-49) are all defined independently of the activity labels and are not fitted to the reported accuracies. The self-citations [30]-[32] supply standard SIFT hyperparameters (I_sigma=1.6, oct=3, KCor=30) and comparative network settings, but they are not load-bearing: the segmentation and matching logic is derived in this paper, and the ablation tables show the proposed components outperform external alternatives such as LBF, GAC, DRLSE, Hausdorff, and Wasserstein. The only potential by-construction element is in the evaluation protocol: Section IV-A says '1/5 of the data is used for performance verification' and '20 sets of data are randomly selected from each type of activity for template matching' without stating that the two sets are disjoint, so if coincident samples exist, Eq. (48) yields similarity 1 and Eq. (49) forces the template's class. That is a data-partition and validity concern that should be fixed by reporting an explicit disjoint split; it does not make the method's derivation circular, and the remaining non-overlapping validation samples still provide independent evidence. Overall, no significant circularity.

Assumptions & free parameters 11 free parameters · 6 assumptions · 0 invented entities

The ledger shows a supervised, heavily hand-tuned pipeline rather than a first-principles derivation. The central empirical claim rests on eleven groups of hand-set hyperparameters, six domain assumptions, and no invented physical entities. The most fragile assumptions are the SIFT-based initialization and the arbitrary choice of Omega_{+-} as the signature region.

free parameters (11)
  • k0 (EMD mode cutoff) = 3
    Hand-set starting IMF index for RTM denoising; affects SNR of subsequent maps.
  • LWind (STFT window length) = 0.5 s
    Hand-set window length for Doppler-time map generation.
  • PWind (STFT step) = 0.05 s
    Hand-set step between STFT windows; controls slow-time resolution.
  • CutThreshold = 0.3
    Fraction of maximum pixel value used to binarize the input map before SIFT detection.
  • I_sigma and oct (SIFT scales) = 1.6 and 3
    Hand-set Gaussian scale and octaves for SIFT keypoint detection.
  • KCor (target number of corners) = 30
    Number of SIFT keypoints retained for foreground and background center estimation.
  • lambda and mu (Chan-Vese weights) = 1 and 0.5
    Hand-set data-fitting and length-regularization weights in the level-set energy.
  • epsilon and tStep (level-set solver) = 1 and 0.1
    Hand-set Heaviside smoothing parameter and gradient-descent time step.
  • rho1 and rho2 (initial level-set radii) = 64 simulated, 32 measured
    Hand-set radii of the two initial circles; chosen differently for simulated and measured images.
  • Maximum iterations of Algorithm 1 = 20/20/30/50
    Stopping iterations for simulated RTM/DTM and measured RTM/DTM.
  • nx, ny and of (Mapper grid) = 100, 100, 1.5
    Hand-set grid resolution and overlap factor for the Mapper-style point-cloud matching.
assumptions (6)
  • standard math Chan-Vese multiphase energy minimization via Euler-Lagrange gradient descent converges to a useful segmentation of radar images.
    Invoked throughout Section III-A; standard image-segmentation mathematics, not proven for radar maps.
  • domain assumption The simplified six-scatterer human motion model (Eq. 9) with sinusoidal limb swings is adequate for generating simulated data.
    Used in Section II to generate simulated radar echoes; simulated results inherit the model's fidelity.
  • domain assumption The center of gravity of SIFT keypoints and the pixel with maximum average distance to keypoints fall respectively inside the human micro-Doppler signature and in the noise background.
    Section III-A Eqs. (21)-(25); if false, the level-set initialization is wrong. The paper admits poor extraction on measured DTMs.
  • ad hoc to paper Region Omega_{+-} (phi1>=0, phi2<0) is the correct quadrant to extract as the micro-Doppler signature.
    Section III-A end states Omega_{+-} is considered the signature region without justifying why not Omega_{++}, Omega_{-+}, or Omega_{--}.
  • domain assumption Grid-edge Jaccard similarity between contour point clouds is a valid activity-discriminative topological measure.
    Section III-B assumes the Mapper-style graph similarity separates activities; no theoretical or empirical guarantee beyond the reported tables.
  • domain assumption The wall is represented by fixed thickness and relative dielectric constant and the noise is additive Gaussian.
    Section IV-A uses an estimated dielectric constant of 6; measured performance depends on this scene model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Through-the-Wall Radar Human Activity Recognition WITHOUT Using Neural Networks." pith.science (2026). https://pith.science/paper/HOM66XN6

