Pith. sign in

REVIEW 4 major objections 9 minor 67 references

Impact Detection in Fall Events: Leveraging Spatio-Temporal Graph Convolutional Networks and Recurrent Neural Networks Using 3D Skeletons Data

T0 review · 4 major / 9 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read This paper claims that the ground-contact frame inside a fall can be recovered from a 3D skeleton graph, and that its STGCN-GRU-BiLSTM detector reaches 97.50% accuracy.

desk verdict Useful dataset and honest task framing, but the subject-overlapping test split invalidates the headline accuracy claim. read the letter →

arxiv 2607.25710 v1 pith:G7HRNLRW submitted 2026-07-28 cs.CV

classification cs.CV
keywords impactdetectionfall3DskeletongraphconvolutionalnetworkGRUBiLSTMUP-Falldatasetcross-datasetgeneralization
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 is trying to establish that the precise impact moment in a fall—the frame where the body actually contacts the ground—can be detected from 3D joint skeleton data by treating the skeleton as a graph and reading the sequence in both time directions. Most fall detectors classify whole events and trigger during the descent or on near-falls; the proposed architecture, a Spatio-Temporal Graph Convolutional Network combined with Gated Recurrent Unit and Bidirectional Long Short-Term Memory layers (STGCN-GRU-BiLSTM), labels individual frames instead. Trained on the authors' improved 3D-skeleton version of the UP-Fall dataset, it reaches 97.50% accuracy overall, 96.50% for backward falls, and 97.50% for falls while sitting, with 94–97% accuracy in cross-dataset tests on UMAFall. If these results hold, camera-based monitoring could report the actual moment of ground contact rather than the occurrence of a fall-like motion, which is what caregivers need to filter false alarms and dispatch help immediately.

What carries the argument

The load-bearing object is the human body-joint graph $G=(A,V,E)$, built from 33 three-dimensional joints extracted by MediaPipe BlazePose and processed through a two-hop adjacency matrix. Spatial features are gathered by graph convolution over this graph, then temporal features are extracted by three temporal convolutions, and the sequence is passed through GRU layers followed by a bidirectional LSTM so each frame is classified with both past and future context. An attention vector multiplies the adjacency matrix to up-weight joints that are likely initial contact points; the paper identifies head, right shoulder, left elbow, right wrist, left hip, and right knee as the joints with consistently high attention. The same mechanism also carries the semi-automatic labeling: a Signal Magnitude Vector from synchronized accelerometer data is thresholded at $\beta=2g$ to mark impact frames, and each candidate is visually checked.

What would settle it

Re-collect a small set of falls with synchronized force-plate or floor-vibration contact timestamps, apply the paper's exact preprocessing and labeling rules, and compare the SMV-generated labels with the physical contact frame. If the threshold labels disagree with physical contact in a substantial fraction of falls, the reported 97.5% would measure label consistency rather than real impact detection; alternatively, retrain the same architecture on independently annotated labels and check whether the accuracy survives.

Watch

Extended reading notes

Core claim

The central claim is that re-framing fall analysis as frame-level impact detection, and modeling the body as a graph with bidirectional temporal context, yields a detector that can identify the exact frame of ground contact from skeleton sequences. The authors report that their STGCN-GRU-BiLSTM model outperforms the STGCN baseline by 5.34 accuracy points and reaches 97.50% accuracy, 98.11% precision, and 97.26% F1-score on the improved UP-Fall dataset. On individual fall scenarios it achieves 96.50% for backward falls while standing and 97.50% for falls while attempting to sit; in cross-dataset evaluation it holds 94–97% accuracy on UMAFall scenarios. The paper also claims that the preprocessing pipeline—cropping, background removal, MediaPipe BlazePose skeleton extraction, and semi-automatic accelerometer-based labeling—improves every compared algorithm, with the STGCN baseline rising from 72% to 87.43% accuracy. The same architectural components are shown through an ablation to contribute progressively: STGCN alone 92.16%, with ConvLSTM 93.00%, with GRU 93.50%, and the full model 97.50%.

Load-bearing premise

The load-bearing assumption is that an impact frame is correctly defined by a single combined accelerometer reading (Signal Magnitude Vector) exceeding $\beta=2g$, with the authors' visual check as confirmation; if that threshold does not match true physical ground contact, the reported accuracies measure agreement with a self-created label, not actual impact detection.

