pith. sign in

arxiv: 2511.22554 · v3 · submitted 2025-11-27 · 💻 cs.NE

Privacy-preserving fall detection at the edge using Sony IMX636 event-based vision sensor and Intel Loihi 2 neuromorphic processor

Lyes Khacef , Philipp Weidel , Susumu Hogyoku , Harry Liu , Claire Alexandra Br\"auer , Shunsuke Koshino , Takeshi Oyakawa , Vincent Parret
show 3 more authors
Yoshitaka Miyatani Mike Davies Mathis Richter
This is my paper

Pith reviewed 2026-05-17 04:43 UTC · model grok-4.3

classification 💻 cs.NE
keywords fall detectionevent-based visionneuromorphic computingedge AIprivacy preservationLoihi 2spiking neural networksstate space models
0
0 comments X

The pith

A neuromorphic edge system using event-based sensing and Loihi 2 achieves 84% F1 score in fall detection at 90 mW power consumption.

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

The paper develops a fall detection system for elderly care that processes data directly at the vision sensor to avoid sending images to the cloud, thereby preserving privacy. It integrates the Sony IMX636 event-based sensor with the Intel Loihi 2 neuromorphic processor through an FPGA interface to exploit data sparsity for efficient, low-power operation. Experiments on a new dataset show that different neural network designs trade off accuracy against computational cost, with one hybrid approach reaching the best accuracy while staying within the power limits of the chip. This matters because always-on monitoring in homes requires both reliability and minimal energy use without compromising resident privacy.

Core claim

The central discovery is that combining an MCUNet feature extractor with patched inference and an S4D state space model on the Loihi 2 processor yields an F1 score of 84% for fall detection, with 2x synaptic operations sparsity and a total power draw of 90 mW, while an LIF convolutional SNN with graded spikes provides higher efficiency at lower accuracy.

What carries the argument

The FPGA-based interface connecting the IMX636 event sensor to the Loihi 2 chip, which enables asynchronous sparse event processing for real-time inference.

Load-bearing premise

The newly recorded dataset under diverse conditions is representative of real-world elderly fall events.

What would settle it

An independent test on a larger set of real home falls that drops the MCUNet-S4D F1 score below 70% would disprove the performance claims.

Figures

Figures reproduced from arXiv: 2511.22554 by Claire Alexandra Br\"auer, Harry Liu, Lyes Khacef, Mathis Richter, Mike Davies, Philipp Weidel, Shunsuke Koshino, Susumu Hogyoku, Takeshi Oyakawa, Vincent Parret, Yoshitaka Miyatani.

Figure 1
Figure 1. Figure 1: Hardware system overview showing the whole system [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Hardware system pipeline with details of Max10 FPGA interface. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Input-patched inference of MCU13B model for a single event frame, [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Spike (activation) functions (left) and surrogate gradient functions (right) of LIF neurons with binary and graded spikes. SLAYER is a form of Backpropagation Through Time (BPTT) which uses surrogate gradients in the backward pass [47], as shown in [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Fall detection algorithmic-level benchmarking at 16 predictions/s of single Loihi 2 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
read the original abstract

Fall detection for elderly care using non-invasive vision-based systems remains an important yet unsolved problem. Driven by strict privacy requirements, inference must run at the edge of the vision sensor, demanding robust, real-time, and always-on perception under tight hardware constraints. To address these challenges, we propose a neuromorphic fall detection system that integrates the Sony IMX636 event-based vision sensor with the Intel Loihi 2 neuromorphic processor via a dedicated FPGA-based interface, leveraging the sparsity of event data together with near-memory asynchronous processing. Using a newly recorded dataset under diverse environmental conditions, we explore the design space of sparse neural networks deployable on a single Loihi 2 chip and analyze the tradeoffs between detection F1 score and computational cost. Notably, on the Pareto front, our LIF-based convolutional SNN with graded spikes achieves the highest computational efficiency, reaching a 55x synaptic operations sparsity for an F1 score of 58%. The LIF with graded spikes shows a gain of 6% in F1 score with 5x less operations compared to binary spikes. Furthermore, our MCUNet feature extractor with patched inference, combined with the S4D state space model, achieves the highest F1 score of 84% with a synaptic operations sparsity of 2x and a total power consumption of 90 mW on Loihi 2. Overall, our smart security camera proof-of-concept highlights the potential of integrating neuromorphic sensing and processing for edge AI applications where latency, energy consumption, and privacy are critical.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a neuromorphic edge system for privacy-preserving fall detection that integrates the Sony IMX636 event-based vision sensor with the Intel Loihi 2 processor through an FPGA interface. On a newly recorded dataset under diverse conditions, the authors explore sparse SNN architectures and report concrete hardware results, including a LIF convolutional SNN with graded spikes reaching 55x synaptic sparsity at 58% F1, a 6% F1 gain with 5x fewer operations versus binary spikes, and an MCUNet feature extractor plus patched inference combined with an S4D state-space model achieving the highest F1 of 84% at 2x sparsity and 90 mW total power on Loihi 2.