@misc{pith2026250605169,
  author       = {Pith},
  title        = {Pith review of: Through-the-Wall Radar Human Activity Recognition WITHOUT Using Neural Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HOM66XN6}},
  note         = {Machine review of arXiv:2506.05169}
}
read the original abstract

After a few years of research in the field of through-the-wall radar (TWR) human activity recognition (HAR), I found that we seem to be stuck in the mindset of training on radar image data through neural network models. The earliest related works in this field based on template matching did not require a training process, and I believe they have never died. Because these methods possess a strong physical interpretability and are closer to the basis of theoretical signal processing research. In this paper, I would like to try to return to the original path by attempting to eschew neural networks to achieve the TWR HAR task and challenge to achieve intelligent recognition as neural network models. In detail, the range-time map and Doppler-time map of TWR are first generated. Then, the initial regions of the human target foreground and noise background on the maps are determined using corner detection method, and the micro-Doppler signature is segmented using the multiphase active contour model. The micro-Doppler segmentation feature is discretized into a two-dimensional point cloud. Finally, the topological similarity between the resulting point cloud and the point clouds of the template data is calculated using Mapper algorithm to obtain the recognition results. The effectiveness of the proposed method is demonstrated by numerical simulated and measured experiments. The open-source code of this work is released at: https://github.com/JoeyBGOfficial/Through-the-Wall-Radar-Human-Activity-Recognition-Without-Using-Neural-Networks.

Figures

Figures reproduced from arXiv: 2506.05169 by the authors.

Figure 1
Figure 1. Current works in this field take neural network-based methods as the research hotspot. This work returns to rethink the value of traditional mindsets. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. TWR human echo model and data processing. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Schematic diagram of the proposed ACM-based micro-Doppler signature extraction method. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Schematic diagram of the proposed indoor HAR method based on point cloud topological structure similarity using Mapper algorithm. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Simulated visualization results of the proposed method: The first row presents RTMs after corner detection, the second row presents RTMs after [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Measured visualization results of the proposed method: The first row presents RTMs after corner detection, the second row presents RTMs after [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Simulated training and validation accuracy under different methods. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Measured training and validation accuracy under different methods. [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: A sketch of my grandma in her youth. I would like to thank my family: My grandfather, mother, fa￾ther, aunt, for the impeccable care you gave to my grandmother on her deathbed and for making me feel the most precious affection on earth. Thank you to all the elders and …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 58 canonical work pages

  1. [1]

    Through-the-wall human motion indication using sparsity-driven change detection,

    F. Ahmad and M. G. Amin, “Through-the-wall human motion indication using sparsity-driven change detection,”IEEE Trans. Geosci. Remote Sens., vol. 51, no. 2, pp. 881-890, Feb. 2013

    work page 2013

  2. [2]

    WiFi- based through-the-wall presence detection of stationary and moving humans analyzing the doppler spectrum,

    S. Di Domenico, M. De Sanctis, E. Cianca, and M. Ruggieri, “WiFi- based through-the-wall presence detection of stationary and moving humans analyzing the doppler spectrum,”IEEE Aerosp. Electron. Syst. Mag., vol. 33, no. 5–6, pp. 14–19, May 2018

    work page 2018

  3. [3]

    Design of dual polarized FMCW radar system with high gain lens antenna for through-the-wall human detection,

    C. Lee, K. H. Park, H. Kim, J. Lee, H. Kim, K.-H. Shin, and J.-W. Kim, “Design of dual polarized FMCW radar system with high gain lens antenna for through-the-wall human detection,”IEEE Access, p. 1, Jan. 2024

    work page 2024

  4. [4]

    A study on 3D human pose estimation using Through-Wall IR-UWB radar and transformer,

    G. W. Kim, S. W. Lee, H. Y . Son, and K. W. Choi, “A study on 3D human pose estimation using Through-Wall IR-UWB radar and transformer,” IEEE Access, vol. 11, pp. 15082–15095, Jan. 2023

    work page 2023

  5. [5]