Editorial extensions

If this is right

  • If the accuracy figures hold, a camera-only system can issue a fall alert at the ground-contact frame rather than during the descent, which is the timing needed to suppress alerts caused by near-falls and stumbles.
  • The improved UP-Fall dataset with frame-level impact labels becomes a reusable benchmark for comparing impact-localization methods, not just fall-event classifiers.
  • Backward falls and seated falls are the scenarios with the largest reported gains, implying the model's joint-attention and bidirectional context capture the movement patterns specific to those fall types; a deployed detector should therefore be more reliable in two situations where elderly falls are common.
  • The cross-dataset results on UMAFall suggest the learned impact signature transfers across camera and sensor setups when the same preprocessing is applied, which is a necessary condition for practical deployment beyond a single lab.
  • The BiLSTM feature analysis indicates that falls of different speeds form separable clusters, so impact detection could be extended to report fall direction or velocity from the same features.

Reading between the lines

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

  • Because the ground-truth labels are generated from an accelerometer threshold and then visually verified, the true quantity being learned is probably 'high-acceleration frame that coincides with visible ground contact'; in deployment, falls with low impact force (slow collapses, soft surfaces) may fall below the 2g threshold and be systematically missed even if the skeleton signal contains the con
  • The occlusion results suggest a practical design rule the paper does not state: confidence should be gated on which body parts are visible, since upper-body-visible occlusion keeps accuracy at 95.2% while lower-body-only visibility drops it to 76.6%.
  • The attention ranking of head, shoulder, elbow, wrist, hip, and knee yields a testable prediction for injury prevention—real-world fall injuries should cluster at these joints—which could be checked against clinical records independently of model accuracy.
  • Because the falls are simulated by healthy young adults, an obvious next test is whether the same skeleton features and the 2g label threshold survive falls by older adults with slower descent and lower impact; the paper lists real-elderly validation as future work.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 9 minor

Summary. The paper proposes a frame-level impact-detection architecture, STGCN-GRU-BiLSTM, that takes MediaPipe BlazePose 3D skeletons from RGB video and classifies each frame as impact or non-impact. The authors introduce an 'improved' version of the UP-Fall dataset with labels obtained by a semi-automatic accelerometer SMV threshold (β = 2g) plus visual inspection, and they report 97.50% accuracy on this dataset, per-fall-type accuracies between 86% and 97.50%, and 94–97% accuracy in cross-dataset experiments on UMAFall. The paper also includes an ablation study, preprocessing comparisons, an occlusion robustness experiment, a joint-attention analysis, and a computational-cost table.

Significance. If the quantitative claims were valid, this would be a practically useful contribution: precise impact-moment localization from skeleton data could reduce false alarms in elderly fall monitoring, and the released improved dataset could support further research. The paper deserves credit for making the dataset public and for comparing several baselines under a common protocol, and the occlusion and computational analyses are informative. However, the evaluation protocol has load-bearing flaws that currently invalidate the headline accuracy numbers, and the cross-dataset setup is not described at a level that would allow reproduction or interpretation.

