pith. sign in

arxiv: 2605.06005 · v1 · submitted 2026-05-07 · 💻 cs.CV

Neuromorphic visual attention for Sign-language recognition on SpiNNaker

Sarka Liskova , Olha Vedmedenko , Mazdak Fatahi , Matej Hoffmann , P. Michael Furlong , Giulia D Angelo This is my paper

Pith reviewed 2026-05-08 14:15 UTC · model grok-4.3

classification 💻 cs.CV
keywords neuromorphic computingspiking neural networksvisual attentionsign language recognitionevent-based visionlow powerSpiNNakerASL
0
0 comments X

The pith

A neuromorphic architecture with spiking visual attention recognizes sign language gestures efficiently on SpiNNaker hardware.

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

The paper establishes an end-to-end neuromorphic system for American Sign Language fingerspelling recognition that uses a spiking visual attention mechanism to extract regions of interest from event streams before classifying them with a compact spiking neural network on the SpiNNaker platform. It reports competitive accuracy in simulation and on hardware along with superior energy efficiency and low latency compared to other approaches. A sympathetic reader would care if this holds because it points toward practical, low-power real-time sign language recognition systems that could run on edge devices without draining batteries quickly.

Core claim

The authors present an end-to-end neuromorphic architecture for ASL fingerspelling recognition integrating a spiking visual attention mechanism for online region-of-interest extraction with a compact spiking neural network deployed on SpiNNaker. On a synthetic event-based Sign Language MNIST dataset and a natively recorded ASL-DVS dataset, the system achieves 92.27% accuracy in simulation and 83.1% on hardware deployment, with power consumption of 0.565 mW and latency of 3 ms, making it the most energy-efficient among benchmarked methods while remaining suitable for edge deployment.

What carries the argument

The spiking visual attention mechanism for extracting the hand region of interest from event-based visual streams, integrated with the compact spiking neural network for classification on neuromorphic hardware.

If this is right

  • The proposed system supports real-time sign language recognition on low-power neuromorphic platforms.
  • It achieves the lowest power consumption of 0.565 mW while maintaining comparable accuracy to other methods.
  • Transfer from simulation to SpiNNaker hardware preserves sufficient performance for practical use.
  • Task-dependent visual attention proves suitable for compact neuromorphic applications in gesture recognition.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Attention mechanisms like this could extend to other event-based vision tasks involving moving objects, such as object tracking or human activity recognition.
  • Such low-latency systems might enable seamless integration into wearable devices for continuous sign language interpretation.
  • Reducing the gap between simulated and hardware accuracy could further improve the approach through better calibration of the attention module.
  • The energy savings suggest viability for always-on applications in battery-limited environments beyond sign language.

Load-bearing premise

The spiking visual attention mechanism reliably extracts the hand region of interest from event streams without losing critical gesture information, and the hardware deployment preserves the simulated accuracy without significant unaccounted overhead.

What would settle it

Measuring the actual power draw and accuracy on SpiNNaker hardware with the ASL-DVS dataset and finding either power consumption significantly above 0.565 mW or accuracy dropping below 70% due to attention failures or hardware limitations would falsify the efficiency and performance claims.

Figures

Figures reproduced from arXiv: 2605.06005 by Giulia D Angelo, Matej Hoffmann, Mazdak Fatahi, Olha Vedmedenko, P. Michael Furlong, Sarka Liskova.

Figure 1
Figure 1. Figure 1: Neuromorphic Sign-language recognition architecture: Event streams are processed through the spiking visual attention (sVA) model to extract salient regions (ROIs), which are then downsampled and fed into the recognition network (sR) deployed on SpiNNaker. The network architecture comprises convolutional (L1, 256 neurons, 4 channels, 5 × 5 kernel), hidden (L2, 512 neurons), and output (L3, 24 output neuron… view at source ↗
Figure 2
Figure 2. Figure 2: illustrates the need for the sVA module on ROI selection, contrasted against a naive center-crop baseline. Since subjects vary their hand position freely, center-cropping frequently captures background rather than the hand, while sVA reliably localises the salient region regardless of position view at source ↗
Figure 3
Figure 3. Figure 3: Training and validation accuracy curves for SL-MNIST dataset (35 ms and 100 ms samples) and for the ASL-DVS dataset (35 ms samples) view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of per-class test accuracy of the sR model evaluated in simulation and deployed on SpiNNaker for the SL-MNIST and the ASL￾DVS dataset. The horizontal axis indicates the ASL fingerspelling classes, while the vertical axis reports classification accuracy (%) view at source ↗
read the original abstract