    Joint compressed sensing and spread spectrum through-the-wall radar imaging,

    M. Li, X. Xi, X. Zhang, and G. Liu, “Joint compressed sensing and spread spectrum through-the-wall radar imaging,”IEEE Access, vol. 9, pp. 6259–6267, Dec. 2020

    work page 2020

  6. [6]

    Target localization and wall parameters estimation via distributed through-wall imaging radar,

    Y . Yao, J. Chen, S. Guo, X. Hao, N. Li, G. Cui, and L. Kong, “Target localization and wall parameters estimation via distributed through-wall imaging radar,”IEEE Trans. Instrum. Meas., vol. 74, pp. 1–11, Nov. 2024

    work page 2024

  7. [7]

    An effective image reconstruction enhancement method with convolutional reweight- ing for near-field sar,

    S. Song, Y . Dai, T. Jin, X. Wang, Y . Hua, and X. Zhou, “An effective image reconstruction enhancement method with convolutional reweight- ing for near-field sar,”IEEE Antennas Wirel. Propag. Lett., vol. 23, no. 8, pp. 2486–2490, May 2024

    work page 2024

  8. [8]

    Range coherence factor for down range sidelobes suppression in radar imaging through multilayered dielectric media,

    Q. An, A. Hoorfar, W. Zhang, S. Li, and J. Wang, “Range coherence factor for down range sidelobes suppression in radar imaging through multilayered dielectric media,”IEEE Access, vol. 7, pp. 66910–66918, Jan. 2019

    work page 2019

Show all 63 references
  1. [9]

    CNN-based time–frequency image enhancement algorithm for target tracking using Doppler through- wall radar,

    M. Ding, Y . Ding, Y . Peng, and J. Cao, “CNN-based time–frequency image enhancement algorithm for target tracking using Doppler through- wall radar,”IEEE Geosci. Remote Sens. Lett., vol. 20, pp. 1–5, Jan. 2023

    2023

  2. [10]

    Motion compensation method based on MFDF of moving target for UWB MIMO through-wall radar system,

    J. Pan, Z.-K. Ni, C. Shi, Z. Zheng, S. Ye, and G. Fang, “Motion compensation method based on MFDF of moving target for UWB MIMO through-wall radar system,”IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, Oct. 2021

    2021

  3. [11]

    First demonstration of using signal processing approach to suppress signal ringing in impulse UWB through-wall radar,

    Y . Jin, J. Chen, B. Liang, D. Yang, M. Xing, and L. Liu, “First demonstration of using signal processing approach to suppress signal ringing in impulse UWB through-wall radar,”IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, Apr. 2021

    2021

  4. [12]

    RESNET-Based counting algorithm for moving targets in through- the-wall radar,

    Y . Jia, Y . Guo, R. Song, G. Wang, S. Chen, X. Zhong, and G. Cui, “RESNET-Based counting algorithm for moving targets in through- the-wall radar,”IEEE Geosci. Remote Sens. Lett., vol. 18, no. 6, pp. 1034–1038, May 2020

    2020

  5. [13]

    Gen- eralization of channel micro-Doppler capacity evaluation for improved finer-grained human activity classification using MIMO UWB radar,

    F. Qi, Z. Li, Y . Ma, F. Liang, H. Lv, J. Wang, and A. E. Fathy, “Gen- eralization of channel micro-Doppler capacity evaluation for improved finer-grained human activity classification using MIMO UWB radar,” IEEE Trans. Microw. Theory Tech., vol. 69, no. 11, pp. 4748–4761, May 2021

    2021

  6. [14]

    Through-the-wall human activity recognition using radar technologies: A review,

    J. Yousaf, S. Yakoub, S. Karkanawi, T. Hassan, E. Almajali, H. Zia, and M. Ghazal, “Through-the-wall human activity recognition using radar technologies: A review,”IEEE Open J. Antennas Propag., p. 1, Jan. 2024

    2024

  7. [15]

    Sense-through-wall human detection based on UWB radar sensors,

    S. D. Liang, “Sense-through-wall human detection based on UWB radar sensors,”Signal Process., vol. 126, pp. 117–124, Oct. 2015

    2015

  8. [16]

