pith. sign in

arxiv: 2606.08370 · v1 · pith:P3L3B2LTnew · submitted 2026-06-06 · 📡 eess.IV · cs.CV

Programmable Silicon Retina on Pixel Processor Array

Maciej Lewandowski , Prince Philip , Alexandre Marcireau , Chetan Singh Thakur , Andr\'e van Schaik , Piotr Dudek This is my paper

Pith reviewed 2026-06-27 18:51 UTC · model grok-4.3

classification 📡 eess.IV cs.CV
keywords silicon retinapixel processor arraydynamic vision sensorsaliency predictionevent-based visionbio-inspired processingSCAMP-5intensity reconstruction
0
0 comments X

The pith

A multi-stage silicon retina model on pixel processor hardware yields events that improve saliency prediction while cutting the event rate nearly in half.

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

The paper implements a biologically inspired retina model that adds spatial filtering and gain control to the usual temporal contrast detection of dynamic vision sensors. This model runs on the SCAMP-5 pixel processor array and is also simulated on GPU. When the resulting events feed a lightweight neural network for saliency prediction, loss drops 13 percent and the event rate falls about 47 percent relative to standard event streams, though intensity reconstruction becomes less accurate. A sympathetic reader would care because the work points to a hardware route for compressing visual data into forms that downstream networks can use with less bandwidth on edge devices.

Core claim

The silicon retina model, which performs temporal contrast detection followed by spatial filtering and gain control, is realized on the SCAMP-5 pixel processor array; when its output events are supplied to an adapted FireNet-style network for video saliency prediction, the model produces a 13 percent reduction in prediction loss and an approximately 47 percent reduction in event rate compared with standard DVS event streams, although the same events perform worse than DVS events at absolute intensity frame reconstruction.

What carries the argument

Multi-stage silicon retina model that adds spatial filtering and gain control to temporal contrast detection on the SCAMP-5 pixel processor array.

If this is right

  • The model supplies a more compact event representation suited to bandwidth-limited saliency prediction on edge hardware.
  • The same processing stages can be executed in real time on existing pixel processor array chips.
  • A GPU simulation of the model allows rapid exploration of parameter choices before hardware deployment.
  • Bio-inspired distillation stages can reduce the data volume that must reach a downstream neural network.

Where Pith is reading between the lines

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

  • The same stages might be evaluated on other tasks such as optical flow or object detection to check whether the efficiency pattern generalizes.
  • Programmable pixel arrays could host task-specific retina variants tuned for different downstream networks.
  • The observed event-rate reduction could be measured directly for power or latency impact in a full hardware pipeline.

Load-bearing premise

That the measured gains with this particular lightweight network and these video datasets reflect benefits that would appear for other networks or tasks.

What would settle it

Testing the identical silicon retina event streams with a different saliency-prediction network architecture and observing no reduction in loss or no drop in event rate would falsify the efficiency claim.

Figures

Figures reproduced from arXiv: 2606.08370 by Alexandre Marcireau, Andr\'e van Schaik, Chetan Singh Thakur, Maciej Lewandowski, Piotr Dudek, Prince Philip.

Figure 1
Figure 1. Figure 1: Visual summary of the proposed framework. (a) Standard intensity frames captured by a digital camera. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Visualisation of the multi-stage retinal processing pipeline. (a) Input stimulus: Frame from a 100 FPS [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Superpixel architecture on SCAMP-5. The physical array is partitioned into [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Overview of the SCAMP-5 silicon retina software interface. a) and b) show the host control GUI used to [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Overview of the experimental pipeline and evaluation benchmarks. (a) Input video sequence. (b, c) Frame [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Performance on Video Intensity Reconstruction. (a) Training loss and (b) Validation loss over epochs. [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Performance on Video Saliency Prediction. (a) Training loss and (b) Validation loss over epochs ( the [PITH_FULL_IMAGE:figures/full_fig_p017_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Performance (Training and Validation loss) and grid occupancy of FIRENet on video saliency reconstruction [PITH_FULL_IMAGE:figures/full_fig_p018_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Impact of high-pass filter formulations on training loss and grid occupancy. Performance is evaluated across [PITH_FULL_IMAGE:figures/full_fig_p019_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Impact of spatial kernel size on neural network performance during the contrast gain control stage. Validation [PITH_FULL_IMAGE:figures/full_fig_p019_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Saliency prediction performance as a function of grid occupancy. The plot illustrates the Sparsity [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Training and validation convergence curves isolating the upsampling methodology. Forcing the DVS model [PITH_FULL_IMAGE:figures/full_fig_p021_12.png] view at source ↗
read the original abstract