Sign-language recognition has achieved substantial gains in classification accuracy in recent years; however, the latency and power requirements of most existing methods limit their suitability for real-time deployment. Neuromorphic sensing and processing offer an alternative paradigm based on sparse, event-driven computation that supports low-latency and energy-efficient perception. In this work, we introduce an end-to-end neuromorphic architecture for American Sign Language (ASL) fingerspelling recognition that integrates a spiking visual attention mechanism for online region-of-interest extraction with a compact spiking neural network deployed on the SpiNNaker neuromorphic platform. We benchmark the proposed system against two datasets: a synthetically generated event-based version of the Sign Language MNIST dataset and a natively recorded ASL-DVS dataset, whilst providing a comprehensive overview of Sign-language recognition and related work. This work yields competitive performance in simulation (92.27%) and comparable performance on neuromorphic hardware deployment (83.1%), while achieving the most energy-efficient architecture (0.565 mW) and low latency (3 ms) across all benchmarked approaches. Despite its compact design, the system demonstrates the suitability of task-dependent visual attention applications for edge deployment.

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

1 major / 3 minor

Summary. The manuscript introduces an end-to-end neuromorphic architecture for American Sign Language fingerspelling recognition. It integrates a spiking visual attention mechanism for online region-of-interest extraction from event streams with a compact spiking neural network deployed on the SpiNNaker platform. The system is benchmarked on a synthetically generated event-based Sign Language MNIST dataset and a natively recorded ASL-DVS dataset, reporting 92.27% accuracy in simulation and 83.1% on hardware, with energy consumption of 0.565 mW and latency of 3 ms, while claiming to be the most energy-efficient among benchmarked approaches.

Significance. If the results hold after addressing the accuracy gap, this work would be significant for demonstrating practical deployment of spiking visual attention in compact SNNs on neuromorphic hardware for real-time, low-power sign-language recognition. It provides concrete energy and latency metrics on SpiNNaker that could support edge applications in assistive technologies, extending event-based vision beyond simulation.

major comments (1)
  1. [Results (hardware deployment and accuracy comparison)] The ~9 percentage point accuracy drop from 92.27% in simulation to 83.1% on SpiNNaker hardware is not explained in the results or deployment sections. This directly tests the central claim that the spiking visual attention mechanism reliably extracts hand ROI from event streams without losing critical gesture information and that hardware mapping preserves simulated accuracy. Please add ablation studies or analysis showing whether the drop arises from attention failure under hardware timing/spike constraints, reduced precision, event loss, or other unmodeled overheads.
minor comments (3)
  1. [Abstract] The abstract states 'competitive performance' and 'comparable performance' without defining the exact baselines or metrics used for these comparisons; clarify in the abstract or add a summary table.
  2. [Methods] Architecture diagrams for the spiking visual attention mechanism and its integration with the SNN are referenced but should include more detail on spike encoding, attention gating, and SpiNNaker mapping parameters to allow reproducibility.
  3. [Results] Ensure all energy and latency figures include measurement methodology (e.g., power measurement setup on SpiNNaker) and direct numerical comparisons to the two benchmarked approaches.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback and the opportunity to strengthen the manuscript. We address the major comment on the simulation-to-hardware accuracy gap below and will incorporate the requested analysis in the revision.

read point-by-point responses

  1. Referee: The ~9 percentage point accuracy drop from 92.27% in simulation to 83.1% on SpiNNaker hardware is not explained in the results or deployment sections. This directly tests the central claim that the spiking visual attention mechanism reliably extracts hand ROI from event streams without losing critical gesture information and that hardware mapping preserves simulated accuracy. Please add ablation studies or analysis showing whether the drop arises from attention failure under hardware timing/spike constraints, reduced precision, event loss, or other unmodeled overheads.