    A real- time human activity recognition method for through-the-wall radar,

    C. Cheng, F. Ling, S. Guo, G. Cui, Q. Jian, C. Jia, and Q. Ran, “A real- time human activity recognition method for through-the-wall radar,” in Proc. IEEE Radar Conf., pp. 1–5, Sep. 2020

    2020

  9. [17]

    Detailed feature representation and analysis of low frequency UWB radar range profile for improving through-wall human activity recognition,

    Y . Peng and S. Guo, “Detailed feature representation and analysis of low frequency UWB radar range profile for improving through-wall human activity recognition,” inProc. IEEE Radar Conf., pp. 1–6, Sep. 2020

    2020

  10. [18]

    Through-the-wall human activity recognition using ultra- wideband radar and deep learning,

    S. Yakoub, S. Karkanawi, G. E. Hassan, T. Hassan, M. Ghazal, and J. Yousaf, “Through-the-wall human activity recognition using ultra- wideband radar and deep learning,” inProc. IEEE Int. Conf. Future Internet Things Cloud, pp. 223–228, Aug. 2024

    2024

  11. [19]

    Spectro-Temporal modeling for human activity recognition using a radar sensor network,

    F. Luo, E. Bodanese, S. Khan, and K. Wu, “Spectro-Temporal modeling for human activity recognition using a radar sensor network,”IEEE Trans. Geosci. Remote Sens., vol. 61, pp. 1–13, Jan. 2023

    2023

  12. [20]

    Human motion serialization recognition with through-the-wall radar,

    X. Yang, P. Chen, M. Wang, S. Guo, C. Jia, and G. Cui, “Human motion serialization recognition with through-the-wall radar,”IEEE Access, vol. 8, pp. 186879–186889, Jan. 2020

    2020

  13. [21]

    Research on human activity recognition technology under the condition of through-the-wall,

    R. Cao, X. Yang, Z. Yang, M. Zhou, and L. Xie, “Research on human activity recognition technology under the condition of through-the-wall,” inProc. IEEE Int. Conf. Commun. China, pp. 501–506, Aug. 2020

    2020

  14. [22]

    MHHT-Based method for analysis of micro-Doppler signatures for human finer-grained activity using through-wall SFCW radar,

    F. Qi, H. Lv, F. Liang, Z. Li, X. Yu, and J. Wang, “MHHT-Based method for analysis of micro-Doppler signatures for human finer-grained activity using through-wall SFCW radar,”Remote Sens., vol. 9, no. 3, p. 260, Mar. 2017. ARXIV PREPRINT, JUNE, 2025 15

    2017

  15. [23]

    GCN-Enhanced mul- tidomain fusion network for through-wall human activity recognition,

    X. Wang, S. Guo, J. Chen, P. Chen, and G. Cui, “GCN-Enhanced mul- tidomain fusion network for through-wall human activity recognition,” IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, Jan. 2022

    2022

  16. [24]

    Dataset of human motion status using IR-UWB through-wall radar,

    Z. Zhengliang, Y . Degui, Z. Junchao, and T. Feng, “Dataset of human motion status using IR-UWB through-wall radar,”J. Syst. Eng. Electron., vol. 32, no. 5, pp. 1083–1096, Oct. 2021

    2021

  17. [25]

    Cross-Modal supervised human body pose recognition techniques for through-wall radar,

    D. Xu, Y . Liu, Q. Wang, L. Wang, and Q. Shen, “Cross-Modal supervised human body pose recognition techniques for through-wall radar,”Sensors, vol. 24, no. 7, p. 2207, Mar. 2024

    2024

  18. [26]

    Through wall human detection based on support tensor machines,

    L. Zhang, W. Wang, Y . Jiang, D. Wang, and M. Zhang, “Through wall human detection based on support tensor machines,” inProc. Lect. Notes Electr. Eng., 2019, pp. 746–751

    2019

  19. [27]

    TWR-MCAE: a data augmen- tation method for through-the-wall radar human motion recognition,

    W. Gao, X. Yang, X. Qu, and T. Lan, “TWR-MCAE: a data augmen- tation method for through-the-wall radar human motion recognition,” IEEE Trans. Geosci. Remote Sens., vol. 60, pp. 1–17, Jan. 2022

    2022

  20. [28]

    Indoor human behavior recognition method based on wavelet scattering network and conditional random field model,