Significance. If the central performance numbers hold under proper validation, the work provides one of the first end-to-end hardware demonstrations of event-based sensing plus neuromorphic processing for always-on, low-power fall detection, with explicit sparsity and power measurements that quantify the practical trade-offs for privacy-critical edge AI.

major comments (2)
  1. [Dataset and Experimental Setup] The central 84% F1 claim (abstract and results) rests on a newly recorded dataset whose representativeness for real elderly falls is not established. No quantitative statistics are supplied on total event counts, fall/non-fall balance, subject demographics, body types, clothing/occlusion variability, labeling protocol, or comparison to public fall datasets; without these the reported F1 cannot be trusted to generalize beyond the recording conditions.
  2. [Results and Evaluation] Results section: the headline F1 scores (84%, 58%) and sparsity factors are presented without cross-validation details, error bars, or statistical significance tests. This weakens the Pareto-front comparison between the LIF graded-spike model and the MCUNet+S4D configuration.
minor comments (2)
  1. [Abstract] Abstract: the power figure of 90 mW should be accompanied by a brief breakdown (sensor + FPGA + Loihi 2) to allow readers to assess the contribution of each component.
  2. [Methods] Notation: the term 'synaptic operations sparsity' is used without an explicit definition or reference to how it is computed from the event-driven execution on Loihi 2.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment below and have revised the manuscript to incorporate additional details and analyses where possible.

read point-by-point responses

  1. Referee: [Dataset and Experimental Setup] The central 84% F1 claim (abstract and results) rests on a newly recorded dataset whose representativeness for real elderly falls is not established. No quantitative statistics are supplied on total event counts, fall/non-fall balance, subject demographics, body types, clothing/occlusion variability, labeling protocol, or comparison to public fall datasets; without these the reported F1 cannot be trusted to generalize beyond the recording conditions.

    Authors: We agree that quantitative dataset statistics are necessary to support claims of representativeness. In the revised manuscript we have added a new subsection under Experimental Setup that reports: total event counts across the recording sessions, fall versus non-fall sample balance, subject demographics (age range 65–85 years, gender distribution, body types), clothing and occlusion variations, the multi-annotator labeling protocol with inter-rater agreement metrics, and a direct comparison table against existing public fall datasets. These additions allow readers to assess the scope of the 84 % F1 result. We note that the dataset was collected under controlled yet diverse indoor/outdoor conditions as a proof-of-concept; we do not claim exhaustive coverage of all real-world elderly fall scenarios. revision: yes

  2. Referee: [Results and Evaluation] Results section: the headline F1 scores (84%, 58%) and sparsity factors are presented without cross-validation details, error bars, or statistical significance tests. This weakens the Pareto-front comparison between the LIF graded-spike model and the MCUNet+S4D configuration.

    Authors: We accept that the original presentation lacked sufficient statistical rigor. The revised Results section now specifies the cross-validation protocol (subject-wise 5-fold cross-validation), includes error bars showing standard deviation across folds for all F1 and sparsity values, and reports statistical significance tests (paired t-tests) between model variants with associated p-values. These changes provide a clearer and more defensible basis for the Pareto-front comparisons and the reported gains (e.g., the 6 % F1 improvement with graded spikes). revision: yes

Circularity Check

0 steps flagged

No circularity: results are direct empirical hardware measurements on recorded dataset

full rationale

The paper reports experimental outcomes from training and deploying sparse neural networks (LIF SNNs, MCUNet + S4D) on a newly recorded fall-detection dataset, with F1 scores, sparsity, and power measured directly on Loihi 2 hardware via FPGA interface. No mathematical derivation chain, fitted parameters renamed as predictions, self-definitional equations, or load-bearing self-citations appear in the abstract or described results; the central claims rest on observed performance metrics rather than quantities forced by construction from the inputs themselves.