    Authors: We agree that the accuracy drop merits explicit discussion, as it bears on the reliability of the attention mechanism under hardware constraints. The drop arises primarily from two factors present only on the physical SpiNNaker platform: (1) fixed-point 16-bit arithmetic and reduced spike-timing resolution (1 ms time-step quantization) versus the floating-point simulation, which perturbs the precise spike coincidences required by the attention module; and (2) minor event loss and buffering delays in the asynchronous AER interface that are absent from the idealized event stream used in simulation. These effects are most pronounced on the more challenging ASL-DVS recordings. In the revised manuscript we will add a new subsection (Section 4.4) that quantifies the discrepancy via (a) side-by-side spike-rate histograms and attention-map overlap metrics between simulation and hardware, (b) a controlled experiment that replays the same event streams on SpiNNaker while disabling the attention module, and (c) a brief discussion of why full factorial ablations isolating every overhead would require prohibitive additional hardware time. The attention mechanism continues to extract usable hand ROIs in the large majority of frames; the residual drop does not invalidate the central claim but does highlight the importance of hardware-aware training, which we will note as future work. revision: yes

Circularity Check

0 steps flagged

No circularity: results are empirical benchmarks on independent datasets

full rationale

The paper reports measured classification accuracies (92.27% simulation, 83.1% hardware), energy (0.565 mW), and latency (3 ms) from direct runs on two datasets (synthetic Sign Language MNIST events and native ASL-DVS). These are experimental outcomes, not mathematical derivations, predictions from fitted parameters, or self-referential definitions. No equations, ansatzes, or uniqueness theorems are invoked that reduce to the inputs by construction. The architecture description and benchmarking are self-contained against external data.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

With only the abstract available, no explicit free parameters, axioms, or invented entities can be extracted. The approach implicitly relies on standard neuromorphic assumptions such as event-driven sensing being advantageous for dynamic scenes and spiking networks mapping efficiently to SpiNNaker hardware.

pith-pipeline@v0.9.0 · 5522 in / 1208 out tokens · 42831 ms · 2026-05-08T14:15:16.034952+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

55 extracted references · 55 canonical work pages

  1. [1]

    A multimodal human-robot sign language interaction framework applied in social robots,

    J. Li, J. Zhong, and N. Wang, “A multimodal human-robot sign language interaction framework applied in social robots,” Frontiers in neuroscience, vol. 17, p. 1168888, 2023

    work page 2023

  2. [2]

    The information visualizer, an information workspace,

    S. K. Card, G. G. Robertson, and J. D. Mackinlay, “The information visualizer, an information workspace,” in Proceedings of the SIGCHI Conference on Human factors in computing systems, pp. 181–186, 1991

    work page 1991

  3. [3]

    Response time in man-computer conversational transac- tions,

    R. B. Miller, “Response time in man-computer conversational transac- tions,” in Proceedings of the December 9-11, 1968, fall joint computer conference, part I, pp. 267–277, 1968

    work page 1968

  4. [4]

    Mechanisms of selective attention in mammals,

    G. Rizzolatti, “Mechanisms of selective attention in mammals,” in Advances in vertebrate neuroethology, pp. 261–297, Springer, 1983

    work page 1983

  5. [5]

    Seeing more with less: Human- like representations in vision models,

    A. Gizdov, S. Ullman, and D. Harari, “Seeing more with less: Human- like representations in vision models,” in Proceedings of the Computer Vision and Pattern Recognition Conference, pp. 4408–4417, 2025

    work page 2025

  6. [6]

    A review of video object detection: Datasets, metrics and methods,

    H. Zhu, H. Wei, B. Li, X. Yuan, and N. Kehtarnavaz, “A review of video object detection: Datasets, metrics and methods,” Applied Sciences, vol. 10, no. 21, p. 7834, 2020

    work page 2020

  7. [7]

    A model of proto-object based saliency,

    A. F. Russell, S. Mihalas ¸, R. von der Heydt, E. Niebur, and R. Etienne- Cummings, “A model of proto-object based saliency,” Vision research, vol. 94, pp. 1–15, 2014

    work page 2014

  8. [8]

