Pith. sign in

REVIEW 4 major objections 5 minor 49 references

EEvAct: Early Event-Based Action Recognition with High-Rate Two-Stream Spiking Neural Networks

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A two-stream spiking neural network hits 94.9% Top-1 accuracy on the THU EACT-50 event-action benchmark, beating the 3D CNN baseline by two points with a fifth of the parameters and under one percent of the synaptic operations.

desk verdict Useful benchmark and architecture, but the headline 2% SOTA margin is not established because the baseline is scored on full sequences while the proposed model is scored at a fixed 2.0 s window. read the letter →

arxiv 2507.07734 v1 pith:R5V7JWFA submitted 2025-07-10 cs.CV cs.NE

Michael Neumeier , Jules Lecomte , Nils Kazinski , Soubarna Banik , Bing Li , Axel von Arnim This is my paper
classification cs.CVcs.NE
keywords earlyactionrecognitionevent-basedvisionspikingneuralnetworkstwo-streamarchitectureTHUEACT-50neuromorphiccomputinggatedrecurrentunitssynapticoperations
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 show that spiking neural networks can match or beat frame-based deep networks on large-scale event-based action recognition, while keeping the low latency and sparse computation that event cameras promise. It introduces a high-rate two-stream spiking architecture and evaluates it on the THU EACT-50 dataset, reporting 94.9% Top-1 accuracy at two seconds of observation, about two percentage points above the previous 3D CNN state of the art, with only a fifth of the parameters and under one percent of the synaptic operations. The paper also formalizes an evaluation framework that reports accuracy at growing observation times, and shows the model can reach roughly 50% Top-5 accuracy within the first 100 ms. A sympathetic reader would take the central claim to be that high-rate spiking processing, rather than frame accumulation, is now a practical route to early and accurate action recognition.

What carries the argument

The load-bearing mechanism is the two-stream spiking backbone plus a gated temporal fusion. The architecture splits a shared convolutional spiking layer into a ventral stream (stride-2, growing channels) and a dorsal stream (alternating strides, slower-growing channels), each built from convolutional layers with either parametric leaky integrate-and-fire (PLIF) or adaptive LIF (adLIF) neurons whose decay, feedback, and gating parameters are trainable per channel. The streams are concatenated and fused by an event-based gated recurrent unit (EGRU), or a lighter event-based gated unit (EGU) that drops state-dependent gates and needs about a third of the EGRU parameters; this fusion emits sparse real-valued events. A non-spiking leaky-integrator readout with trainable per-class decay accumulates the fused events, and the network is trained with a combined loss that averages the membrane-potential cross-entropy over time (LCEM) and samples the potential at multiple steps (LTET), which together are designed to balance early approximate predictions with final accuracy.

What would settle it

Run the baseline 3D CNN under the same fixed 2.0 second partial-observation protocol and the same test split; if its Top-1 accuracy reaches or exceeds the reported 94.9%, the paper's headline superiority claim would be overturned.

Watch

Extended reading notes

Core claim

The central claim is that a fully trainable two-stream spiking neural network—a shared spiking convolutional layer feeding separate ventral and dorsal streams, fused by an event-based gated recurrent unit and read out by leaky-integrator neurons—achieves state-of-the-art accuracy on THU EACT-50 while operating at high temporal rate. The reported 94.9% Top-1 accuracy at a fixed 2.0 s observation window outperforms the EV-ACT 3D CNN (92.72% on complete samples) by more than two points, while requiring about 20% of the parameters and under 1% of the synaptic operations. The paper further claims that combining a mean-potential cross-entropy loss with a temporally sampled TET loss, and reading out through leaky-integrator neurons, yields both strong early recognition and strong final accuracy.

Load-bearing premise

The headline accuracy claim assumes that the 3D CNN baseline's accuracy, measured on complete action samples of 2 to 5 seconds, can be fairly compared with the proposed model's accuracy measured after exactly 2.0 seconds of observation.