Axiom & Free-Parameter Ledger

1 free parameters · 0 axioms · 0 invented entities

The central claims rest on standard assumptions from neuromorphic engineering and event-based vision; no new physical entities or ad-hoc axioms are introduced beyond typical hardware and dataset assumptions.

free parameters (1)
  • SNN architecture hyperparameters
    Network sizes, spike thresholds, and learning rates chosen to fit the Loihi 2 constraints and dataset.

pith-pipeline@v0.9.0 · 5632 in / 1132 out tokens · 46928 ms · 2026-05-17T04:43:24.320459+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

63 extracted references · 63 canonical work pages · 1 internal anchor

  1. [1]

    Chapter 1 - the global demography of aging: Facts, explanations, future,

    D. Bloom and D. Luca, “Chapter 1 - the global demography of aging: Facts, explanations, future,” ser. Handbook of the Economics of Population Aging, J. Piggott and A. Woodland, Eds. North-Holland, 2016, vol. 1, pp. 3–56. [Online]. Available: https://www.sciencedirect. com/science/article/pii/S2212007616300062

    work page 2016

  2. [2]

    The methods of fall detection: A literature review,

    N. T. Newaz and E. Hanada, “The methods of fall detection: A literature review,”Sensors, vol. 23, no. 11, 2023. [Online]. Available: https://www.mdpi.com/1424-8220/23/11/5212

    work page 2023

  3. [3]

    [Online]

    “Falls,”World Health Organization, 2021. [Online]. Available: https://www.who.int/news-room/fact-sheets/detail/falls

    work page 2021

  4. [4]

    Elderly fall detection systems: A literature survey,

    X. Wang, J. Ellul, and G. Azzopardi, “Elderly fall detection systems: A literature survey,”Frontiers in Robotics and AI, vol. V olume 7 - 2020, 2020. [Online]. Available: https://www.frontiersin.org/journals/ robotics-and-ai/articles/10.3389/frobt.2020.00071

    work page doi:10.3389/frobt.2020.00071 2020

  5. [5]

    A survey on fall detection: Principles and approaches,

    M. Mubashir, L. Shao, and L. Seed, “A survey on fall detection: Principles and approaches,”Neurocomputing, vol. 100, pp. 144– 152, 2013, special issue: Behaviours in video. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0925231212003153

    work page 2013

  6. [6]

    A comprehensive review of elderly fall detection using wireless communication and artificial intelligence techniques,

    S. K. Gharghan and H. A. Hashim, “A comprehensive review of elderly fall detection using wireless communication and artificial intelligence techniques,”Measurement, vol. 226, p. 114186, 2024. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S0263224124000708

    work page 2024

  7. [7]

    Falls and fear of falling: Which comes first? a longitudinal prediction model suggests strategies for primary and secondary prevention,

    S. M. Friedman, B. Munoz, S. K. West, G. S. Rubin, and L. P. Fried, “Falls and fear of falling: Which comes first? a longitudinal prediction model suggests strategies for primary and secondary prevention,” Journal of the American Geriatrics Society, vol. 50, no. 8, pp. 1329–1335, 2002. [Online]. Available: https://agsjournals.onlinelibrary. wiley.com/doi/...

    work page doi:10.1046/j.1532-5415.2002.50352.x 2002

  8. [8]

    A survey of depth and inertial sensor fusion for human action recognition,

    C. Chen, R. Jafari, and N. Kehtarnavaz, “A survey of depth and inertial sensor fusion for human action recognition,”Multimedia Tools and Applications, vol. 76, 02 2017

    work page 2017

  9. [9]

    Vision-based human fall detection systems using deep learning: A review,

    E. Alam, A. Sufian, P. Dutta, and M. Leo, “Vision-based human fall detection systems using deep learning: A review,”Computers in Biology and Medicine, vol. 146, p. 105626, 2022. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0010482522004188

    work page 2022

  10. [10]