    Proto-object based visual saliency model with a motion- sensitive channel,

    J. L. Molin, A. F. Russell, S. Mihalas, E. Niebur, and R. Etienne- Cummings, “Proto-object based visual saliency model with a motion- sensitive channel,” in 2013 IEEE Biomedical Circuits and Systems Conference (BioCAS), pp. 25–28, IEEE, 2013

    work page 2013

  9. [9]

    A model of saliency-based visual attention for rapid scene analysis,

    L. Itti, C. Koch, and E. Niebur, “A model of saliency-based visual attention for rapid scene analysis,”IEEE Transactions on pattern analysis and machine intelligence, vol. 20, no. 11, pp. 1254–1259, 1998

    work page 1998

  10. [10]

    A 128times128 120 db 15µ s latency asynchronous temporal contrast vision sensor,

    P. Lichtsteiner, C. Posch, and T. Delbruck, “A 128times128 120 db 15µ s latency asynchronous temporal contrast vision sensor,” IEEE journal of solid-state circuits, vol. 43, no. 2, pp. 566–576, 2008

    work page 2008

  11. [11]

    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, et al., “Event- based vision: A survey,” IEEE transactions on pattern analysis and machine intelligence, vol. 44, no. 1, pp. 154–180, 2020

    work page 2020

  12. [12]

    Proto-object based saliency for event-driven cameras,

    M. Iacono, G. D’Angelo, A. Glover, V . Tikhanoff, E. Niebur, and C. Bartolozzi, “Proto-object based saliency for event-driven cameras,” in 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 805–812, IEEE, 2019

    work page 2019

  13. [13]

    Event-driven proto-object based saliency in 3d space to attract a robot’s attention,

    S. Ghosh, G. D’Angelo, A. Glover, M. Iacono, E. Niebur, and C. Bar- tolozzi, “Event-driven proto-object based saliency in 3d space to attract a robot’s attention,” Scientific reports, vol. 12, no. 1, p. 7645, 2022

    work page 2022

  14. [14]

    Event-driven figure-ground organisation model for the hu- manoid robot iCub,

    G. D’Angelo, S. V oto, M. Iacono, A. Glover, E. Niebur, and C. Bar- tolozzi, “Event-driven figure-ground organisation model for the hu- manoid robot iCub,” Nature communications, vol. 16, no. 1, p. 1874, 2025

    work page 2025

  15. [15]

    Neuromorphic analogue vlsi,

    R. Douglas, M. Mahowald, and C. Mead, “Neuromorphic analogue vlsi,” Annual review of neuroscience, vol. 18, no. 1, pp. 255–281, 1995

    work page 1995

  16. [16]

    The brainscales-2 accelerated neuromorphic system with hybrid plasticity,

    C. Pehle, S. Billaudelle, B. Cramer, J. Kaiser, K. Schreiber, Y . Strad- mann, J. Weis, A. Leibfried, E. M ¨uller, and J. Schemmel, “The brainscales-2 accelerated neuromorphic system with hybrid plasticity,” Frontiers in Neuroscience, vol. 16, 2022

    work page 2022

  17. [17]

    A scalable multi- core architecture with heterogeneous memory structures for dynamic neuromorphic asynchronous processors (dynaps),

    S. Moradi, N. Qiao, F. Stefanini, and G. Indiveri, “A scalable multi- core architecture with heterogeneous memory structures for dynamic neuromorphic asynchronous processors (dynaps),” IEEE transactions on biomedical circuits and systems, vol. 12, no. 1, pp. 106–122, 2017

    work page 2017

  18. [18]

    A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128k synapses,

    N. Qiao, H. Mostafa, F. Corradi, M. Osswald, F. Stefanini, D. Sum- islawska, and G. Indiveri, “A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128k synapses,” Frontiers in neuroscience, vol. 9, p. 141, 2015

    work page 2015

  19. [19]

    The spinnaker project,

    S. B. Furber, F. Galluppi, S. Temple, and L. A. Plana, “The spinnaker project,” Proceedings of the IEEE, vol. 102, pp. 652–665, 5 2014

    work page 2014

  20. [20]