Standard dynamic vision sensors approximate retinal processing by detecting temporal contrast changes, offering high speed and high dynamic range. In this work, we explore whether incorporating additional biologically inspired processing stages - specifically spatial filtering and gain control - can offer advantages for certain downstream tasks such as saliency prediction. We present the first implementation of a multi-stage Silicon Retina model on the SCAMP-5 Pixel Processor Array, along with a GPU-based simulation framework. We evaluate the performance of our model on Video Intensity Reconstruction and Video Saliency Prediction. While the bio-inspired model is less effective at reconstructing absolute intensity frames, it achieves a 13\% reduction in saliency prediction loss in comparison to standard DVS event representation, while reducing the event rate by approximately 47\%. These experiments are obtained using a lightweight $\approx 100$k-parameter FireNet-style network, adapted from event-based reconstruction to saliency prediction. These results suggest that the silicon retina's "information distillation" mechanism can achieve a more efficient representation for downstream neural networks, particularly in bandwidth-constrained edge applications.

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 paper claims to present the first hardware implementation of a multi-stage Silicon Retina model (incorporating temporal contrast detection, spatial filtering, and gain control) on the SCAMP-5 Pixel Processor Array, together with a matching GPU simulation. On video intensity reconstruction the model underperforms standard DVS; on saliency prediction with a single adapted ~100 k-parameter FireNet-style network it reports a 13 % reduction in loss and a 47 % reduction in event rate relative to standard DVS, which the authors interpret as evidence of an “information distillation” benefit for bandwidth-constrained edge applications.

Significance. A working PPA implementation of additional biologically inspired stages is a concrete engineering contribution. If the reported saliency gains prove robust, the work supplies a concrete example of hardware-level preprocessing that can lower event rate while improving a downstream task. The narrow evaluation scope, however, prevents any strong claim of generality across tasks or architectures.

major comments (2)
  1. [Abstract and Experiments section] Abstract and Experiments section: the central claim that the silicon retina supplies a general “information distillation” benefit for downstream neural networks rests entirely on results from one ~100 k-parameter FireNet-style network adapted from reconstruction to saliency prediction; no other architectures, tasks (detection, flow, classification), or datasets are evaluated, so the extrapolation is unsupported.
  2. [Abstract] Abstract: quantitative results (13 % loss reduction, 47 % event-rate reduction) are stated without error bars, statistical tests, dataset identifiers, or a reproducible experimental protocol, preventing assessment of whether the observed differences are reliable or architecture-specific.
minor comments (2)
  1. The description of how the FireNet-style network was adapted from reconstruction to saliency prediction is brief; a short architectural diagram or layer table would improve clarity.
  2. Dataset names, resolutions, and preprocessing steps for both the reconstruction and saliency experiments are not stated in the provided text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments highlighting the scope of evaluation and reporting details. Our work focuses on a concrete hardware implementation and specific results for saliency prediction; we clarify that no broad generality is claimed and address the points below.

read point-by-point responses

  1. Referee: [Abstract and Experiments section] Abstract and Experiments section: the central claim that the silicon retina supplies a general “information distillation” benefit for downstream neural networks rests entirely on results from one ~100 k-parameter FireNet-style network adapted from reconstruction to saliency prediction; no other architectures, tasks (detection, flow, classification), or datasets are evaluated, so the extrapolation is unsupported.

