Pith. sign in

REVIEW 4 major objections 7 minor 52 references

Hybrid Spiking Vision Transformer for Object Detection with Event Cameras

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a hybrid ANN-SNN vision Transformer, HsVT, matches or beats the RVT reference on event-based object detection while using fewer parameters, and introduces a privacy-preserving fall-detection event benchmark.

desk verdict HsVT is a coherent hybrid ANN-SNN detector with a real spiking temporal module and a useful simulated benchmark, but the headline gains over RVT are within run-to-run noise and the significance claim is not supported. read the letter →

arxiv 2505.07715 v1 pith:HBYNH6NB submitted 2025-05-12 cs.CV cs.AI

Qi Xu , Jie Deng , Jiangrong Shen , Biwu Chen , Huajin Tang , Gang Pan This is my paper · ORCID
classification cs.CVcs.AI
keywords event-basedobjectdetectionspikingneuralnetworksvisiontransformerhybridANN-SNNeventcamerafalldatasetparameterefficiencytemporalfeatureextraction
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

HsVT is a four-block hybrid network for event-camera object detection: ANN-style block and grid self-attention paired with SpikingMLP extract spatial features, while LSTM cells in the first three blocks and a spiking temporal-feature module in the last block carry information across time. The paper's claim is that this combination is competitive with the strong RVT baseline on GEN1—HsVT-B reaches 0.478 mAP 50:95 with 17.2M parameters against RVT's 0.472 with 18.5M—and better than RVT on the FALL and AIR datasets for every model size. The paper also releases the Fall Detection Dataset, an event-stream benchmark produced by converting Le2i fall videos through the ESIM simulator, arguing that event representation protects privacy and reduces storage. A sympathetic reader would see this as evidence that spiking temporal modules can replace heavier recurrent units without giving up accuracy, which matters for low-power, privacy-conscious monitoring.

What carries the argument

The load-bearing mechanism is the four-block HsVT backbone with two complementary paths. Vertically, each block passes spatial features down the network; horizontally, each block receives the temporal output of the adjacent time step, letting information accumulate over time. Spatial extraction combines MaxViT's Block-SA and Grid-SA self-attention with SpikingMLP; temporal extraction uses LSTM in blocks 1 through 3 and STFE, a convolutional-batchnorm-spiking-neuron module with an LSTM-like recurrent section, in block 4. This layout is what lets the paper attribute spatial coverage to ANN attention and temporal modeling to cheap spiking units, and the ablation isolating STFE placement is the evidence that final-block placement carries the performance.

What would settle it

Record the same fall scenarios with a real event camera and run HsVT and RVT under identical training and evaluation; if the real-data mAP gap does not reproduce the synthetic-data gap, the fall-detection advantage is an artifact of simulation. A simpler check is to corrupt or re-render the synthetic events with realistic sensor noise and see whether the claimed margin survives.

Watch

Extended reading notes

Core claim

The core discovery is that a deliberately hybrid computation—ANN self-attention for spatial structure, spiking neurons for temporal structure—yields a parameter-efficient event detector. On GEN1 the base HsVT reports 0.478 mAP 50:95 with 17.2M parameters, slightly above RVT's 0.472 with 18.5M; tiny and small variants reach 0.449 and 0.465. On the FALL and AIR datasets, HsVT reports higher mAP than RVT across all variants (for example, AIR tiny 0.641 versus 0.613, and FALL small 0.492 versus 0.466). The authors attribute the gain to replacing LSTM-only temporal encoding with a final-block spiking temporal feature extraction module, which in ablation reaches 0.640 mAP on AIR with 0.20M parameters and 101.19M FLOPs, together with horizontal propagation of temporal information between blocks.

Load-bearing premise

The load-bearing assumption is that the synthetic event streams generated by the ESIM simulator from Le2i frame videos are faithful enough to real event-camera data, including sensor noise, dynamic range, and temporal correlations, that the reported fall-detection numbers predict real-world performance.