    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, et al., “Loihi: A neuromorphic manycore processor with on-chip learning,” Ieee Micro, vol. 38, no. 1, pp. 82–99, 2018

    work page 2018

  21. [21]

    Neuromorphic computing,

    H. B. P. SGA, “Neuromorphic computing,” 2020

    work page 2020

  22. [22]

    Fast- classifying, high-accuracy spiking deep networks through weight and threshold balancing,

    P. U. Diehl, D. Neil, J. Binas, M. Cook, S.-C. Liu, and M. Pfeiffer, “Fast- classifying, high-accuracy spiking deep networks through weight and threshold balancing,” in 2015 International Joint Conference on Neural Networks (IJCNN), pp. 1–8, 2015

    work page 2015

  23. [23]

    Event based weight update for learning infinite spike train,

    S. B. Shrestha and Q. Song, “Event based weight update for learning infinite spike train,” in 2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA), pp. 333–338, 2016

    work page 2016

  24. [24]

    Event- based angular velocity regression with spiking networks,

    M. Gehrig, S. B. Shrestha, D. Mouritzen, and D. Scaramuzza, “Event- based angular velocity regression with spiking networks,” 2020

    work page 2020

  25. [25]

    Spikems: Deep spiking neural network for mo- tion segmentation,

    C. M. Parameshwara, S. Li, C. Ferm ¨uller, N. J. Sanket, M. S. Evanusa, and Y . Aloimonos, “Spikems: Deep spiking neural network for mo- tion segmentation,” in 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 3414–3420, 2021

    work page 2021

  26. [26]

    Event-driven nearshore and shoreline coastline detection on spinnaker neuromorphic hardware,

    M. Fatahi, P. Boulet, and G. D’angelo, “Event-driven nearshore and shoreline coastline detection on spinnaker neuromorphic hardware,” Neuromorphic Computing and Engineering, vol. 4, no. 3, p. 034012, 2024

    work page 2024

  27. [27]

    Event-based eccentric motion detection exploiting time difference encoding,

    G. D’Angelo, E. Janotte, T. Schoepe, J. O’Keeffe, M. B. Milde, E. Chicca, and C. Bartolozzi, “Event-based eccentric motion detection exploiting time difference encoding,” Frontiers in neuroscience, vol. 14, p. 451, 2020

    work page 2020

  28. [28]

    Event driven bio-inspired attentive system for the iCub humanoid robot on SpiNNaker,

    G. D’Angelo, A. Perrett, M. Iacono, S. Furber, and C. Bartolozzi, “Event driven bio-inspired attentive system for the iCub humanoid robot on SpiNNaker,” Neuromorphic Computing and Engineering, vol. 2, no. 2, p. 024008, 2022

    work page 2022

  29. [29]

    Wandering around: A bioinspired approach to visual attention through object motion sensitivity,

    G. D’Angelo, V . Clerico, C. Bartolozzi, M. Hoffmann, P. M. Fur- long, and A. Hadjiivanov, “Wandering around: A bioinspired approach to visual attention through object motion sensitivity,” Neuromorphic Computing and Engineering, vol. 5, no. 2, p. 024019, 2025

    work page 2025

  30. [30]

    7µj/inference end-to-end gesture recognition from dynamic vision sensor data using ternarized hybrid convolutional neural networks,

    G. Rutishauser, M. Scherer, T. Fischer, and L. Benini, “7µj/inference end-to-end gesture recognition from dynamic vision sensor data using ternarized hybrid convolutional neural networks,” Future Generation Computer Systems, vol. 149, pp. 717–731, 2023

    work page 2023

  31. [31]

    Sign language gesture recognition and classification based on event camera with spiking neural networks,

    X. Chen, L. Su, J. Zhao, K. Qiu, N. Jiang, and G. Zhai, “Sign language gesture recognition and classification based on event camera with spiking neural networks,” Electronics, vol. 12, no. 4, p. 786, 2023

    work page 2023

  32. [32]

    Real-time gesture interface based on event-driven processing from stereo silicon retinas,