    Authors: The manuscript does not assert a general benefit across architectures or tasks; the abstract states that the results 'suggest' an efficient representation 'particularly in bandwidth-constrained edge applications' based on the presented saliency experiment. We agree the scope is narrow and will revise the abstract and discussion to explicitly note the limitation to this network and task, avoiding any implication of broader applicability. revision: partial

  2. Referee: [Abstract] Abstract: quantitative results (13 % loss reduction, 47 % event-rate reduction) are stated without error bars, statistical tests, dataset identifiers, or a reproducible experimental protocol, preventing assessment of whether the observed differences are reliable or architecture-specific.

    Authors: The full paper specifies the datasets and protocol used for the saliency experiments. We will revise the abstract and experiments section to include dataset identifiers, add error bars and any applicable statistical details, and expand the protocol description for reproducibility. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical measurements only

full rationale

The paper reports measured performance numbers (13% saliency loss reduction, 47% event rate reduction) obtained by running a fixed lightweight FireNet-style network on outputs from the implemented silicon retina hardware and simulation. No equations, derivations, fitted parameters, or predictions appear in the provided text. The central claims rest on direct experimental comparison rather than any self-referential construction or self-citation chain that reduces to the inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no free parameters, axioms, or invented entities are identifiable from the provided text. The contribution is an engineering implementation rather than a theoretical derivation.

pith-pipeline@v0.9.1-grok · 5724 in / 1208 out tokens · 20951 ms · 2026-06-27T18:51:00.155271+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

48 extracted references · 33 canonical work pages · 6 internal anchors

  1. [1]

    Event-Based Vision Meets Deep Learning on Steering Prediction for Self-Driving Cars

    Ana I. Maqueda et al. “Event-Based Vision Meets Deep Learning on Steering Prediction for Self-Driving Cars”. In:2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Salt Lake City, UT: IEEE, June 2018, pp. 5419–5427. DOI: 10.1109/CVPR.2018.00568.URL: https://ieeexp...

    work page doi:10.1109/cvpr.2018.00568.url: 2018

  2. [2]

    Fast Image Reconstruction with an Event Camera

    Cedric Scheerlinck et al. “Fast Image Reconstruction with an Event Camera”. In:2020 IEEE Winter Conference on Applications of Computer Vision (WACV). 2020 IEEE Winter Conference on Applications of Computer Vision (WACV). Snowmass Village, CO, USA: IEEE, Mar. 2020, pp. 156–163.DOI: 10.1109/WACV45572.2020. 9093366.URL:https://ieeexplore.ieee.org/document/90...

    work page doi:10.1109/wacv45572.2020 2020

  3. [3]

    Drone Detection with Event Cameras

    Gabriele Magrini et al. “Drone Detection with Event Cameras”. In: ().URL: https://arxiv.org/abs/2508. 04564

  4. [4]

    Event-Based Vision: A Survey ,

    Guillermo Gallego et al. “Event-based Vision: A Survey”. In:IEEE Transactions on Pattern Analysis and Machine Intelligence44.1 (Jan. 1, 2022), pp. 154–180.DOI: 10.1109/TPAMI.2020.3008413. arXiv: 1904.08405[cs]. URL:http://arxiv.org/abs/1904.08405(visited on 12/06/2025)

    work page doi:10.1109/tpami.2020.3008413 2022

  5. [5]

    Retinomorphic Event-Based Vision Sensors: Bioinspired Cameras With Spiking Output

    Christoph Posch et al. “Retinomorphic Event-Based Vision Sensors: Bioinspired Cameras With Spiking Output”. In:Proceedings of the IEEE102.10 (Oct. 2014), pp. 1470–1484.DOI: 10.1109/JPROC.2014.2346153.URL: https://ieeexplore.ieee.org/document/6887319/(visited on 12/06/2025)

    work page doi:10.1109/jproc.2014.2346153.url: 2014

  6. [6]