    Video-based human fall detection in smart homes using deep learn- ing,

    A. Shojaei-Hashemi, P. Nasiopoulos, J. J. Little, and M. T. Pourazad, “Video-based human fall detection in smart homes using deep learn- ing,” in2018 IEEE International Symposium on Circuits and Systems (ISCAS), 2018, pp. 1–5

    work page 2018

  11. [11]

    Security and privacy in distributed smart cameras,

    D. N. Serpanos and A. Papalambrou, “Security and privacy in distributed smart cameras,”Proceedings of the IEEE, vol. 96, no. 10, pp. 1678– 1687, 2008

    work page 2008

  12. [12]

    Challenges, issues and trends in fall detection systems,

    R. Igual, C. Medrano, and I. Plaza, “Challenges, issues and trends in fall detection systems,”Biomedical engineering online, vol. 12, p. 66, 07 2013

    work page 2013

  13. [13]

    New advances and challenges of fall detection systems: A survey,

    T. Xu, Y . Zhou, and J. Zhu, “New advances and challenges of fall detection systems: A survey,”Applied Sciences, vol. 8, no. 3, 2018. [Online]. Available: https://www.mdpi.com/2076-3417/8/3/418

    work page 2018

  14. [14]

    Finateu, A

    T. Finateu, A. Niwa, D. Matolin, K. Tsuchimoto, A. Mascheroni, E. Reynaud, P. Mostafalu, F. Brady, L. Chotard, F. LeGoff, H. Takahashi, H. Wakabayashi, Y . Oike, and C. Posch, “5.10 a 1280×720 back- illuminated stacked temporal contrast event-based vision sensor with 4.86µm pixels, 1.066geps readout, programmable event-rate controller and compressive data...

    work page 2020

  15. [15]

    Advancing neuromorphic computing with loihi: A survey of results and outlook,

    M. Davies, A. Wild, G. Orchard, Y . Sandamirskaya, G. Fonseca Guerra, P. Joshi, P. Plank, and S. Risbud, “Advancing neuromorphic computing with loihi: A survey of results and outlook,”Proceedings of the IEEE, vol. PP, pp. 1–24, 04 2021

    work page 2021

  16. [16]

    Event-based vision: A survey,

    G. Gallego, T. Delbr ¨uck, G. Orchard, C. Bartolozzi, B. Taba, A. Censi, S. Leutenegger, A. J. Davison, J. Conradt, K. Daniilidis, and D. Scaramuzza, “Event-based vision: A survey,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 44, no. 1, p. 154–180, Jan. 2022. [Online]. Available: https://doi.org/10.1109/TPAMI.2020.3008413

    work page doi:10.1109/tpami.2020.3008413 2022

  17. [17]

    An application-driven survey on event-based neuromorphic computer vision,

    D. Cazzato and F. Bono, “An application-driven survey on event-based neuromorphic computer vision,”Information, vol. 15, no. 8, 2024. [Online]. Available: https://www.mdpi.com/2078-2489/15/8/472

    work page 2024

  18. [18]

    Event-based eye tracking. 2025 event-based vision workshop,

    Q. Chen, C. Gao, M. Liu, D. Perrone, Y . R. Pei, Z. Wang, Z. Zou, S. Tan, T. Han, G. Lu, Z. Xu, J. Ding, Z. Wang, Z. Wu, H. Han, Y . Wu, J. Chen, W. Zhai, Y . Cao, Z.-j. Zha, N. Bandara, T. Kandappu, A. Misra, X. Lin, H. Huang, H. Ren, B. Cheng, H. M. Truong, V .-T. Ly, H. G. Tran, T.-P. Nguyen, and T. T. Doan, “Event-based eye tracking. 2025 event-based ...

    work page 2025

  19. [19]

    Temporal-guided spiking neural networks for event-based human action recognition,

    S. Yang, S. Lu, S. Wang, M. H. Er, Z. Zheng, and A. C. Kot, “Temporal-guided spiking neural networks for event-based human action recognition,” 2025. [Online]. Available: https://arxiv.org/abs/2503.17132

    work page arXiv 2025

  20. [20]