major comments (4)
  1. [Section 4.2, Training and Testing Strategy] The test split is not subject-disjoint. The paper states that 'we selected two fall sequences from each subject for inclusion in the test set' after an 80/10/10 split, which means sequences from the same subjects also appear in the training set. For skeleton-based models, subject identity (body proportions, habitual joint angles, recording setup) is a strong confound, so the reported 97.50% accuracy in Table 7 and the per-fall-type numbers in Table 8 cannot be interpreted as generalization to unseen individuals. A subject-disjoint split with per-subject results and error bars is required before the central claim can be accepted.
  2. [Section 3.5, Eq. (2)] The ground-truth impact labels are generated by an accelerometer SMV exceeding a fixed threshold β = 2g, followed by visual inspection by the authors. No independent validation of this threshold against true physical ground contact is provided, and the paper itself acknowledges that 'discrepancies between the semi-automatic labeling and the actual impact event can be identified.' The reported accuracies therefore measure agreement with a self-defined label, not necessarily detection of actual ground contact. The authors should validate the labeling procedure (e.g., against manual annotation by multiple raters or an independent sensor modality), report inter-annotator agreement, or explicitly restrict the claims to the defined label.
  3. [Sections 4.6 and 4.7, Cross-dataset experiments on UMAFall] The UMAFall cross-dataset evaluation is not sufficiently described. UMAFall is an inertial-sensor dataset, and the paper does not explain how camera images or video were obtained and synchronized with the sensor data, how MediaPipe skeletons were extracted from UMAFall data, or how ground-truth impact labels were defined for the UMAFall test frames. Without this information, the 94–97% accuracies in Tables 9 and 11 cannot be reproduced, and the claim of cross-dataset generalization is not verifiable.
  4. [Table 8, Individual fall scenario performance] Several rows in Table 8 contain identical metric values across different models, which is implausible. For example, in the FKnee+Standing block, LSTM and BiLSTM share identical Precision, F1 Score, Specificity, Recall, AUC ROC, and MCC, while in the FH+Standing block, CNN and STGCN share identical Accuracy, Precision, Specificity, and MCC. These repetitions suggest reporting errors and undermine the comparative claim that the proposed method outperforms the state of the art in each fall scenario. The authors should verify and correct the table, and ideally provide error bars over multiple runs.
minor comments (9)
  1. [Section 3.4 and 3.5] The terms 'Manual Labelization' and 'Semi-Automatic Labelization' should be replaced with 'Manual Labeling' and 'Semi-Automatic Labeling'.
  2. [Section 3.7, Eq. (4)] In Eq. (4), the symbol V is used in the term Γ_µ ⊗ V, but V was previously defined as the set of graph vertices; the intended operand appears to be the input feature tensor, not the vertex set.
  3. [Section 4.2] The description of the 80/10/10 split should state explicitly whether the split is performed at the subject level or the sequence level; the current text is self-contradictory because selecting two sequences per subject for the test set conflicts with a random split.
  4. [Table 8 caption] The abbreviation 'FSitting' is used in the table but is not defined in the caption; the corresponding activity is listed as 'FE' in Table 4.
  5. [Section 4.10] The section reports testing with fixed-length inputs of 60, 90, and 180 frames, but Table 6 lists a sequence length of 100 frames; the paper should explain how variable-length inputs are handled by the BiLSTM layer.
  6. [General] No code, random seeds, or number of repeated runs are reported; the authors should provide the implementation and report mean and standard deviation over multiple seeds.
  7. [Figure 2] Figure 2 is very dense and difficult to read at the current resolution; it should be enlarged or simplified for clarity.
  8. [Abstract] The abstract states 'accuracy exceeding 90%' while the body reports 97.50%; consider reporting the headline accuracy directly in the abstract.
  9. [Section 3.1.3 and 3.4] There are minor typos: 'sleleton' appears in Section 3.1.3 and 'However, While manual labeling' appears in Section 3.4.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: the model predicts accelerometer-threshold labels from independent 3D-skeleton input; the only self-referential element is a minor novelty self-citation.

full rationale

The paper's derivation chain is: (1) MediaPipe BlazePose extracts 3D skeletons from RGB images; (2) frame-level impact labels are produced independently from synchronized accelerometer data using the SMV threshold of Eq. (2) plus visual inspection (Section 3.5); (3) the STGCN-GRU-BiLSTM model maps skeleton sequences to those labels. No equation or protocol makes the model's prediction a function of the label-generating accelerometer signal, and no fitted parameter is renamed as a prediction: beta = 2g is a fixed labeling threshold, not an optimized parameter, and the skeleton input is not derived from SMV. The architecture uses standard external components (STGCN from [57,58], GRU, BiLSTM) and is not derived from the labels. The self-citation [38] in Section 2.2 only supports a novelty claim about prior work on impact detection; it is not used as evidence for the reported 97.50% accuracy. The UMAFall cross-dataset experiments provide an external benchmark. The subject-overlap in the UP-Fall split (Section 4.2) and the fact that impact is defined by the authors' threshold are validity/generalization concerns, not circularity reductions under the required standard. Therefore no significant circularity is present.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central performance claim rests on a self-defined label threshold, a pose estimator chosen by convenience, a compatibility assumption for UMAFall, and an unverifiable implementation of standard neural blocks. The released dataset is real and reusable, but it does not supply independent ground truth for impact.

free parameters (4)
  • SMV impact threshold beta = 2g
    Used in Eq. (2) to define impact versus non-impact for all labels; model accuracy is measured against this definition rather than an independent physical ground truth.
  • Sequence window size = 100 frames (approximately 5.6 s)
    Section 4.2 chooses 100 frames as the guaranteed usable length; all sequences are standardized to this window, which affects which frames the model sees.
  • k-hops in STGCN adjacency = 2
    Section 4.2 says a two-hop adjacency matrix was selected after experiments; this changes the graph receptive field and is not derived from first principles.
  • Training hyperparameter set = epochs=300, batch=8, dropout=0.25, learning rate=0.0001, hidden sizes 80/40/40/80, GRU units 33 each
    Listed in Table 6 and chosen without sensitivity analysis; the reported results are conditional on these choices.
assumptions (4)
  • domain assumption An SMV greater than 2g indicates that a fall impact occurred.
    This is the operational definition of impact in Eq. (2); the paper does not compare it against independent impact labels or clinical ground truth.
  • domain assumption MediaPipe BlazePose returns accurate 3D joint coordinates on the cropped and background-removed UP-Fall images.
    Section 3.2 motivates BlazePose with frame detection counts rather than joint accuracy; the entire input representation depends on this.
  • ad hoc to paper UMAFall data can be processed with the same image-based skeleton pipeline.
    Section 4.7 applies the preprocessing to UMAFall, but the paper never states UMAFall's sensor modalities; if UMAFall lacks RGB video, the experiment cannot be executed as described.
  • standard math The standard GCN, GRU, and BiLSTM update equations (Eqs. 5, 10, 11) are implemented exactly as stated.
    No implementation is provided, so the paper implicitly assumes the equations, once implemented, produce the reported numbers.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Impact Detection in Fall Events: Leveraging Spatio-Temporal Graph Convolutional Networks and Recurrent Neural Networks Using 3D Skeletons Data." pith.science (2026). https://pith.science/paper/G7HRNLRW