    X. Qu, W. Gao, H. Meng, Y . Zhao, and X. Yang, “Indoor human behavior recognition method based on wavelet scattering network and conditional random field model,”IEEE Trans. Geosci. Remote Sens., vol. 61, pp. 1–15, Jan. 2023

    2023

  21. [29]

    A lightweight multiscale neural network for indoor human activity recog- nition based on macro and micro-Doppler features,

    X. Yang, W. Gao, X. Qu, P. Yin, H. Meng, and A. E. Fathy, “A lightweight multiscale neural network for indoor human activity recog- nition based on macro and micro-Doppler features,”IEEE Intern. Things J., vol. 10, no. 24, pp. 21836–21854, Aug. 2023

    2023

  22. [30]

    Through-the-Wall radar human activity micro-doppler signature representation method based on joint Boulic-sinusoidal pendulum model,

    X. Yang, W. Gao, X. Qu, Z. Ma, and H. Zhang, “Through-the-Wall radar human activity micro-doppler signature representation method based on joint Boulic-sinusoidal pendulum model,”IEEE Trans. Microw. Theory Tech., vol. 73, no. 2, pp. 1248–1263, Aug. 2024

    2024

  23. [31]

    Adaptive micro- Doppler corner feature extraction method based on difference of gaus- sian filter and deformable convolution,

    W. Gao, X. Qu, H. Meng, X. Sun, and X. Yang, “Adaptive micro- Doppler corner feature extraction method based on difference of gaus- sian filter and deformable convolution,”IEEE Signal Process. Lett., vol. 31, pp. 860–864, Jan. 2024

    2024

  24. [32]

    Generalizable indoor human activity recognition method based on micro-Doppler corner point cloud and dynamic graph learning,

    X. Yang, W. Gao, X. Qu, and H. Meng, “Generalizable indoor human activity recognition method based on micro-Doppler corner point cloud and dynamic graph learning,”IEEE Trans. Aerosp. Electron. Syst., pp. 1–15, Jan. 2024

    2024

  25. [33]

    A survey of deep learning-based human activity recognition in radar,

    X. Li, Y . He, and X. Jing, “A survey of deep learning-based human activity recognition in radar,”Remote Sens., vol. 11, no. 9, p. 1068, May 2019

    2019

  26. [34]

    Radar micro-Doppler signatures of various human activities,

    R. M. Narayanan and M. Zenaldin, “Radar micro-Doppler signatures of various human activities,”IET Radar Sonar Navig., vol. 9, no. 9, pp. 1205–1215, Sep. 2015

    2015

  27. [35]

    Analysis of micro- Doppler signatures,

    V . C. Chen, F. Li, S. -s. Ho, and H. Wechsler, “Analysis of micro- Doppler signatures,” inIEE Proc.-Radar Sonar Navig., vol. 150, no. 4, p. 271, Jan. 2003

    2003

  28. [36]

    Hilbert-Huang transform (HHT) processing of through-wall noise radar data for human activity characterization,

    C. P. Lai, Q. Ruan, and R. M. Narayanan, “Hilbert-Huang transform (HHT) processing of through-wall noise radar data for human activity characterization,” inProc. IEEE Work. Signal Process. Appl. Public Secur. Forensics, pp. 1–6, Apr. 2007

    2007

  29. [37]

    Through-the-Wall target separation using Low-Rank and variational mode decomposition,

    F. H. C. Tivive and A. Bouzerdoum, “Through-the-Wall target separation using Low-Rank and variational mode decomposition,”IEEE Trans. Geosci. Remote Sens., vol. 57, no. 12, pp. 9928–9940, Aug. 2019

    2019

  30. [38]

    Front-Wall clutter removal in through-the-wall radar based on weighted nuclear norm minimization,

    Y . Zhou, C. Huang, H. Liu, D. Li, and T.-K. Truong, “Front-Wall clutter removal in through-the-wall radar based on weighted nuclear norm minimization,”IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, Nov. 2020

    2020

  31. [39]

    Lightweight multi- attention enhanced fusion network for omnidirectional human activity recognition with FMCW radar,

    X. Li, Y . Qiu, Z. Deng, X. Liu, and X. Huang, “Lightweight multi- attention enhanced fusion network for omnidirectional human activity recognition with FMCW radar,”IEEE Intern. Things J., vol. 12, no. 5, pp. 5755–5768, Nov. 2024