    Prince Philip et al.Neuromorphic Retina: An FPGA-based Emulator. Jan. 15, 2025.DOI: 10.48550/arXiv. 2501 . 08943. arXiv: 2501 . 08943[eess].URL: http : / / arxiv . org / abs / 2501 . 08943(visited on 12/06/2025)

    work page internal anchor Pith review doi:10.48550/arxiv 2025

  7. [7]

    Bharatesh Chakravarthi et al.Recent Event Camera Innovations: A Survey. Aug. 27, 2024.DOI: 10.48550/ arXiv.2408.13627 . arXiv: 2408.13627[cs] .URL: http://arxiv.org/abs/2408.13627 (visited on 12/06/2025)

    arXiv 2024

  8. [8]

    Astrometric calibration and source characterisation of the latest generation neu- romorphic event-based cameras for space imaging

    Nicholas Owen Ralph et al. “Astrometric calibration and source characterisation of the latest generation neu- romorphic event-based cameras for space imaging”. In:Astrodynamics7.4 (Dec. 1, 2023), pp. 415–443.DOI: 10.1007/s42064- 023- 0168- 2 .URL: https://doi.org/10.1007/s42064- 023- 0168- 2 (visited on 12/06/2025)

    work page doi:10.1007/s42064- 2023

  9. [9]

    Sami Arja et al.Noise Filtering Benchmark for Neuromorphic Satellites Observations. Nov. 18, 2024.DOI: 10.48550/arXiv.2411.11233 . arXiv: 2411.11233[cs].URL: http://arxiv.org/abs/2411.11233 (visited on 12/09/2025). 22 APREPRINT- JUNE9, 2026

    work page doi:10.48550/arxiv.2411.11233 2024

  10. [10]

    Robust Event-Based Object Tracking Combining Correlation Filter and CNN Representation

    Hongmin Li and Luping Shi. “Robust Event-Based Object Tracking Combining Correlation Filter and CNN Representation”. In:Frontiers in Neurorobotics13 (Oct. 10, 2019).DOI: 10 . 3389 / fnbot . 2019 . 00082. URL: https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2019. 00082/full(visited on 12/06/2025)

    work page doi:10.3389/fnbot.2019 2019

  11. [11]

    Simultaneous Optical Flow and Intensity Estima- tion from an Event Camera

    Patrick Bardow, Andrew J. Davison, and Stefan Leutenegger. “Simultaneous Optical Flow and Intensity Estima- tion from an Event Camera”. In:2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA: IEEE, June 2016, pp. 884–892.DOI: 10.1109/CVPR.2016.10...

    work page doi:10.1109/cvpr.2016.102 2016

  12. [12]

    High Speed and High Dynamic Range Video with an Event Camera

    Henri Rebecq et al.High Speed and High Dynamic Range Video with an Event Camera. June 15, 2019.DOI: 10.48550/arXiv.1906.07165 . arXiv: 1906.07165[cs].URL: http://arxiv.org/abs/1906.07165 (visited on 12/06/2025)

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1906.07165 2019

  13. [13]

    Issue: arXiv:2105.14753

    Amélie Gruel and Jean Martinet.Bio-inspired visual attention for silicon retinas based on spiking neural networks applied to pattern classification. Issue: arXiv:2105.14753. May 31, 2021.DOI: 10.48550/arXiv.2105.14753. arXiv:2105.14753[cs].URL:http://arxiv.org/abs/2105.14753(visited on 12/06/2025)

    work page doi:10.48550/arxiv.2105.14753 2021

  14. [14]

    Issue: arXiv:2504.06748

    Sirine Arfa et al.Efficient Deployment of Spiking Neural Networks on SpiNNaker2 for DVS Gesture Recognition Using Neuromorphic Intermediate Representation. Issue: arXiv:2504.06748. Apr. 9, 2025.DOI: 10.48550/ arXiv.2504.06748 . arXiv: 2504.06748[cs] .URL: http://arxiv.org/abs/2504.06748 (visited on 12/08/2025)

    arXiv 2025

  15. [15]