Editorial extensions

If this is right

  • If the reported numbers hold, SNNs become a credible alternative to frame-accumulating 3D CNNs for event-camera action recognition, with about a fifth of the parameters and over two orders of magnitude fewer effective synaptic operations.
  • The formalized evaluation scheme (accuracy at fixed observation times, Top-K scores, effective SynOps) gives the field a common yardstick for early event-based recognition, replacing single final-accuracy comparisons.
  • The 50% Top-5 accuracy within 100 ms suggests that coarse action hypotheses can be formed within the first few frames' worth of events, which is directly useful for always-on monitoring and triggering systems.
  • The reported transfer to tennis-serve triggering in a motion-capture setting indicates the architecture can be applied to real-world streaming tasks without retraining on dense frames.
  • Because the model is designed for event-by-event-like processing with small time bins, it is positioned for asynchronous neuromorphic deployment, where sparse spikes directly map to hardware operations.

Reading between the lines

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

  • A natural next check is to re-evaluate the EV-ACT 3D CNN under the same fixed 2.0 s observation window; the paper compares its number against a full-sequence baseline, so the true gap under matched conditions remains to be measured.
  • The high-rate processing idea could be pushed further by testing bin durations below 2 ms with hardware-in-the-loop, since the paper finds 1 ms bins degrade accuracy; the optimum may depend on event density and class dynamics.
  • The EGU's removal of state-dependent gates suggests a path toward even cheaper fusion, but the paper's own results show EGRU still wins on accuracy, so the trade-off between parameter savings and recognition quality is not yet settled.
  • The early Top-5 capability is promising for interactive systems, but the paper does not report Top-5 accuracy for the 3D CNN baseline; a matched comparison would show whether this early-hypothesis advantage is unique to SNNs or shared by frame-based approaches.
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 / 5 minor

Summary. The paper proposes EEvAct, a two-stream spiking neural network for early event-based action recognition, and a formalized evaluation framework that reports accuracy at growing observation times together with effective synaptic operation counts. The architecture combines a common convolutional spiking layer, two parallel streams with trainable PLIF or adLIF neurons, an event-based gated fusion layer (EGRU or the proposed EGU), and a leaky-integrator readout with a combined loss. On THU EACT-50, the best variant is reported to reach 94.9% Top-1 accuracy at a 2.0 s observation, outperforming the EV-ACT 3D CNN baseline by 2.18 points while using 4.46M parameters and 0.046G effective ACs. The paper also reports about 50% Top-5 accuracy within 100 ms for its best early-recognition configuration and demonstrates the method on a tennis-serve triggering use case.

Significance. If the central comparison were sound, the paper would make a useful contribution: it provides a concrete early-recognition benchmark for event-based cameras, a practical architecture with several ablated components, and a real-world application that illustrates the value of high-rate spiking processing. I credit the authors for moving beyond final-accuracy-only reporting, for the large-scale evaluation, and for the transparent presentation of parameter counts and operation counts in Table I. However, the two headline claims—outperforming the prior SOTA by more than 2% and using less than 1% of the synaptic operations—both rest on comparisons that are not matched in evaluation condition or cost metric. The stress-test concern raised by the reader is valid: without a same-protocol baseline or a full-sample EEvAct accuracy, the SOTA claim is not established.