@misc{pith2026260725710,
  author       = {Pith},
  title        = {Pith review of: Impact Detection in Fall Events: Leveraging Spatio-Temporal Graph Convolutional Networks and Recurrent Neural Networks Using 3D Skeletons Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G7HRNLRW}},
  note         = {Machine review of arXiv:2607.25710}
}
read the original abstract

Fall represents a significant risk of accidental death among individuals aged over 65, presenting a global health concern. A fall is defined as any event where a person loses balance and moves to an off-position, which may or may not result in an impact where the person hits the ground. While fall detection systems have achieved good results in general, impact detection within falls remains challenging. This study proposes an efficient methodology for accurately detecting impacts within fall events by incorporating 3D joints skeleton data treated as a graph using Spatio-Temporal Graph Convolutional Networks (STGCN), Gated Recurrent Unit (GRU), and Bidirectional Long Short-Term Memory (BiLSTM) layers. By pinpointing impact moments, our approach enhances precision by distinguishing between false falls and actual impacts, contributing to better healthcare resource allocation. Our methodology, evaluated using the improved 3D skeletons UP-Fall dataset, achieves accuracy exceeding 90\% across various fall scenarios. We have made this improved dataset publicly available at https://zenodo.org/records/12773013 to facilitate further research.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

67 extracted references · 62 canonical work pages

  1. [1]

    British journal of hospital medicine81(2), 1–9 (2020)

    Mitchell, E., Walker, R.: Global ageing: successes, challenges and opportunities. British journal of hospital medicine81(2), 1–9 (2020)

  2. [2]

    Gait & posture26(2), 194–199 (2007)

    Bourke, A.K., O’brien, J., Lyons, G.M.: Evaluation of a threshold-based tri-axial accelerometer fall detection algorithm. Gait & posture26(2), 194–199 (2007)

  3. [3]

    ´A., Barros, L.M.: Impact of educational intervention in the perception of hospitalised patients about the risk of falling and associated factors

    Ximenes, M.A.M., Oliveira, I.K.M., Cavalcante, F.M.L., Neto, N.M.G., Cae- tano, J. ´A., Barros, L.M.: Impact of educational intervention in the perception of hospitalised patients about the risk of falling and associated factors. Authorea Preprints (2024)

  4. [4]

    In: 2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp

    Noury, N., Fleury, A., Rumeau, P., Bourke, A.K., Laighin, G., Rialle, V., Lundy, J.-E.: Fall detection-principles and methods. In: 2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp. 1663– 1666 (2007). IEEE

  5. [5]

    Neurocomputing100, 144–152 (2013)

    Mubashir, M., Shao, L., Seed, L.: A survey on fall detection: Principles and approaches. Neurocomputing100, 144–152 (2013)

  6. [6]

    IEEE Sensors Journal (2023)

    Wang, X., Talavera, E., Karastoyanova, D., Azzopardi, G.: Fall detection with a non-intrusive and first-person vision approach. IEEE Sensors Journal (2023)

  7. [7]

    Measurement192, 34 110870 (2022)

    Yang, Y., Yang, H., Liu, Z., Yuan, Y., Guan, X.: Fall detection system based on infrared array sensor and multi-dimensional feature fusion. Measurement192, 34 110870 (2022)

  8. [8]

    Wang, Y., Sarvari, P.A., Khadraoui, D.: Fusion of machine learning and threshold- based approaches for fall detection in healthcare using inertial sensors

Show all 67 references
  1. [9]

    IEEE Access7, 77702–77722 (2019)

    Ren, L., Peng, Y.: Research of fall detection and fall prevention technologies: A systematic review. IEEE Access7, 77702–77722 (2019)

  2. [10]

    IEEE transactions on cognitive and developmental systems12(3), 588–600 (2019)