    Issue: arXiv:2203.17149

    Simon Schaefer, Daniel Gehrig, and Davide Scaramuzza.AEGNN: Asynchronous Event-based Graph Neural Networks. Issue: arXiv:2203.17149. Nov. 1, 2022.DOI: 10 . 48550 / arXiv . 2203 . 17149. arXiv: 2203 . 17149[cs].URL:http://arxiv.org/abs/2203.17149(visited on 12/06/2025)

    arXiv 2022

  16. [16]

    EvGNN: An Event-driven Graph Neural Network Ac- celerator for Edge Vision

    Yufeng Yang, Adrian Kneip, and Charlotte Frenkel. “EvGNN: An Event-driven Graph Neural Network Ac- celerator for Edge Vision”. In:IEEE Transactions on Circuits and Systems for Artificial Intelligence2.1 (Mar. 2025), pp. 37–50.DOI: 10 . 1109 / TCASAI . 2024 . 3520905. arXiv: 2404 . 19489[cs].URL: http : //arxiv.org/abs/2404.19489(visited on 12/06/2025)

    arXiv 2025

  17. [17]

    Continuous-time Intensity Estimation Using Event Cameras

    Cedric Scheerlinck, Nick Barnes, and Robert Mahony.Continuous-time Intensity Estimation Using Event Cam- eras. Issue: arXiv:1811.00386. Nov. 1, 2018.DOI: 10.48550/arXiv.1811.00386. arXiv: 1811.00386[cs]. URL:http://arxiv.org/abs/1811.00386(visited on 12/06/2025)

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1811.00386 2018

  18. [18]

    PROPHESEE.URL: https://www.prophesee.ai/event-based-sensors/ (visited on 02/09/2026)

    Metavision® Sensors. PROPHESEE.URL: https://www.prophesee.ai/event-based-sensors/ (visited on 02/09/2026). [19]Buy – iniVation.URL:https://inivation.com/buy/(visited on 02/09/2026)

    2026

  19. [19]

    Thomas Finateu et al. “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- Formatting Pipeline”. In:2020 IEEE International Solid-State Circuits Conference - (ISSCC). 2020, pp. 112–114. DOI:10.1109/ISSCC19947.2020.9063149

    work page doi:10.1109/isscc19947.2020.9063149 2020

  20. [20]

    The Silicon Retina

    Misha A. Mahowald and Carver Mead. “The Silicon Retina”. In:Scientific American264.5 (1991), pp. 76–83. URL:http://www.jstor.org/stable/24936904(visited on 06/04/2026)

    arXiv 1991

  21. [21]

    Virtual Retina: A biological retina model and simulator, with contrast gain control

    Adrien Wohrer and Pierre Kornprobst. “Virtual Retina: A biological retina model and simulator, with contrast gain control”. In:Journal of Computational Neuroscience26.2 (Apr. 1, 2009), pp. 219–249.DOI: 10.1007/s10827- 008-0108-4.URL:https://doi.org/10.1007/s10827-008-0108-4(visited on 12/06/2025)

    work page doi:10.1007/s10827- 2009

  22. [22]

    Identification of a possible superconducting transition above room temperature in natural graphite crystals

    G.D. Field and E.J. Chichilnisky. “Information Processing in the Primate Retina: Circuitry and Coding”. In: Annual Review of Neuroscience30.1 (July 1, 2007), pp. 1–30.DOI: 10.1146/annurev.neuro.30.051606. 094252.URL: https://www.annualreviews.org/doi/10.1146/annurev.neuro.30.051606.094252 (visited on 12/06/2025)

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1146/annurev.neuro.30.051606 2007

  23. [23]

    Sensor-level computer vision with pixel processor arrays for agile robots

    Piotr Dudek et al. “Sensor-level computer vision with pixel processor arrays for agile robots”. In:Science Robotics7.67 (June 29, 2022), eabl7755.DOI: 10 . 1126 / scirobotics . abl7755.URL: https : / / www . science.org/doi/10.1126/scirobotics.abl7755(visited on 12/06/2025)

    work page doi:10.1126/scirobotics.abl7755(visited 2022

  24. [24]