major comments (4)
  1. [Section V, Table I (footnote 2); Section I, Contribution 2] The headline SOTA claim is based on unmatched evaluation conditions. The 94.9% accuracy is reported for the two-stream SNN at a fixed S=2.0 s partial observation, whereas the 92.72% accuracy for EV-ACT 3D CNN [12] is explicitly reported for complete samples of 2–5 s duration. The paper does not provide either (a) EV-ACT evaluated under the same 2.0 s partial-observation protocol, or (b) EEvAct evaluated on the complete-sample protocol. The direction of the mismatch is not obvious, so without one of these matched evaluations the central 'outperforms by more than 2%' claim in the abstract, introduction, and conclusion is unverified.
  2. [Section V, Table I (efficiency columns)] The 'less than 1% of synaptic operations' claim compares 0.046 G effective ACs for the SNN against 29.0 G MACs for the 3D CNN. Effective ACs (which skip zero or spike-free operations) and MACs (dense multiply-accumulates) are not commensurable without a precise counting rule that maps both to the same operation definition. Moreover, the AC count is for a 2.0 s partial observation while the MAC count is for complete 2–5 s samples, which compounds the mismatch. Please report a matched cost comparison, e.g., both models evaluated on the same observation length and both counted as MACs, or both as effective operations under an identical counting rule.
  3. [Section IV-B, Eqs. (3)–(5); Section V-A] The paper never specifies the surrogate gradient used to train through the Heaviside spike function in Eq. (5), nor the event threshold in Eq. (7) for the EGU. This is essential for reproducibility of any SNN result. Relatedly, Table I reports single-run accuracies without standard deviations or number of seeds; given that the claimed margin over the baseline is just over 2 percentage points, seed-to-seed variation could change the conclusion. Please specify the surrogate, the training randomness, and report error bars over multiple seeds.
  4. [Section V-A (training and evaluation protocol)] There is a potential mismatch between training and evaluation data sampling: models are trained on randomly cropped 1000 ms windows taken from anywhere in the 2–5 s samples, whereas early-recognition evaluation starts from the beginning of each recording. The paper should clarify whether these training crops are aligned to action onsets or include arbitrary phases of the action. If they are not aligned, the early-recognition scores could be systematically affected, and the authors should discuss how the model transfers from randomly phased crops to action-onset evaluation.
minor comments (5)
  1. [Section V-C, Figure 7] The sentence 'The best early recognition model allows for 50% Top-5 accuracy already within 100ms' does not identify which configuration is plotted; please name the exact model/loss/readout combination (e.g., the TET/Last-membrane variant in Table I).
  2. [Equation (9)] The symbol L_TET is typeset with inconsistent spacing and the acronym TET is not expanded; please clarify the notation and define TET when first used.
  3. [Table I caption] The terms SynOps, MACs, and ACs are not defined in the caption; please provide a sentence defining effective accumulations and how they are counted.
  4. [Abstract and Section I] The phrase 'outperforming previous work by 2%' should name the exact baseline and evaluation condition, since the percentage is not interpretable without the protocol.
  5. [Section V-D] The SpikingBody dataset description gives only total train/test counts; for reproducibility, please provide per-class sample counts and describe the recording protocol (e.g., camera placement, action segmentation).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the accuracy and efficiency claims are empirical measurements against external baselines; the 2% margin rests on an unmatched evaluation condition, but that is a comparability flaw rather than a circular derivation.

full rationale

The paper does not derive its headline results from its own definitions or fitted constants. The 94.9% Top-1 accuracy on THU EACT-50 is an empirical evaluation result obtained by training the proposed two-stream SNN and measuring its accuracy on the dataset's test split; it is not a parameter fitted to the test data and then relabeled as a prediction. The architectural choices (PLIF/adLIF neurons, EGRU/EGU fusion, loss/readout variants, time-bin size) are design choices evaluated by ablation, not inputs that force the reported outcome by construction. The cited neuron models and gated units (PLIF, adLIF, EGRU, MinGRU) are external prior work with independent support; they are not self-citations carrying the argument. The closest issue is Table I footnote 2, which states that the EV-ACT 3D CNN baseline accuracy is reported for complete 2–5 s samples while EEvAct is evaluated at S=2.0 s; this makes the claimed >2% improvement over the baseline not strictly matched. However, a mismatched evaluation protocol is a correctness/validation concern, not a circularity: the paper's own result is still an externally measured quantity, and no equation reduces the claimed margin to the framework's definitions. There is no self-citation chain, no uniqueness theorem imported from the authors' prior work, and no renaming of a known empirical pattern as a derivation. Accordingly, no specific circular step can be quoted and exhibited under the required standard.

Assumptions & free parameters 2 free parameters · 3 assumptions · 1 invented entities

The central empirical claims rest on standard deep learning assumptions (optimization, generalization) and on the dataset's correctness, plus a few hand-chosen design parameters (2ms time bin, 1000ms training window). No physical entities are invented beyond the EGU architecture component, which is tested only indirectly.