    2024

  32. [40]

    An EMD-Based micro-Doppler signature for moving vehicles classification,

    S. Zhai, T. Mao, and B. Hu, “An EMD-Based micro-Doppler signature for moving vehicles classification,” inProc. IEEE Int. Conf. Commun. China, pp. 1322–1326, Dec. 2019

    2019

  33. [41]

    Unsupervised SAR image change detection based on SIFT keypoints and region information,

    Y . Wang, L. Du, and H. Dai, “Unsupervised SAR image change detection based on SIFT keypoints and region information,”IEEE Geosci. Remote Sens. Lett., vol. 13, no. 7, pp. 931–935, May 2016

    2016

  34. [42]

    Ship contour extraction from SAR images based on faster R-CNN and Chan–Vese model,

    M. Jiang, L. Gu, X. Li, F. Gao, and T. Jiang, “Ship contour extraction from SAR images based on faster R-CNN and Chan–Vese model,”IEEE Trans. Geosci. Remote Sens., vol. 61, pp. 1–14, Jan. 2023

    2023

  35. [43]

    A multiphase level set framework for image segmentation using the Mumford and Shah model,

    L. A. Vese and T. F. Chan, “A multiphase level set framework for image segmentation using the Mumford and Shah model,”Intern. J. Comput. Vis., vol. 50, no. 3, pp. 271–293, Jan. 2002

    2002

  36. [44]

    Meaningful object segmen- tation from SAR images via a multiscale nonlocal active contour model,

    G.-S. Xia, G. Liu, W. Yang, and L. Zhang, “Meaningful object segmen- tation from SAR images via a multiscale nonlocal active contour model,” IEEE Trans. Geosci. Remote Sens., vol. 54, no. 3, pp. 1860–1873, Oct. 2015

    2015

  37. [45]

    A coastline detection method in polari- metric SAR images mixing the Region-Based and Edge-Based active contour models,

    C. Liu, Y . Xiao, and J. Yang, “A coastline detection method in polari- metric SAR images mixing the Region-Based and Edge-Based active contour models,”IEEE Trans. Geosci. Remote Sens., vol. 55, no. 7, pp. 3735–3747, Mar. 2017

    2017

  38. [46]

    A topological data analysis guided fusion algorithm: Mapper-Regularized manifold alignment,

    J. Hu, D. Hong, Y . Wang, and X. X. Zhu, “A topological data analysis guided fusion algorithm: Mapper-Regularized manifold alignment,” in Proc. IEEE Intern. Geosci. Remote Sens. Symp., pp. 2822–2825, Jul. 2019

    2019

  39. [47]

    Pose measurement and motion estimation of space non-cooperative targets based on laser radar and stereo-vision fusion,

    J. Peng, W. Xu, B. Liang, and A.-G. Wu, “Pose measurement and motion estimation of space non-cooperative targets based on laser radar and stereo-vision fusion,”IEEE Sens. J., vol. 19, no. 8, pp. 3008–3019, Dec. 2018

    2018

  40. [48]

    Detection and classfication of subsurface objects by polarimetric radar imaging,

    C. N. Koyama and M. Sato, “Detection and classfication of subsurface objects by polarimetric radar imaging,” inProc. IEEE Radar Conf., Oct. 2015

    2015

  41. [49]

    Simhumalator: An open-source end-to-end radar simulator for human activity recognition,

    S. Vishwakarma, W. Li, C. Tang, K. Woodbridge, R. Adve, and K. Chetty, “Simhumalator: An open-source end-to-end radar simulator for human activity recognition,”IEEE Aerosp. Electron. Syst. Mag., vol. 37, no. 3, pp. 6–22, Dec. 2021

    2021

  42. [50]

    A hybrid CNN–LSTM network for the classification of human activities based on micro-Doppler radar,

    J. Zhu, H. Chen, and W. Ye, “A hybrid CNN–LSTM network for the classification of human activities based on micro-Doppler radar,”IEEE Access, vol. 8, pp. 24713–24720, Jan. 2020

    2020

  43. [51]