    A 100,000 fps vision sensor with embedded 535GOPS/W 256×256 SIMD processor array

    Stephen J. Carey et al. “A 100,000 fps vision sensor with embedded 535GOPS/W 256×256 SIMD processor array”. In:2013 Symposium on VLSI Circuits. 2013 Symposium on VLSI Circuits. June 2013, pp. C182–C183. URL:https://ieeexplore.ieee.org/document/6578654(visited on 12/06/2025)

    arXiv 2013

  25. [25]

    Descriptor-In-Pixel : Point-Feature Tracking For Pixel Processor Arrays

    Laurie Bose, Jianing Chen, and Piotr Dudek. “Descriptor-In-Pixel : Point-Feature Tracking For Pixel Processor Arrays”. In:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 2025, pp. 5392–5400. 23 APREPRINT- JUNE9, 2026

    2025

  26. [26]

    Pixel Processor Arrays For Low Latency Gaze Estimation

    Laurie Bose et al. “Pixel Processor Arrays For Low Latency Gaze Estimation”. In:2022 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW). 2022 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW). Christchurch, New Zealand: IEEE, Mar. 2022, pp. 970– 971.DOI: 10.1109/VRW55335.2022.00336...

    work page doi:10.1109/vrw55335.2022.00336.url: 2022

  27. [27]

    So et al.PixelRNN: In-pixel Recurrent Neural Networks for End-to-end-optimized Perception with Neural Sensors

    Haley M. So et al.PixelRNN: In-pixel Recurrent Neural Networks for End-to-end-optimized Perception with Neural Sensors. Apr. 11, 2023.DOI: 10.48550/arXiv.2304.05440 . arXiv: 2304.05440[cs].URL: http: //arxiv.org/abs/2304.05440(visited on 12/06/2025)

    work page doi:10.48550/arxiv.2304.05440 2023

  28. [28]

    Pixel interlacing to trade off the resolution of a cellular processor array against more registers

    Julien N. P. Martel et al. “Pixel interlacing to trade off the resolution of a cellular processor array against more registers”. In:2015 European Conference on Circuit Theory and Design (ECCTD). 2015, pp. 1–4.DOI: 10.1109/ECCTD.2015.7300011

    work page doi:10.1109/ecctd.2015.7300011 2015

  29. [29]

    Daniel Gehrig et al.Video to Events: Recycling Video Datasets for Event Cameras. Apr. 1, 2020.DOI: 10. 48550/arXiv.1912.03095. arXiv: 1912.03095[cs].URL: http://arxiv.org/abs/1912.03095 (visited on 12/06/2025)

    arXiv 2020

  30. [30]

    Fast and Slow Contrast Adaptation in Retinal Circuitry

    Stephen A. Baccus and Markus Meister. “Fast and Slow Contrast Adaptation in Retinal Circuitry”. In:Neuron 36.5 (Dec. 2002), pp. 909–919.DOI: 10.1016/S0896- 6273(02)01050- 4 .URL: https://linkinghub. elsevier.com/retrieve/pii/S0896627302010504(visited on 12/06/2025)

    work page doi:10.1016/s0896- 2002

  31. [31]

    Understanding the Retina: A Review of Computational Models of the Retina from the Single Cell to the Network Level

    Tianruo Guo et al. “Understanding the Retina: A Review of Computational Models of the Retina from the Single Cell to the Network Level”. In:Critical Reviews in Biomedical Engineering42.5 (2014), pp. 419–436. DOI: 10.1615/CritRevBiomedEng.2014011732 .URL: http://www.dl.begellhouse.com/journals/ 4b27cbfc562e21b8,64a3e6f7290a8a6e,64cd0e236c1f5579.html(visite...

    work page doi:10.1615/critrevbiomedeng.2014011732 2014

  32. [32]

    Tau-Cell-Based Analog Silicon Retina With Spatio-Temporal Filtering and Contrast Gain Control