free parameters (2)
  • Time bin size = 2 ms
    Selected as the best among 1, 2, 5, 10, 20 ms based on early recognition accuracy over observation time (Section V-B). This value controls the temporal resolution and the reported SynOps counts.
  • Training window length = 1000 ms
    Random crops of 1000 ms are used for training (Section V-A); the evaluation uses varying observation times up to 2s, so the model is trained on shorter windows than the longest evaluation window.
assumptions (3)
  • domain assumption The surrogate gradient method used for backpropagation through the non-differentiable spike function is not described.
    The spike emission in Eq. 5 uses a Heaviside step function, so training must rely on a surrogate gradient; its choice affects training stability and final accuracy, but no surrogate is identified in the paper.
  • domain assumption THU EACT-50 labels and the official train/test split are correct.
    The central empirical claim depends on the dataset being correctly labeled and the standard split being used, as stated in Section V-A.
  • domain assumption Event binning by summing events per polarity in 2ms bins preserves the information needed for early recognition.
    Events within a bin are summed, losing their exact sub-bin timing; the paper argues this is sufficient for 'event-by-event like' processing, but this is an assumption about the task.
invented entities (1)
  • Event-based Gated Unit (EGU)
    purpose: Lightweight fusion of ventral and dorsal stream features, replacing the heavier EGRU while maintaining gating behavior
    Introduced in Section IV-C; it is a new architecture component evaluated only inside this paper, not on external tasks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of EEvAct: Early Event-Based Action Recognition with High-Rate Two-Stream Spiking Neural Networks." pith.science (2026). https://pith.science/paper/R5V7JWFA

@misc{pith2026250707734,
  author       = {Pith},
  title        = {Pith review of: EEvAct: Early Event-Based Action Recognition with High-Rate Two-Stream Spiking Neural Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R5V7JWFA}},
  note         = {Machine review of arXiv:2507.07734}
}
read the original abstract

Recognizing human activities early is crucial for the safety and responsiveness of human-robot and human-machine interfaces. Due to their high temporal resolution and low latency, event-based vision sensors are a perfect match for this early recognition demand. However, most existing processing approaches accumulate events to low-rate frames or space-time voxels which limits the early prediction capabilities. In contrast, spiking neural networks (SNNs) can process the events at a high-rate for early predictions, but most works still fall short on final accuracy. In this work, we introduce a high-rate two-stream SNN which closes this gap by outperforming previous work by 2% in final accuracy on the large-scale THU EACT-50 dataset. We benchmark the SNNs within a novel early event-based recognition framework by reporting Top-1 and Top-5 recognition scores for growing observation time. Finally, we exemplify the impact of these methods on a real-world task of early action triggering for human motion capture in sports.

Figures

Figures reproduced from arXiv: 2507.07734 by the authors.

Figure 1
Figure 1. Event stream processing: Event-by-event, high-rate [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Spiking Two-Stream Net: a common spiking convolution layer, parallel ventral and dorsal streams, an event-based gated [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Early recognition plots for PLIF+EGRU Two-Stream [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Early recognition plots for PLIF+EGRU Two-Stream [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 7
Figure 7. Figure 7: Top-5 early recognition plots for Two-Stream SNNs [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Block diagram of our neuromorphic trigger system [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Early recognition plots for evaluation with Spiking [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

49 extracted references · 37 canonical work pages

  1. [12]

    Action Recognition and Benchmark Using Event Cam- eras,

    Y . Gao et al., “Action Recognition and Benchmark Using Event Cam- eras,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, pp. 14 081–14 097, 2023

    work page 2023

  2. [1]

    Scaling Egocentric Vision: The EPIC-KITCHENS Dataset,

    D. Damen et al. , “Scaling Egocentric Vision: The EPIC-KITCHENS Dataset,” ArXiv, vol. abs/1804.02748, 2018

    arXiv 2018

  3. [2]

    Online human action detection and anticipation in videos: A survey,

    X. Hu et al., “Online human action detection and anticipation in videos: A survey,” Neurocomputing, vol. 491, pp. 395–413, 2022

    work page 2022

  4. [3]

    Pedestrian Action Antici- pation using Contextual Feature Fusion in Stacked RNNs,

    A. Rasouli, I. Kotseruba, and J. K. Tsotsos, “Pedestrian Action Antici- pation using Contextual Feature Fusion in Stacked RNNs,” ArXiv, vol. abs/2005.06582, 2020

    arXiv 2005

  5. [4]

    Event-based Vision for Early Prediction of Manipulation Actions

    D. Deniz, C. Fermuller, E. Ros, M. Rodr ´ıguez- ´Alvarez, and F. Barranco, “Event-based Vision for Early Prediction of Manipulation Actions,” ArXiv, vol. abs/2307.14332, 2023

    work page Pith review arXiv 2023

  6. [5]