    A survey on Vision Transformer,

    K. Han, Y . Wang, H. Chen, X. Chen, J. Guo, Z. Liu, Y . Tang, A. Xiao, C. Xu, Y . Xu, Z. Yang, Y . Zhang, and D. Tao, “A survey on Vision Transformer,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 45, no. 1, pp. 87–110, Jan. 2023

    2023

  44. [52]

    A ConvNet for the 2020s,

    Z. Liu, H. Mao, C.-Y . Wu, C. Feichtenhofer, T. Darrell, and S. Xie, “A ConvNet for the 2020s,” inProc. IEEE/CVF Conf. Comput. Vis. Pattern Recog., Jun. 2022

    2022

  45. [53]

    Through-Wall human motion recognition based on transfer learning and ensemble learning,

    P. Chen, S. Guo, H. Li, X. Wang, G. Cui, C. Jiang, and L. Kong, “Through-Wall human motion recognition based on transfer learning and ensemble learning,”IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, Apr. 2021

    2021

  46. [54]

    Capsule network with multiscale feature fusion for hidden human activity classification,

    X. Wang, Y . Wang, S. Guo, L. Kong, and G. Cui, “Capsule network with multiscale feature fusion for hidden human activity classification,” IEEE Trans. Instrum. Meas., vol. 72, pp. 1–12, Jan. 2023

    2023

  47. [55]

    RPCA-based high resolution through-the-wall human motion feature extraction and classification,

    Q. An, S. Wang, L. Yao, W. Zhang, H. Lv, J. Wang, S. Li, and A. Hoorfar, “RPCA-based high resolution through-the-wall human motion feature extraction and classification,”IEEE Sens. J., vol. 21, no. 17, pp. 19058–19068, Jun. 2021

    2021

  48. [56]

    Comparisons of feature extraction algorithm based on unmanned aerial vehicle image,

    W. Xi, Z. Shi, and D. Li, “Comparisons of feature extraction algorithm based on unmanned aerial vehicle image,”Open Phys., vol. 15, no. 1, pp. 472–478, Jan. 2017

    2017

  49. [57]

    Faster and better: A machine learning approach to corner detection,

    E. Rosten, R. Porter, and T. Drummond, “Faster and better: A machine learning approach to corner detection,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 32, no. 1, pp. 105–119, Jan. 2010

    2010

  50. [58]

    ECFRNet: Effective corner feature representations network for image corner detection,

    J. Jing, C. Liu, W. Zhang, Y . Gao, and C. Sun, “ECFRNet: Effective corner feature representations network for image corner detection,” Expert Syst. Appl., vol. 211, Jan. 2023, Art. no. 118673

    2023

  51. [59]

    Implicit active con- tours driven by local binary fitting energy,

    C. Li, C. -Y . Kao, J. C. Gore and Z. Ding, “Implicit active con- tours driven by local binary fitting energy,” inIEEE Conf. Comput. Vis. Pattern Recognit., Minneapolis, MN, USA, 2007, pp. 1-7, doi: 10.1109/CVPR.2007.383014

    2007

  52. [60]

    Geodesic active contours,

    V . Caselles, R. Kimmel, and G. Sapiro, “Geodesic active contours,” Intern. J. Comput. Vis., vol. 22, no. 1, pp. 61–79, Jan. 1997

    1997

  53. [61]

    Distance regularized level set evolution and its application to image segmentation,

    N. C. Li, N. C. Xu, N. C. Gui, and M. D. Fox, “Distance regularized level set evolution and its application to image segmentation,”IEEE Trans. Image Process., vol. 19, no. 12, pp. 3243–3254, Aug. 2010

    2010

  54. [62]

    Efficient and accurate Hausdorff distance computation based on diffusion search,

    D. Zhang, L. Zou, Y . Chen, and F. He, “Efficient and accurate Hausdorff distance computation based on diffusion search,”IEEE Access, vol. 6, pp. 1350–1361, Dec. 2017

    2017

  55. [63]

    Earth Mover’s Distance vs. quadratic form distance: An analytical and empirical comparison,

    C. Beecks, M. S. Uysal, and T. Seidl, “Earth Mover’s Distance vs. quadratic form distance: An analytical and empirical comparison,” in IEEE Intern. Symp. Multimedia, pp. 233–236, Dec. 2015

    2015

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.