    2022 roadmap on neuromorphic computing and engineering

    D. V . Christensen, R. Dittmann, B. Linares-Barranco, A. Sebastian, M. Le Gallo, A. Redaelli, S. Slesazeck, T. Mikolajick, S. Spiga, S. Menzel, I. Valov, G. Milano, C. Ricciardi, S.-J. Liang, F. Miao, M. Lanza, T. J. Quill, S. T. Keene, A. Salleo, J. Grollier, D. Markovi ´c, A. Mizrahi, P. Yao, J. J. Yang, G. Indiveri, J. P. Strachan, S. Datta, E. Vianell...

    work page doi:10.1088/2634-4386/ac4a83 2022

  21. [21]

    Opportunities for neuromorphic computing algorithms and applica- tions,

    C. Schuman, S. Kulkarni, M. Parsa, J. Mitchell, P. Date, and B. Kay, “Opportunities for neuromorphic computing algorithms and applica- tions,”Nature Computational Science, vol. 2, pp. 10–19, 01 2022

    work page 2022

  22. [22]

    Neuromorphic computing at scale,

    D. Kudithipudi, C. Schuman, C. Vineyard, T. Pandit, C. Merkel, R. Kubendran, J. Aimone, G. Orchard, C. Mayr, R. Benosman, J. Hays, C. Young, C. Bartolozzi, A. Majumdar, S. Cardwell, M. Payvand, S. Buckley, S. Kulkarni, H. Gonzalez, and S. Furber, “Neuromorphic computing at scale,”Nature, vol. 637, pp. 801–812, 01 2025

    work page 2025

  23. [23]

    Fully neuromorphic vision and control for autonomous drone flight,

    F. Paredes-Vall ´es, J. J. Hagenaars, J. Dupeyroux, S. Stroobants, Y . Xu, and G. C. H. E. de Croon, “Fully neuromorphic vision and control for autonomous drone flight,”Science Robotics, vol. 9, no. 90, p. eadi0591, 2024. [Online]. Available: https://www.science.org/doi/abs/ 10.1126/scirobotics.adi0591

    work page doi:10.1126/scirobotics.adi0591 2024

  24. [24]

    Hand-gesture recognition based on emg and event-based camera sensor fusion: A benchmark in neuromorphic computing,

    E. Ceolini, C. Frenkel, S. B. Shrestha, G. Taverni, L. Khacef, M. Payvand, and E. Donati, “Hand-gesture recognition based on emg and event-based camera sensor fusion: A benchmark in neuromorphic computing,”Frontiers in Neuroscience, vol. V olume 14 - 2020, 2020. [Online]. Available: https://www.frontiersin.org/journals/neuroscience/ articles/10.3389/fnins...

    work page doi:10.3389/fnins.2020.00637 2020

  25. [25]

    Towards low-latency event-based obstacle avoidance on a fpga-drone,

    P. Bonazzi, C. V ogt, M. Jost, L. Khacef, F. Paredes-Valles, and M. Magno, “Towards low-latency event-based obstacle avoidance on a fpga-drone,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, June 2025, pp. 4947–4955

    work page 2025

  26. [26]

    Colibries: A milliwatts risc-v based embedded system lever- aging neuromorphic and neural networks hardware accelerators for low- latency closed-loop control applications,

    G. Rutishauser, R. Hunziker, A. Di Mauro, S. Bian, L. Benini, and M. Magno, “Colibries: A milliwatts risc-v based embedded system lever- aging neuromorphic and neural networks hardware accelerators for low- latency closed-loop control applications,” in2023 IEEE International Symposium on Circuits and Systems (ISCAS), 2023, pp. 1–5

    work page 2023

  27. [27]

    Spike-based dynamic computing with asynchronous sensing-computing neuromorphic chip,

    M. Yao, O. Richter, G. Zhao, N. Qiao, Y . Xing, W. Dingheng, T. Hu, W. Fang, T. Demirci, M. Marchi, T. Yan, C. Nielsen, S. Sheik, C. Wu, Y . Tian, B. Xu, and G. Li, “Spike-based dynamic computing with asynchronous sensing-computing neuromorphic chip,”Nature Commu- nications, vol. 15, 05 2024

    work page 2024

  28. [28]