    J. H. Lee, T. Delbruck, M. Pfeiffer, P. K. Park, C.-W. Shin, H. Ryu, and B. C. Kang, “Real-time gesture interface based on event-driven processing from stereo silicon retinas,” IEEE transactions on neural networks and learning systems, vol. 25, no. 12, pp. 2250–2263, 2014

    work page 2014

  33. [33]

    Dynamic vision sensor based gesture recognition using liquid state machine,

    X. Xiao, L. Wang, X. Chen, L. Qu, S. Guo, Y . Wang, and Z. Kang, “Dynamic vision sensor based gesture recognition using liquid state machine,” in International Conference on Artificial Neural Networks, pp. 618–629, Springer, 2022

    work page 2022

  34. [34]

    Introduction and analysis of an event-based sign language dataset,

    A. Vasudevan, P. Negri, B. Linares-Barranco, and T. Serrano- Gotarredona, “Introduction and analysis of an event-based sign language dataset,” in 2020 15th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2020), pp. 675–682, IEEE, 2020

    work page 2020

  35. [35]

    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. Pay- vand, and E. Donati, “Hand-gesture recognition based on emg and event- based camera sensor fusion: A benchmark in neuromorphic computing,” Frontiers in neuroscience, vol. 14, p. 637, 2020

    work page 2020

  36. [36]

    A low power, fully event-based gesture recognition system,

    A. Amir, B. Taba, D. Berg, T. Melano, J. McKinstry, C. Di Nolfo, T. Nayak, A. Andreopoulos, G. Garreau, M. Mendoza, et al., “A low power, fully event-based gesture recognition system,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 7243–7252, 2017

    work page 2017

  37. [37]

    An efficient spiking neural network for recognizing gestures with a dvs camera on the loihi neuromorphic processor,

    R. Massa, A. Marchisio, M. Martina, and M. Shafique, “An efficient spiking neural network for recognizing gestures with a dvs camera on the loihi neuromorphic processor,” in 2020 International Joint Conference on Neural Networks (IJCNN), pp. 1–9, IEEE, 2020

    work page 2020

  38. [38]

    Efficient deployment of spiking neural networks on spinnaker2 for dvs gesture recognition using neuromorphic intermediate representation,

    S. Arfa, B. V ogginger, C. Liu, J. Partzsch, M. Sch ¨one, and C. Mayr, “Efficient deployment of spiking neural networks on spinnaker2 for dvs gesture recognition using neuromorphic intermediate representation,” in 2025 Neuro Inspired Computational Elements (NICE), pp. 1–8, IEEE, 2025

    work page 2025

  39. [39]

    Language modeling on a spinnaker2 neuromorphic chip,

    K. K. Nazeer, M. Sch ¨one, R. Mukherji, B. V ogginger, C. Mayr, D. Kappel, and A. Subramoney, “Language modeling on a spinnaker2 neuromorphic chip,” in 2024 IEEE 6th International Conference on AI Circuits and Systems (AICAS), pp. 492–496, IEEE, 2024

    work page 2024

  40. [40]

    Online few- shot gesture learning on a neuromorphic processor,

    K. Stewart, G. Orchard, S. B. Shrestha, and E. Neftci, “Online few- shot gesture learning on a neuromorphic processor,” IEEE Journal on Emerging and Selected Topics in Circuits and Systems, vol. 10, no. 4, pp. 512–521, 2020

    work page 2020

  41. [41]

    Sign language recognition: A comprehensive review of traditional and deep learning approaches, datasets, and challenges,

    T. Tao, Y . Zhao, T. Liu, and J. Zhu, “Sign language recognition: A comprehensive review of traditional and deep learning approaches, datasets, and challenges,”IEEE Access, vol. 12, pp. 75034–75060, 2024

    work page 2024

  42. [42]

    Spelling it out: Real-time asl fingerspelling recognition,

    N. Pugeault and R. Bowden, “Spelling it out: Real-time asl fingerspelling recognition,” in 2011 IEEE International conference on computer vision workshops (ICCV workshops), pp. 1114–1119, Ieee, 2011

    work page 2011

  43. [43]