    An overview of Human Action Recognition in sports based on Computer Vision,

    K. Host and M. Ivasic-Kos, “An overview of Human Action Recognition in sports based on Computer Vision,” Heliyon, vol. 8, 2022

    work page 2022

  7. [6]

    IMU-based Human Activity Recognition using Machine Learning and Deep Learning models,

    S. AlKharji, A. Alteneiji, and K. Poon, “IMU-based Human Activity Recognition using Machine Learning and Deep Learning models,” ICSPIS, pp. 62–66, 2023

    work page 2023

  8. [7]

    Lightweight Deep Learning Model in Mobile-Edge Computing for Radar-Based Human Activity Recognition,

    J. Zhu, X. Lou, and W. Ye, “Lightweight Deep Learning Model in Mobile-Edge Computing for Radar-Based Human Activity Recognition,” IEEE Internet of Things Journal , vol. 8, pp. 12 350–12 359, 2021

    work page 2021

Show all 49 references
  1. [8]

    UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild,

    K. Soomro, A. Zamir, and M. Shah, “UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild,” ArXiv, vol. abs/1212.0402, 2012

    2012 arXiv

  2. [9]

    A Survey on Deep Learning Techniques for Action Anticipation,

    Z. Zhong, M. Martin, M. V oit, J. Gall, and J. Beyerer, “A Survey on Deep Learning Techniques for Action Anticipation,” ArXiv, vol. abs/2309.17257, 2023

    2023 arXiv

  3. [10]

    Streaming egocentric action anticipa- tion: An evaluation scheme and approach,

    A. Furnari and G. M. Farinella, “Streaming egocentric action anticipa- tion: An evaluation scheme and approach,” Comput. Vis. Image Underst., vol. 234, p. 103763, 2023

    2023

  4. [11]

    Event-Based Vision: A Survey,

    G. Gallego et al., “Event-Based Vision: A Survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, pp. 154–180, 2019

    2019

  5. [13]

    Temporal Binary Representation for Event-Based Action Recognition,

    S. U. Innocenti, F. Becattini, F. Pernici, and A. Bimbo, “Temporal Binary Representation for Event-Based Action Recognition,” ICPR, pp. 10 426– 10 432, 2020

    2020

  6. [14]

    Scalable Event-by-Event Processing of Neuromorphic Sensory Signals with Deep State-Space Models,

    M. Sch ¨one et al., “Scalable Event-by-Event Processing of Neuromorphic Sensory Signals with Deep State-Space Models,” ICONS, pp. 124–131, 2024

    2024

  7. [15]

    Networks of spiking neurons: The third generation of neural network models,

    W. Maass, “Networks of spiking neurons: The third generation of neural network models,” Neural Networks, vol. 10, no. 9, pp. 1659–1671, 1997

    1997

  8. [16]

    Efficient Neuromorphic Signal Processing with Loihi 2,

    G. Orchard et al., “Efficient Neuromorphic Signal Processing with Loihi 2,” SiPS, pp. 254–259, 2021

    2021

  9. [17]

    HMDB: A large video database for human motion recognition,

    H. Kuehne, H. Jhuang, E. Garrote, T. A. Poggio, and T. Serre, “HMDB: A large video database for human motion recognition,” ICCV, pp. 2556– 2563, 2011

    2011

  10. [18]

    Long-Term Recurrent Convolutional Networks for Visual Recognition and Description,

    J. Donahue et al. , “Long-Term Recurrent Convolutional Networks for Visual Recognition and Description,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 39, pp. 677–691, 2017

    2017

  11. [19]

    Learning Spatiotemporal Features with 3D Convolutional Networks,

    D. Tran et al., “Learning Spatiotemporal Features with 3D Convolutional Networks,” ICCV, pp. 4489–4497, 2015

    2015

  12. [20]

    Disentangling and Unifying Graph Convolutions for Skeleton-Based Action Recognition,

    K. Z. Liu et al. , “Disentangling and Unifying Graph Convolutions for Skeleton-Based Action Recognition,” CVPR, pp. 140–149, 2020

    2020

  13. [21]

    Two-Stream Convolutional Networks for Action Recognition in Videos,

    K. Simonyan and A. Zisserman, “Two-Stream Convolutional Networks for Action Recognition in Videos,” ArXiv, vol. abs/1406.2199, 2014

    2014 arXiv

  14. [22]

    Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset,

    J. Carreira and A. Zisserman, “Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset,” CVPR, pp. 4724–4733, 2017

    2017

  15. [23]

    SlowFast Networks for Video Recognition,

    C. Feichtenhofer, H. Fan, J. Malik, and K. He, “SlowFast Networks for Video Recognition,” ICCV, pp. 6201–6210, 2019

    2019

  16. [24]

    Enhancing Video Trans- formers for Action Understanding with VLM-aided Training,

    H. Lu, H. Jian, R. Poppe, and A. A. Salah, “Enhancing Video Trans- formers for Action Understanding with VLM-aided Training,” ArXiv, vol. abs/2403.16128, 2024

    2024 arXiv

  17. [25]

    MANTA: Diffusion Mamba for Efficient and Effective Stochastic Long- Term Dense Anticipation,

    O. Zatsarynna, E. Bahrami, Y . A. Farha, G. Francesca, and J. Gall, “MANTA: Diffusion Mamba for Efficient and Effective Stochastic Long- Term Dense Anticipation,” ArXiv, vol. abs/2501.08837, 2025

    2025 arXiv

  18. [26]

    Early event detection based on dynamic images of surveillance videos,

    Y . Fan, G. Wen, D. Li, S. Qiu, and M. D. Levine, “Early event detection based on dynamic images of surveillance videos,”J. Vis. Commun. Image Represent., vol. 51, pp. 70–75, 2018

    2018

  19. [27]

    A Low Power, Fully Event-Based Gesture Recognition System,

    A. Amir et al., “A Low Power, Fully Event-Based Gesture Recognition System,” CVPR, pp. 7388–7397, 2017

    2017

  20. [28]

    Neuromorphic Vision Datasets for Pedestrian Detection, Action Recognition, and Fall Detection,

    S. Miao et al., “Neuromorphic Vision Datasets for Pedestrian Detection, Action Recognition, and Fall Detection,” Frontiers in Neurorobotics , vol. 13, 2019

    2019

  21. [29]

    Event-based Action Recognition Using Motion Information and Spiking Neural Networks,

    Q. Liu, D. Xing, H. Tang, D. Ma, and G. Pan, “Event-based Action Recognition Using Motion Information and Spiking Neural Networks,” in International Joint Conference on Artificial Intelligence , 2021

    2021

  22. [30]

    Graph-Based Spatio-Temporal Feature Learning for Neu- romorphic Vision Sensing,

    Y . Bi et al., “Graph-Based Spatio-Temporal Feature Learning for Neu- romorphic Vision Sensing,” IEEE Transactions on Image Processing , vol. 29, pp. 9084–9098, 2020

    2020

  23. [31]

    Eventtransact: A video transformer-based framework for event-camera based action recognition,

    T. de Blegiers, I. R. Dave, A. Yousaf, and M. Shah, “Eventtransact: A video transformer-based framework for event-camera based action recognition,” IROS, pp. 1–7, 2023

    2023

  24. [32]

    Space-Time Event Clouds for Gesture Recognition: From RGB Cameras to Event Cameras,

    Q. Wang, Y . Zhang, J. Yuan, and Y . Lu, “Space-Time Event Clouds for Gesture Recognition: From RGB Cameras to Event Cameras,” WACV, pp. 1826–1835, 2019

    2019

  25. [33]

    Training Spiking Neural Networks Using Lessons From Deep Learning,

    J. K. Eshraghian et al. , “Training Spiking Neural Networks Using Lessons From Deep Learning,” Proceedings of the IEEE , vol. 111, pp. 1016–1054, 2021

    2021

  26. [34]

    Spiking Neural Networks for event-based action recognition: A new task to understand their advantage,

    A. Vicente-Sola et al., “Spiking Neural Networks for event-based action recognition: A new task to understand their advantage,” Neurocomput- ing, vol. 611, p. 128657, 2022

    2022

  27. [35]

    SpikePoint: An Efficient Point-based Spiking Neu- ral Network for Event Cameras Action Recognition,

    H. Ren et al. , “SpikePoint: An Efficient Point-based Spiking Neu- ral Network for Event Cameras Action Recognition,” ArXiv, vol. abs/2310.07189, 2023

    2023 arXiv

  28. [36]

    Temporal-Guided Spiking Neural Networks for Event- Based Human Action Recognition,