    Kottari, K.N., Delibasis, K.K., Maglogiannis, I.G.: Real-time fall detection using uncalibrated fisheye cameras. IEEE transactions on cognitive and developmental systems12(3), 588–600 (2019)

  3. [11]

    Challenges and Trends in Multimodal Fall Detection for Healthcare, 97–120 (2020)

    Espinosa, R., Ponce, H., Guti´ errez, S., Mart´ ınez-Villase˜ nor, L., Brieva, J., Moya- Albor, E.: Application of convolutional neural networks for fall detection using multiple cameras. Challenges and Trends in Multimodal Fall Detection for Healthcare, 97–120 (2020)

  4. [12]

    Mathematics12(24), 3896 (2024)

    Ma, X., Zhang, L., Zhan, J., Chang, S.: Application of dual-stage attention tempo- ral convolutional networks in gas well production prediction. Mathematics12(24), 3896 (2024)

  5. [13]

    Sensors24(24), 8051 (2024)

    Kibet, D., So, M.S., Kang, H., Han, Y., Shin, J.-H.: Sudden fall detection of human body using transformer model. Sensors24(24), 8051 (2024)

  6. [14]

    International Journal of Computer Information Systems and Industrial Management Applications8, 195–204 (2016)

    Mourchid, Y., El Hassouni, M., Cherifi, H.: Image segmentation based on commu- nity detection approach. International Journal of Computer Information Systems and Industrial Management Applications8, 195–204 (2016)

  7. [15]

    In: Complex Networks & Their Applications IX: Volume 1, Proceedings of the Ninth International Con- ference on Complex Networks and Their Applications COMPLEX NETWORKS 2020, pp

    Lafhel, M., Cherifi, H., Renoust, B., El Hassouni, M., Mourchid, Y.: Movie script similarity using multilayer network portrait divergence. In: Complex Networks & Their Applications IX: Volume 1, Proceedings of the Ninth International Con- ference on Complex Networks and Their ...

  8. [16]

    In: Proceedings of Complex Networks 2017 The Sixth Interna- tional Conference on Complex Networks and Their Applications

    Cherifi, C., Cherifi, H., Karsai, M., Musolesi, M.: Complex networks and their applications vi. In: Proceedings of Complex Networks 2017 The Sixth Interna- tional Conference on Complex Networks and Their Applications. Berlin: Springer (2017). Springer

  9. [17]

    Computers in Biology and Medicine165, 107420 (2023)

    Mourchid, Y., Slama, R.: D-stgcnt: A dense spatio-temporal graph conv-gru network based on transformer for assessment of patient physical rehabilitation. Computers in Biology and Medicine165, 107420 (2023)

  10. [18]

    Symmetry12(5), 744 (2020)

    Chen, W., Jiang, Z., Guo, H., Ni, X.: Fall detection based on key points of human- skeleton using openpose. Symmetry12(5), 744 (2020)

  11. [19]

    IEEE Access8, 166117–166137 (2020)

    Islam, M.M., Tayan, O., Islam, M.R., Islam, M.S., Nooruddin, S., Kabir, M.N., 35 Islam, M.R.: Deep learning based systems developed for fall detection: a review. IEEE Access8, 166117–166137 (2020)

  12. [20]

    Medical engineering & physics30(1), 84–90 (2008)

    Bourke, A.K., Lyons, G.M.: A threshold-based fall-detection algorithm using a bi-axial gyroscope sensor. Medical engineering & physics30(1), 84–90 (2008)

  13. [21]

    In: 2022 6th International Conference on Information Technology (InCIT), pp

    Mekruksavanich, S., Jantawong, P., Jitpattanakul, A.: Pre-impact fall detection based on wearable inertial sensors using hybrid deep residual neural network. In: 2022 6th International Conference on Information Technology (InCIT), pp. 450–453 (2022). IEEE

  14. [22]

    In: Proceedings of the 2009 Sixth International Workshop on Wearable and Implantable Body Sensor Networks (2009)

    Li, Q., Stankovic, J.A., Hanson, M.A., Barth, A.T., Lach, J., Zhou, G.: Accurate, fast fall detection using gyroscopes and accelerometer-derived posture informa- tion. In: Proceedings of the 2009 Sixth International Workshop on Wearable and Implantable Body Sensor Networks (20...

  15. [23]

    Digital Health8, 20552076221074128 (2022) https://doi.org/10.1177/20552076221074128

    Lim, A.C.Y., Natarajan, P., Fonseka, R.D., Maharaj, M., Mobbs, R.J.: The application of artificial intelligence and custom algorithms with inertial wear- able devices for gait analysis and detection of gait-altering pathologies in adults: A scoping review of literature. Digita...