    Low- power event-based face detection with asynchronous neuromorphic hardware,

    C. Caccavella, F. Paredes-Vall ´es, M. Cannici, and L. Khacef, “Low- power event-based face detection with asynchronous neuromorphic hardware,” in2024 International Joint Conference on Neural Networks (IJCNN), 2024, pp. 1–10

    work page 2024

  29. [29]

    Event-based vision sensor (evs)

    S. S. S. Corporation, “Event-based vision sensor (evs).” [Online]. Avail- able: https://www.sony-semicon.com/en/products/is/industry/evs.html

    work page

  30. [30]

    Stacked event-based vision sensor with the industry’s smallest pixels

    Prophesee, “Stacked event-based vision sensor with the industry’s smallest pixels.” [Online]. Available: https://www.prophesee.ai/2020/02/ 19/prophesee-sony-stacked-event-based-vision-sensor/

    work page 2020

  31. [31]

    Loihi: A neuromorphic manycore processor with on-chip learning,

    M. Davies, N. Srinivasa, T.-H. Lin, G. Chinya, Y . Cao, S. H. Choday, G. Dimou, P. Joshi, N. Imam, S. Jain, Y . Liao, C.-K. Lin, A. Lines, R. Liu, D. Mathaikutty, S. McCoy, A. Paul, J. Tse, G. Venkataramanan, Y .-H. Weng, A. Wild, Y . Yang, and H. Wang, “Loihi: A neuromorphic manycore processor with on-chip learning,”IEEE Micro, vol. 38, no. 1, pp. 82–99, 2018

    work page 2018

  32. [32]

    Loihi 2 technology brief

    I. N. C. Lab, “Loihi 2 technology brief.” [On- line]. Available: https://www.intel.com/content/www/us/en/research/ neuromorphic-computing-loihi-2-technology-brief.html

    work page

  33. [33]

    Bakhtiarifard, C

    S. B. Shrestha, J. Timcheck, E. P. Frady, L. Campos-Macias, and M. Davies, “Efficient video and audio processing with loihi 2,” inICASSP, 2024, pp. 13 481–13 485. [Online]. Available: https://doi.org/10.1109/ICASSP48485.2024.10448003

    work page doi:10.1109/icassp48485.2024.10448003 2024

  34. [34]

    Exploiting neuro-inspired dynamic sparsity for energy-efficient intelligent perception,

    S. Zhou, C. Gao, T. Delbruck, M. Verhelst, and S. Liu, “Exploiting neuro-inspired dynamic sparsity for energy-efficient intelligent perception,”Nature Communications, vol. 16, no. 1, p. 9928,

    work page

  35. [35]

    Available: https://doi.org/10.1038/s41467-025-65387-7

    [Online]. Available: https://doi.org/10.1038/s41467-025-65387-7

    work page doi:10.1038/s41467-025-65387-7

  36. [36]

    Mcunet: Tiny deep learning on iot devices,

    J. Lin, W.-M. Chen, Y . Lin, C. Gan, and S. Han, “Mcunet: Tiny deep learning on iot devices,”Advances in Neural Information Processing Systems, vol. 33, 2020

    work page 2020

  37. [37]

    Mcunetv2: Memory- efficient patch-based inference for tiny deep learning,

    J. Lin, W.-M. Chen, H. Cai, C. Gan, and S. Han, “Mcunetv2: Memory- efficient patch-based inference for tiny deep learning,” inAnnual Con- ference on Neural Information Processing Systems (NeurIPS), 2021

    work page 2021

  38. [38]

    Mobilenetv2: Inverted residuals and linear bottlenecks,

    M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “Mobilenetv2: Inverted residuals and linear bottlenecks,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2018

    work page 2018

  39. [39]

    A 23.5 tops/w depthwise separable convolution ac- celerator for event-based depth estimation,

    A. Brito, T. Yamasaki, U. Rancon, T. Masquelier, B. R. Cottereau, A. T. Do, and B. Wang, “A 23.5 tops/w depthwise separable convolution ac- celerator for event-based depth estimation,” in2025 IEEE International Symposium on Circuits and Systems (ISCAS), 2025, pp. 1–5

    work page 2025

  40. [40]

    Mamba-360: Survey of state space models as transformer alternative for long sequence modelling: Methods, applications, and challenges,