    Prince Philip et al. “Tau-Cell-Based Analog Silicon Retina With Spatio-Temporal Filtering and Contrast Gain Control”. In:IEEE Transactions on Biomedical Circuits and Systems18.2 (2024), pp. 423–437.DOI: 10.1109/ TBCAS.2023.3332117

    arXiv 2024

  33. [33]

    The effect of contrast on the transfer properties of cat retinal ganglion cells

    R M Shapley and J D Victor. “The effect of contrast on the transfer properties of cat retinal ganglion cells.” In: The Journal of Physiology285 (Dec. 1978), pp. 275–298.DOI: 10.1113/jphysiol.1978.sp012571 .URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC1281756/(visited on 12/09/2025)

    work page doi:10.1113/jphysiol.1978.sp012571 1978

  34. [34]

    Feature Extraction using a Portable Vision System

    Jianing Chen, Stephen J Carey, and Piotr Dudek. “Feature Extraction using a Portable Vision System”. In: ()

  35. [35]

    Real-Time Depth From Focus on a Programmable Focal Plane Processor

    Julien N. P. Martel et al. “Real-Time Depth From Focus on a Programmable Focal Plane Processor”. In:IEEE Transactions on Circuits and Systems I: Regular Papers65.3 (Mar. 2018), pp. 925–934.DOI: 10.1109/TCSI. 2017.2753878.URL:https://ieeexplore.ieee.org/document/8071011/(visited on 12/06/2025)

    work page doi:10.1109/tcsi 2018

  36. [36]

    A 128× 128 120 dB 15 µs Latency Asynchronous Temporal Contrast Vision Sensor

    Patrick Lichtsteiner, Christoph Posch, and Tobi Delbruck. “A 128× 128 120 dB 15 µs Latency Asynchronous Temporal Contrast Vision Sensor”. In:IEEE Journal of Solid-State Circuits43.2 (2008), pp. 566–576.DOI: 10.1109/JSSC.2007.914337

    work page doi:10.1109/jssc.2007.914337 2008

  37. [37]

    A Spatial Contrast Retina With On-Chip Calibration for Neuromorphic Spike-Based AER Vision Systems

    Jess Costas-Santos et al. “A Spatial Contrast Retina With On-Chip Calibration for Neuromorphic Spike-Based AER Vision Systems”. In:IEEE Transactions on Circuits and Systems I: Regular Papers54.7 (July 2007), pp. 1444–1458.DOI: 10.1109/TCSI.2007.900179 .URL: https://ieeexplore.ieee.org/document/ 4268401(visited on 12/14/2025)

    work page doi:10.1109/tcsi.2007.900179 2007

  38. [38]

    Live demonstration: Asynchronous time-based image sensor (ATIS) camera with full- custom AE processor

    Christoph Posch et al. “Live demonstration: Asynchronous time-based image sensor (ATIS) camera with full- custom AE processor”. In:Proceedings of 2010 IEEE International Symposium on Circuits and Systems. 2010 IEEE International Symposium on Circuits and Systems. May 2010, pp. 1392–1392.DOI: 10.1109/ISCAS. 2010.5537265.URL:https://ieeexplore.ieee.org/docu...

    work page doi:10.1109/iscas 2010

  39. [39]

    A Five-Decade Dynamic-Range Ambient-Light-Independent Calibrated Signed-Spatial-Contrast AER Retina With 0.1-ms Latency and Optional Time-to-First-Spike Mode

    Juan Antonio Leñero-Bardallo, Teresa Serrano-Gotarredona, and Bernabé Linares-Barranco. “A Five-Decade Dynamic-Range Ambient-Light-Independent Calibrated Signed-Spatial-Contrast AER Retina With 0.1-ms Latency and Optional Time-to-First-Spike Mode”. In:IEEE Transactions on Circuits and Systems I: Regular Papers57.10 (Oct. 2010), pp. 2632–2643.DOI: 10.1109/...

    work page doi:10.1109/tcsi.2010.2046971.url: 2010

  40. [40]