  16. [24]

    Li, H., Ma, J., Ren, X., Wang, K.: Novel fall detection algorithm based on multi- threshold fall model (2023)

  17. [25]

    Journal of Real-Time Image Processing9(4), 635–646 (2014) https://doi.org/10

    Mastorakis, G., Makris, D.: Fall detection system using kinect’s infrared sensor. Journal of Real-Time Image Processing9(4), 635–646 (2014) https://doi.org/10. 1007/s11554-012-0246-9

  18. [26]

    Multime- dia Tools and Applications81(4), 5113–5136 (2022) https://doi.org/10.1007/ s11042-021-11646-w

    De, A., Saha, A., Kumar, P.: Fall detection approach based on combined displacement of spatial features for intelligent indoor surveillance. Multime- dia Tools and Applications81(4), 5113–5136 (2022) https://doi.org/10.1007/ s11042-021-11646-w

  19. [27]

    IEEE Sensors Journal20(13), 6889–6919 (2020) https://doi.org/10.1109/JSEN.2020.2975522

    Singh, A., Rehman, S.U., Yongchareon, S., Chong, P.H.J.: Sensor technologies for fall detection systems: A review. IEEE Sensors Journal20(13), 6889–6919 (2020) https://doi.org/10.1109/JSEN.2020.2975522

  20. [28]

    Internet of Things9, 100130 (2020)

    Nooruddin, S., Islam, M.M., Sharna, F.A.: An iot based device-type invariant fall detection system. Internet of Things9, 100130 (2020)

  21. [29]

    In: 2022 IEEE Symposium on Future Telecommunication Technologies (SOFTT), pp

    Wasi, M.W.I., Dziyauddin, R.A., Amir, N.I.M., Ahmad, R.: Machine learning algorithm for fall classification using wearable device. In: 2022 IEEE Symposium on Future Telecommunication Technologies (SOFTT), pp. 62–66 (2022). IEEE

  22. [30]

    Procedia Computer Science203, 16–23 (2022)

    Kausar, F., Awadalla, M., Mesbah, M., AlBadi, T.: Automated machine learning 36 based elderly fall detection classification. Procedia Computer Science203, 16–23 (2022)

  23. [31]

    Sensors18(1), 20 (2017)

    Putra, I.P.E.S., Brusey, J., Gaura, E., Vesilo, R.: An event-triggered machine learning approach for accelerometer-based fall detection. Sensors18(1), 20 (2017)

  24. [32]

    Computational intelligence and neuroscience2017(2017)

    Droghini, D., Ferretti, D., Principi, E., Squartini, S., Piazza, F., et al.: A com- bined one-class svm and template-matching approach for user-aided human fall detection by means of floor acoustic features. Computational intelligence and neuroscience2017(2017)

  25. [33]

    Instrumentation Science & Technology48(1), 22–42 (2020)

    Kim, I., Lee, K.-S., Kim, K., Kim, K., Chae, H.-S., Kim, H.-C.: Implementation of a real-time fall detection system for elderly korean farmers using an insole- integrated sensing device. Instrumentation Science & Technology48(1), 22–42 (2020)

  26. [34]

    Multimedia Tools and Applications81(18), 26081–26100 (2022) https://doi.org/10.1007/ s11042-022-11914-3

    De, A., Saha, A., Kumar, P., Pal, G.: Fall detection method based on spatio- temporal feature fusion using combined two-channel classification. Multimedia Tools and Applications81(18), 26081–26100 (2022) https://doi.org/10.1007/ s11042-022-11914-3

  27. [35]

    IEEE Access (2025)

    Mir, A.A., Khalid, A.S., Musa, S., Fauzi, M.F.A., Razak, N.N., Tang, T.B.: Machine learning in ambient assisted living for enhanced elderly healthcare: A systematic literature review. IEEE Access (2025)

  28. [36]

    Artificial Intelligence Review55(4), 3369–3400 (2022) https://doi.org/10.1007/ s10462-021-10093-5

    Miranda, L., Viterbo, J., Bernardini, F.: A survey on the use of machine learning methods in context-aware middlewares for human activity recognition. Artificial Intelligence Review55(4), 3369–3400 (2022) https://doi.org/10.1007/ s10462-021-10093-5

  29. [37]

    IEEE Internet of Things Journal (2025)

    Li, P., Jiang, M., Lin, H., Lv, X., Huang, J.: Multidimensional time series seg- mentation of human activity without prior knowledge. IEEE Internet of Things Journal (2025)

  30. [38]