    B. N. Patro and V . S. Agneeswaran, “Mamba-360: Survey of state space models as transformer alternative for long sequence modelling: Methods, applications, and challenges,”Engineering Applications of Artificial Intelligence, vol. 159, p. 111279, 2025. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0952197625012801

    work page 2025

  41. [41]

    On the parameterization and initialization of diagonal state space models,

    A. Gu, A. Gupta, K. Goel, and C. R ´e, “On the parameterization and initialization of diagonal state space models,” inProceedings of the 36th International Conference on Neural Information Processing Systems, ser. NIPS ’22. Red Hook, NY , USA: Curran Associates Inc., 2022

    work page 2022

  42. [42]

    A diagonal structured state space model on loihi 2 for efficient streaming sequence processing,

    S. M. Meyer, P. Weidel, P. Plank, L. Campos-Macias, S. B. Shrestha, P. Stratmann, and M. Richter, “A diagonal structured state space model on loihi 2 for efficient streaming sequence processing,” 2024. [Online]. Available: https://arxiv.org/abs/2409.15022

    work page arXiv 2024

  43. [43]

    Gerstner and W

    W. Gerstner and W. M. Kistler,Spiking neuron models: Single neurons, populations, plasticity. Cambridge university press, 2002

    work page 2002

  44. [44]

    Impact of spiking neurons leakages and network recurrences on event-based spatio-temporal pattern recognition,

    M. S. Bouanane, D. Cherifi, E. Chicca, and L. Khacef, “Impact of spiking neurons leakages and network recurrences on event-based spatio-temporal pattern recognition,”Frontiers in Neuroscience, vol. V olume 17 - 2023, 2023. [Online]. Available: https://www.frontiersin. org/journals/neuroscience/articles/10.3389/fnins.2023.1244675

    work page doi:10.3389/fnins.2023.1244675 2023

  45. [45]

    P-spikessm: Harnessing probabilistic spiking state space models for long-range dependency tasks,

    M. Bal and A. Sengupta, “P-spikessm: Harnessing probabilistic spiking state space models for long-range dependency tasks,” inThe Thirteenth International Conference on Learning Representations, 2025. [Online]. Available: https://openreview.net/forum?id=Sf4ep9Udjf

    work page 2025

  46. [46]

    SiLIF: Structured State Space Model Dynamics and Parametrization for Spiking Neural Networks

    M. Fabre, L. Dudchenko, and E. Neftci, “Structured state space model dynamics and parametrization for spiking neural networks,” 2025. [Online]. Available: https://arxiv.org/abs/2506.06374

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  47. [47]

    Focal loss for dense object detection,

    T.-Y . Lin, P. Goyal, R. Girshick, K. He, and P. Doll ´ar, “Focal loss for dense object detection,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 42, no. 2, pp. 318–327, 2020

    work page 2020

  48. [48]

    Surrogate gradient learning in spiking neural networks: Bringing the power of gradient-based opti- mization to spiking neural networks,

    E. O. Neftci, H. Mostafa, and F. Zenke, “Surrogate gradient learning in spiking neural networks: Bringing the power of gradient-based opti- mization to spiking neural networks,”IEEE Signal Processing Magazine, vol. 36, no. 6, pp. 51–63, 2019

    work page 2019

  49. [49]

    Training spiking neural networks using lessons from deep learning,

    J. K. Eshraghian, M. Ward, E. O. Neftci, X. Wang, G. Lenz, G. Dwivedi, M. Bennamoun, D. S. Jeong, and W. D. Lu, “Training spiking neural networks using lessons from deep learning,”Proceedings of the IEEE, vol. 111, no. 9, pp. 1016–1054, 2023

    work page 2023

  50. [50]

    Slayer: Spike layer error reassignment in time,

    S. B. Shrestha and G. Orchard, “Slayer: Spike layer error reassignment in time,” inAdvances in Neural Information Processing Systems, S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa- Bianchi, and R. Garnett, Eds., vol. 31. Curran Associates, Inc.,

    work page

  51. [51]

    Available: https://proceedings.neurips.cc/paper files/ paper/2018/file/82f2b308c3b01637c607ce05f52a2fed-Paper.pdf

    [Online]. Available: https://proceedings.neurips.cc/paper files/ paper/2018/file/82f2b308c3b01637c607ce05f52a2fed-Paper.pdf

    work page 2018

  52. [52]