    S. Yang et al., “Temporal-Guided Spiking Neural Networks for Event- Based Human Action Recognition,” ArXiv, vol. abs/2503.17132, 2025

    2025 arXiv

  29. [37]

    Spike-HAR++: an energy-efficient and lightweight parallel spiking transformer for event-based human action recognition,

    X. Lin, M. Liu, and H. Chen, “Spike-HAR++: an energy-efficient and lightweight parallel spiking transformer for event-based human action recognition,” Frontiers in Computational Neuroscience , vol. 18, 2024

    2024

  30. [38]

    Two-Stream Spiking Neural Network for Event-based Action Recognition,

    S. J. Lian et al., “Two-Stream Spiking Neural Network for Event-based Action Recognition,” ICASSP, 2025

    2025

  31. [39]

    Incorporating Learnable Membrane Time Constant to Enhance Learning of Spiking Neural Networks,

    W. Fang et al. , “Incorporating Learnable Membrane Time Constant to Enhance Learning of Spiking Neural Networks,” ICCV, pp. 2641–2651, 2021

    2021

  32. [40]

    Neuronal Dy- namics: From Single Neurons To Networks And Models Of Cognition,

    W. Gerstner, W. M. Kistler, R. Naud, and L. Paninski, “Neuronal Dy- namics: From Single Neurons To Networks And Models Of Cognition,” 2014

    2014

  33. [41]

    Advanc- ing Spatio-Temporal Processing in Spiking Neural Networks through Adaptation,

    M. Baronig, R. Ferrand, S. Sabathiel, and R. A. Legenstein, “Advanc- ing Spatio-Temporal Processing in Spiking Neural Networks through Adaptation,” ArXiv, vol. abs/2408.07517, 2024

    2024 arXiv

  34. [42]

    Neuromorphic Lip-Reading with Signed Spiking Gated Recurrent Units,

    M. Dampfhoffer and T. Mesquida, “Neuromorphic Lip-Reading with Signed Spiking Gated Recurrent Units,” CVPRW, pp. 2141–2151, 2024

    2024

  35. [43]

    EGRU: Event-based GRU for activity-sparse inference and learning,

    A. Subramoney, K. K. Nazeer, M. Sch ¨one, C. Mayr, and D. Kappel, “EGRU: Event-based GRU for activity-sparse inference and learning,” ArXiv, vol. abs/2206.06178, 2022

    2022 arXiv

  36. [44]

    Temporal Contrastive Learning for Spiking Neural Networks,

    H. Qiu et al. , “Temporal Contrastive Learning for Spiking Neural Networks,” ArXiv, vol. abs/2305.13909, 2023

    2023 arXiv

  37. [45]

    ED-sKWS: Early- Decision Spiking Neural Networks for Rapid,and Energy-Efficient Key- word Spotting,

    Z. Song, Q. Liu, Q. Yang, Y . Peng, and H. Li, “ED-sKWS: Early- Decision Spiking Neural Networks for Rapid,and Energy-Efficient Key- word Spotting,” ArXiv, vol. abs/2406.12726, 2024

    2024 arXiv

  38. [46]

    NeuroBench: Advancing Neuromorphic Computing through Collaborative, Fair and Representa- tive Benchmarking,

    Y . Demira˘g, M. A. Petrovici, and J. Timcheck, “NeuroBench: Advancing Neuromorphic Computing through Collaborative, Fair and Representa- tive Benchmarking,” ArXiv, vol. abs/2304.04640, 2023

    2023 arXiv

  39. [47]

    Separate visual pathways for perception and action,

    M. A. Goodale and A. D. Milner, “Separate visual pathways for perception and action,” Trends in Neurosciences , vol. 15, pp. 20–25, 1992

    1992

  40. [48]

    Were RNNs All We Needed?

    L. Feng, F. Tung, M. O. Ahmed, Y . Bengio, and H. Hajimirsadeghi, “Were RNNs All We Needed?” ArXiv, vol. abs/2410.01201, 2024

    2024 arXiv

  41. [49]

    Comprehensive Survey on Human Motion Tracking System,

    Juveriyah, Himanshu, and M. J. Moharkan, “Comprehensive Survey on Human Motion Tracking System,” ICACRS, pp. 1380–1387, 2023

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.