    In: 2023 IEEE 25th International Workshop on Multimedia Signal Processing (MMSP), pp

    Koffi, T.Y., Mourchid, Y., Hindawi, M., Dupuis, Y.: Machine learning and feature ranking for impact fall detection event using multisensor data. In: 2023 IEEE 25th International Workshop on Multimedia Signal Processing (MMSP), pp. 1–6 (2023). IEEE

  31. [39]

    In: Complex, Intelligent, and Software Intensive Systems: Proceedings of the 12th International Conference on Complex, Intelligent, and Software Intensive Systems (CISIS-2018), pp

    Shen, L., Zhang, Q., Cao, G., Xu, H.: Fall detection system based on deep learning and image processing in cloud environment. In: Complex, Intelligent, and Software Intensive Systems: Proceedings of the 12th International Conference on Complex, Intelligent, and Software Intens...

  32. [40]

    37 Biomedical Signal Processing and Control71, 103242 (2022)

    S ¸eng¨ ul, G., Karakaya, M., Misra, S., Abayomi-Alli, O.O., Damaˇ seviˇ cius, R.: Deep learning based fall detection using smartwatches for healthcare applications. 37 Biomedical Signal Processing and Control71, 103242 (2022)

  33. [41]

    In: Journal of Physics: Conference Series, vol

    Xu, J., He, Z., Zhang, Y.: Cnn-lstm combined network for iot enabled fall detec- tion applications. In: Journal of Physics: Conference Series, vol. 1267, p. 012044 (2019). IOP Publishing

  34. [42]

    Diagnostics 13(17), 2746 (2023)

    Balasubramaniam, S., Velmurugan, Y., Jaganathan, D., Dhanasekaran, S.: A modified lenet cnn for breast cancer diagnosis in ultrasound images. Diagnostics 13(17), 2746 (2023)

  35. [43]

    Scientific Reports14(1), 21537 (2024)

    Almukadi, W.S., Alrowais, F., Saeed, M.K., Yahya, A.E., Mahmud, A., Marzouk, R.: Deep feature fusion with computer vision driven fall detection approach for enhanced assisted living safety. Scientific Reports14(1), 21537 (2024)

  36. [44]

    Journal of biomechanical engineering141(8), 081010 (2019)

    Kim, T.H., Choi, A., Heo, H.M., Kim, K., Lee, K., Mun, J.H.: Machine learning- based pre-impact fall detection model to discriminate various types of fall. Journal of biomechanical engineering141(8), 081010 (2019)

  37. [45]

    Frontiers in bioengineering and biotechnology8, 63 (2020)

    Yu, X., Qiu, H., Xiong, S.: A novel hybrid deep neural network to predict pre-impact fall for older people based on wearable inertial sensors. Frontiers in bioengineering and biotechnology8, 63 (2020)

  38. [46]

    In: 2025 Design, Automation & Test in Europe Conference (DATE), pp

    Turetta, C., Ali, M.T., Demrozi, F., Pravadelli, G.: A lightweight cnn for real- time pre-impact fall detection. In: 2025 Design, Automation & Test in Europe Conference (DATE), pp. 1–7 (2025). IEEE

  39. [47]

    In: ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp

    Chi, T.-H., Liu, K.-C., Hsieh, C.-Y., Tsao, Y., Chan, C.-T.: Prefallkd: Pre-impact fall detection via cnn-vit knowledge distillation. In: ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1–5 (2023). IEEE

  40. [48]

    arXiv preprint arXiv:2403.06994 (2024)

    Qu, Z., Huang, T., Ji, Y., Li, Y.: Physics sensor based deep learning fall detection system. arXiv preprint arXiv:2403.06994 (2024)

  41. [49]

    Mathematics11(8), 1965 (2023)

    Wang, Z., Lv, Y., Wu, R., Zhang, Y.: Review of grabcut in image processing. Mathematics11(8), 1965 (2023)

  42. [50]

    Vyas, P.: Pose estimation and action recognition in sports and fitness (2019)

  43. [51]

    arXiv preprint arXiv:2206.11678 (2022)

    Grishchenko, I., Bazarevsky, V., Zanfir, A., Bazavan, E.G., Zanfir, M., Yee, R., Raveendran, K., Zhdanovich, M., Grundmann, M., Sminchisescu, C.: Blazepose ghum holistic: Real-time 3d human landmarks and pose estimation. arXiv preprint arXiv:2206.11678 (2022)

  44. [52]

    arXiv preprint arXiv:1906.08172 (2019) 38

    Lugaresi, C., Tang, J., Nash, H., McClanahan, C., Uboweja, E., Hays, M., Zhang, F., Chang, C.-L., Yong, M.G., Lee, J., et al.: Mediapipe: A framework for building perception pipelines. arXiv preprint arXiv:1906.08172 (2019) 38