    Exodus: Stable and efficient training of spiking neural networks,

    F. C. Bauer, G. Lenz, S. Haghighatshoar, and S. Sheik, “Exodus: Stable and efficient training of spiking neural networks,” Frontiers in Neuroscience, vol. V olume 17 - 2023, 2023. [Online]. Available: https://www.frontiersin.org/journals/neuroscience/articles/10. 3389/fnins.2023.1110444

    work page arXiv 2023

  53. [53]

    Neurobench: A framework for benchmarking neuromorphic computing algorithms and systems,

    J. Yik, K. V . den Berghe, D. den Blanken, Y . Bouhadjar, M. Fabre, P. Hueber, W. Ke, M. A. Khoei, D. Kleyko, N. Pacik-Nelson, A. Pierro, P. Stratmann, P.-S. V . Sun, G. Tang, S. Wang, B. Zhou, S. H. Ahmed, G. V . Joseph, B. Leto, A. Micheli, A. K. Mishra, G. Lenz, T. Sun, Z. Ahmed, M. Akl, B. Anderson, A. G. Andreou, C. Bartolozzi, A. Basu, P. Bogdan, S....

    work page doi:10.1038/s41467-025-56739-4 2025

  54. [54]

    What uncertainties do we need in bayesian deep learning for computer vision?

    A. Kendall and Y . Gal, “What uncertainties do we need in bayesian deep learning for computer vision?” inAdvances in Neural Information Processing Systems, I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, Eds., vol. 30. Curran Associates, Inc., 2017. [Online]. Available: https://proceedings.neurips.cc/paper file...

    work page 2017

  55. [55]

    Realizing fully- integrated, low-power, event-based pupil tracking with neuromorphic hardware,

    F. Paredes-Valles, Y . Miyatani, and K. Y . W. Scheper, “Realizing fully- integrated, low-power, event-based pupil tracking with neuromorphic hardware,” 2025. [Online]. Available: https://arxiv.org/abs/2511.20175

    work page arXiv 2025

  56. [56]

    High speed and high dynamic range video with an event camera,

    H. Rebecq, R. Ranftl, V . Koltun, and D. Scaramuzza, “High speed and high dynamic range video with an event camera,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 6, pp. 1964– 1980, 2021

    work page 1964

  57. [57]

    Back to event basics: Self-supervised learning of image reconstruction for event cameras via photometric constancy,

    F. Paredes-Valles and G. C. H. E. de Croon, “Back to event basics: Self-supervised learning of image reconstruction for event cameras via photometric constancy,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2021, pp. 3446–3455

    work page 2021

  58. [58]

    Event-based video reconstruction via potential-assisted spiking neural network,

    L. Zhu, X. Wang, Y . Chang, J. Li, T. Huang, and Y . Tian, “Event-based video reconstruction via potential-assisted spiking neural network,” in 2022 IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion (CVPR), 2022, pp. 3584–3594

    work page 2022

  59. [59]

    Person re-identification without identification via event anonymization,

    S. Ahmad, P. Morerio, and A. Del Bue, “Person re-identification without identification via event anonymization,” inProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), October 2023, pp. 11 132–11 141

    work page 2023

  60. [60]

    Are snns really more energy-efficient than anns? an in-depth hardware- aware study,

    M. Dampfhoffer, T. Mesquida, A. Valentian, and L. Anghel, “Are snns really more energy-efficient than anns? an in-depth hardware- aware study,”IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 7, no. 3, pp. 731–741, 2023

    work page 2023

  61. [61]

    Neuromorphic computing, memory and sensing 2024,

    Y . Group, “Neuromorphic computing, memory and sensing 2024,”

    work page 2024

  62. [62]

    Available: https://www.yolegroup.com/product/report/ neuromorphic-computing-memory-and-sensing-2024/

    [Online]. Available: https://www.yolegroup.com/product/report/ neuromorphic-computing-memory-and-sensing-2024/

    work page 2024

  63. [63]

    The road to commercial success for neuromor- phic technologies,

    D. Muir and S. Sheik, “The road to commercial success for neuromor- phic technologies,”Nature Communications, vol. 16, 04 2025

    work page 2025