    Live demonstration: The “DA VIS

    C. Brandli et al. “Live demonstration: The “DA VIS” Dynamic and Active-Pixel Vision Sensor”. In:2014 IEEE International Symposium on Circuits and Systems (ISCAS). 2014 IEEE International Symposium on Circuits and Systems (ISCAS). June 2014, pp. 440–440.DOI: 10.1109/ISCAS.2014.6865163.URL: https: //ieeexplore.ieee.org/document/6865163/(visited on 12/14/2025)

    work page doi:10.1109/iscas.2014.6865163.url: 2014

  41. [41]

    Retinal Circuit Emulator With Spatiotemporal Spike Outputs at Millisecond Resolution in Response to Visual Events

    Yuki Hayashida et al. “Retinal Circuit Emulator With Spatiotemporal Spike Outputs at Millisecond Resolution in Response to Visual Events”. In:IEEE transactions on biomedical circuits and systems11.3 (June 2017), pp. 597–611.DOI:10.1109/TBCAS.2017.2662659. 24 APREPRINT- JUNE9, 2026

    work page doi:10.1109/tbcas.2017.2662659 2017

  42. [42]

    Efficient Meshflow and Optical Flow Estimation from Event Cameras

    Xinglong Luo et al. “Efficient Meshflow and Optical Flow Estimation from Event Cameras”. In:2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA, USA: IEEE, June 16, 2024, pp. 19198–19207.DOI: 10 . 1109 / CVPR52733 . 2024 . 01816.URL: https : / /...

    2024

  43. [43]

    Issue: arXiv:2406.09409

    Sachin Shah et al.CodedEvents: Optimal Point-Spread-Function Engineering for 3D-Tracking with Event Cameras. Issue: arXiv:2406.09409. June 13, 2024.DOI: 10 . 48550 / arXiv . 2406 . 09409. arXiv: 2406 . 09409[cs].URL:http://arxiv.org/abs/2406.09409(visited on 12/06/2025)

    arXiv 2024

  44. [44]

    v2e: From Video Frames to Realistic DVS Events

    Yuhuang Hu, Shih-Chii Liu, and Tobi Delbruck. “v2e: From Video Frames to Realistic DVS Events”. In: ()

  45. [45]

    Revisiting Video Saliency Prediction in the Deep Learning Era

    Wenguan Wang et al. “Revisiting Video Saliency Prediction in the Deep Learning Era”. In:IEEE Transactions on Pattern Analysis and Machine Intelligence43.1 (Jan. 2021), pp. 220–237.DOI: 10.1109/TPAMI.2019.2924417. URL:https://ieeexplore.ieee.org/document/8744328(visited on 01/05/2026)

    work page doi:10.1109/tpami.2019.2924417 2021

  46. [46]

    Christian Reinbacher, Gottfried Graber, and Thomas Pock.Real-Time Intensity-Image Reconstruction for Event Cameras Using Manifold Regularisation. Aug. 4, 2016.DOI: 10.48550/arXiv.1607.06283 . arXiv: 1607.06283[cs].URL:http://arxiv.org/abs/1607.06283(visited on 12/06/2025)

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1607.06283 2016

  47. [47]

    Events-To-Video: Bringing Modern Computer Vision to Event Cameras

    Henri Rebecq et al. “Events-To-Video: Bringing Modern Computer Vision to Event Cameras”. In:2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach, CA, USA: IEEE, June 2019, pp. 3852–3861. DOI: 10.1109/CVPR.2019.00398.URL: https://ieeexplore.ieee.or...

    work page doi:10.1109/cvpr.2019.00398.url: 2019

  48. [48]

    Richard Zhang et al.The Unreasonable Effectiveness of Deep Features as a Perceptual Metric. Apr. 10, 2018.DOI: 10.48550/arXiv.1801.03924 . arXiv: 1801.03924[cs].URL: http://arxiv.org/abs/1801.03924 (visited on 12/06/2025). 25

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1801.03924 2018