  45. [53]

    arXiv preprint arXiv:2006.10204 (2020)

    Bazarevsky, V., Grishchenko, I., Raveendran, K., Zhu, T., Zhang, F., Grund- mann, M.: Blazepose: On-device real-time body pose tracking. arXiv preprint arXiv:2006.10204 (2020)

  46. [54]

    PhD thesis, Carnegie Mellon University Pittsburgh, PA, USA (2019)

    Martınez, G.H.: Openpose: Whole-body pose estimation. PhD thesis, Carnegie Mellon University Pittsburgh, PA, USA (2019)

  47. [55]

    IEEE Sensors Letters4(6), 1–4 (2020)

    Lin, C.-L., Chiu, W.-C., Chen, F.-H., Ho, Y.-H., Chu, T.-C., Hsieh, P.-H.: Fall monitoring for the elderly using wearable inertial measurement sensors on eyeglasses. IEEE Sensors Letters4(6), 1–4 (2020)

  48. [56]

    Applied Sciences12(21), 11031 (2022)

    Jeong, S.S., Kim, N.H., Yu, Y.S.: Fall detection system based on simple threshold method and long short-term memory: Comparison with hidden markov model and extraction of optimal parameters. Applied Sciences12(21), 11031 (2022)

  49. [57]

    Abdo, H.A., Amin, K., Hamad, A.M.: Human fall detection using spatial tempo- ral graph convolutional networks. IJCI. International Journal of Computers and Information10(2), 80–98 (2023)

  50. [58]

    IEEE Access 9, 28224–28236 (2021)

    Keskes, O., Noumeir, R.: Vision-based fall detection using st-gcn. IEEE Access 9, 28224–28236 (2021)

  51. [59]

    Sensors19(9), 1988 (2019)

    Mart´ ınez-Villase˜ nor, L., Ponce, H., Brieva, J., Moya-Albor, E., N´ u˜ nez-Mart´ ınez, J., Pe˜ nafort-Asturiano, C.: Up-fall detection dataset: A multimodal approach. Sensors19(9), 1988 (2019)

  52. [60]

    Procedia Computer Science 110, 32–39 (2017)

    Casilari, E., Santoyo-Ram´ on, J.A., Cano-Garc´ ıa, J.M.: Umafall: A multisensor dataset for the research on automatic fall detection. Procedia Computer Science 110, 32–39 (2017)

  53. [61]

    International Journal of Advanced Computer Science and Applications12(6), 599–606 (2021)

    Vujovi´ c,ˇZ.,et al.: Classification model evaluation metrics. International Journal of Advanced Computer Science and Applications12(6), 599–606 (2021)

  54. [62]

    BMC genomics21, 1–13 (2020)

    Chicco, D., Jurman, G.: The advantages of the matthews correlation coeffi- cient (mcc) over f1 score and accuracy in binary classification evaluation. BMC genomics21, 1–13 (2020)

  55. [63]

    Multimedia Tools and Applications83(6), 18091–18118 (2024)

    Ha, T.V., Nguyen, H.M., Thanh, S.H., Nguyen, B.T.: Fall detection using mix- tures of convolutional neural networks. Multimedia Tools and Applications83(6), 18091–18118 (2024)

  56. [64]

    In: WAMWB@ MobileHCI, pp

    Castro, F., Dentamaro, V., Gattulli, V., Impedovo, D.: Fall detection with lstm and attention mechanism. In: WAMWB@ MobileHCI, pp. 37–50 (2023)

  57. [65]

    Computers, Materials 39 & Continua75(2) (2023)

    Eltahir, M.M., Yousif, A., Alrowais, F., Nour, M.K., Marzouk, R., Dafaalla, H., Hassan Elnour, A.A., Aziz, A.S.A., Hamza, M.A.: Deep transfer learning-enabled activity identification and fall detection for disabled people. Computers, Materials 39 & Continua75(2) (2023)

  58. [66]

    International Journal of Technology13(6), 1173–1182 (2022)

    Lau, X.L., Connie, T., Goh, M.K.O., Lau, S.H.: Fall detection and motion analysis using visual approaches. International Journal of Technology13(6), 1173–1182 (2022)

  59. [67]

    In: International Conference on Machine Learning, pp

    Balın, M.F., Abid, A., Zou, J.: Concrete autoencoders: Differentiable feature selec- tion and reconstruction. In: International Conference on Machine Learning, pp. 444–453 (2019). PMLR 40

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.