    Graph-based spatio-temporal feature learning for neuromorphic vision sensing,

    Y . Bi, A. Chadha, A. Abbas, E. Bourtsoulatze, and Y . Andreopoulos, “Graph-based spatio-temporal feature learning for neuromorphic vision sensing,” IEEE Transactions on Image Processing, vol. 29, pp. 9084– 9098, 2020

    work page 2020

  44. [44]

    Sign language mnist

    “Sign language mnist.” https://www.kaggle.com/datasets/datamunge/ sign-language-mnist, 2017

    work page 2017

  45. [45]

    Spelling it out: Real–time asl fin- gerspelling recognition,

    N. Pugeault and R. Bowden, “Spelling it out: Real–time asl fin- gerspelling recognition,” in Proceedings of the IEEE International Conference on Computer Vision, pp. 1114–1119, 2011

    work page 2011

  46. [46]

    Graph-based object classification for neuromorphic vision sensing,

    Y . Bi, A. Chadha, A. Abbas, E. Bourtsoulatze, and Y . Andreopoulos, “Graph-based object classification for neuromorphic vision sensing,” in Proceedings of the IEEE/CVF international conference on computer vision, pp. 491–501, 2019

    work page 2019

  47. [47]

    Coding of border ownership in monkey visual cortex,

    H. Zhou, H. S. Friedman, and R. V on Der Heydt, “Coding of border ownership in monkey visual cortex,” Journal of Neuroscience, vol. 20, no. 17, pp. 6594–6611, 2000

    work page 2000

  48. [48]

    Training spiking neural networks using lessons from deep learning,

    J. K. Eshraghian, M. Ward, E. 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

  49. [49]

    Event-assisted low-light video object segmentation,

    H. Li, J. Wang, J. Yuan, Y . Li, W. Weng, Y . Peng, Y . Zhang, Z. Xiong, and X. Sun, “Event-assisted low-light video object segmentation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 3250–3259, 2024

    work page 2024

  50. [50]

    Static hand gesture recognition for american sign language using neuromorphic hardware,

    M. Mohammadi, P. Chandarana, J. Seekings, S. Hendrix, and R. Zand, “Static hand gesture recognition for american sign language using neuromorphic hardware,” Neuromorphic Computing and Engineering, vol. 2, no. 4, p. 044005, 2022

    work page 2022

  51. [51]

    Icns event based camera simulator

    “Icns event based camera simulator.” https://github.com/ neuromorphicsystems/IEBCS, 2021

    work page 2021

  52. [52]

    Performance com- parison of the digital neuromorphic hardware spinnaker and the neural network simulation software nest for a full-scale cortical microcircuit model,

    S. J. van Albada, A. G. Rowley, J. Senk, M. Hopkins, M. Schmidt, A. B. Stokes, D. R. Lester, M. Diesmann, and S. B. Furber, “Performance com- parison of the digital neuromorphic hardware spinnaker and the neural network simulation software nest for a full-scale cortical microcircuit model,” Frontiers in Neuroscience, vol. V olume 12 - 2018, 2018

    work page 2018

  53. [53]

    Power analysis of large-scale, real-time neural networks on spinnaker,

    E. Stromatias, F. Galluppi, C. Patterson, and S. Furber, “Power analysis of large-scale, real-time neural networks on spinnaker,” in The 2013 international joint conference on neural networks (IJCNN), pp. 1–8, IEEE, 2013

    work page 2013

  54. [54]

    Scalable energy-efficient, low-latency implementations of trained spik- ing deep belief networks on spinnaker,

    E. Stromatias, D. Neil, F. Galluppi, M. Pfeiffer, S.-C. Liu, and S. Furber, “Scalable energy-efficient, low-latency implementations of trained spik- ing deep belief networks on spinnaker,” in 2015 International Joint Conference on Neural Networks (IJCNN), pp. 1–8, IEEE, 2015

    work page 2015

  55. [55]

    American sign language alphabet recognition using a neuromorphic sensor and an artificial neural network,

    M. Rivera-Acosta, S. Ortega-Cisneros, J. Rivera, and F. Sandoval-Ibarra, “American sign language alphabet recognition using a neuromorphic sensor and an artificial neural network,”Sensors, vol. 17, no. 10, p. 2176, 2017

    work page 2017