Editorial extensions

If this is right

  • On GEN1, HsVT-B reaches 0.478 mAP 50:95 with 17.2M parameters, slightly above RVT's 0.472 with 18.5M, while HsVT-T and HsVT-S stay close to RVT at 4.1M and 9.1M parameters.
  • On FALL and AIR, every HsVT variant beats the corresponding RVT variant, with examples including AIR tiny 0.641 versus 0.613 and FALL small 0.492 versus 0.466.
  • The ablation supports replacing LSTM-only temporal encoding with the STFE module in the final block, which reaches the highest AIR mAP with fewer parameters than LSTM.
  • The released Fall Detection Dataset gives the community a public event-stream benchmark for fall detection that avoids facial exposure and stores only contrast-change events, lowering memory use.
  • The paper's energy estimates place HsVT-B at 134.5 mJ on GEN1, above pure-SNN detectors such as SFOD at 7.26 mJ and EAS-SNN-M at 28.10 mJ, with the added cost attributed to the ANN backbone and FPN plus head.

Reading between the lines

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

  • An implication the authors leave implicit: if this hybrid pattern transfers, heavy LSTM temporal modules in other event-based tasks such as tracking, action recognition, or pose estimation could be replaced by spiking temporal modules, cutting parameters without necessarily losing accuracy.
  • The FALL benchmark's privacy and memory advantages come from the event representation itself, but the dataset is synthetic; capturing a real event-camera fall dataset with the same annotations would test whether the advantage survives real sensor noise.
  • The paper reports that larger variants do not consistently beat smaller ones on FALL and AIR and attributes this to dataset size and overfitting; a larger or more diverse real event-camera corpus could change the scaling behavior and model-size rankings.
  • The ablation finding that STFE performs best only in the final block suggests a design principle not stated by the paper: spiking temporal extraction may be most useful after spatial features are already abstract, rather than in early layers.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 7 minor

Summary. The paper proposes HsVT, a hybrid ANN/SNN multi-stage vision transformer for event-based object detection. The architecture combines MaxViT-based spatial feature extraction (Block-SA and Grid-SA with SpikingMLP) with LSTM and a spiking temporal feature extraction (STFE) module in the final block. The authors introduce a new Fall Detection Dataset, generated by converting the frame-based Le2i dataset into event streams with the ESIM simulator, and they evaluate HsVT on GEN1, FALL, and a private aircraft-detection dataset. On GEN1, HsVT-B reaches 0.478 mAP 50:95 versus RVT's 0.472, while HsVT-T and HsVT-S reach 0.449 and 0.465. On FALL and AIR, HsVT variants are reported to outperform RVT by small margins. The paper also reports ablations over spiking neuron models, surrogate gradients, SNN components, and component placement, plus theoretical energy estimates.

Significance. If the performance claims were robust, HsVT would be a useful parameter-efficient hybrid design for event-based detection, and the public release of a privacy-preserving fall-detection event benchmark would be a community contribution. The paper's strengths are the systematic ablation study, the parameter counts for each variant, and the explicit energy-estimation methodology. However, the central quantitative claim of 'significant performance improvements' is not yet supported: the main GEN1 gain over RVT is 0.006 mAP, the FALL/AIR gains are also small, and no error bars, seed counts, or significance tests are reported. The FALL dataset is simulated rather than sensor-captured, contrary to the abstract's wording, which limits the real-world transferability of the FALL results. The two supporting comparisons on non-public or synthetic data cannot compensate for the lack of statistical evidence on GEN1.

major comments (4)
  1. [Abstract; Section 5.4; Table 7] The abstract and Section 5.4 claim 'significant performance improvements', but the GEN1 comparison in Table 7 shows HsVT-B at 0.478 mAP versus RVT at 0.472, a delta of 0.006, and HsVT-T and HsVT-S at 0.449 and 0.465, below RVT. No standard deviations, number of seeds, or significance tests are reported anywhere in the paper, so the claimed significance is not established; a 0.006 mAP difference is within normal run-to-run variance for event-based detection training. Please report multi-seed results with error bars or confidence intervals for all three datasets and adjust the wording of the central claim accordingly.
  2. [Table 7] Table 7 contains internally inconsistent values: STAT is listed as 49.9 and SpikSSD as 40.8, while every other row uses mAP values between 0 and 1. These entries appear to be percentages and should likely read 0.499 and 0.408. As printed, the table cannot support the comparative claims; please correct these values and verify all baseline numbers against their original sources.
  3. [Section 3.1; Abstract] The abstract states that the Fall Detection Dataset was 'captured using an event-based camera', but Section 3.1 explicitly says the dataset was generated by converting the frame-based Le2i videos into event streams with the ESIM simulator. This is a direct contradiction. Because FALL is one of the two datasets where HsVT is reported to consistently outperform RVT, the simulated nature of the events must be stated accurately, and the paper should discuss how well ESIM-simulated events approximate real event-camera data in this deployment scenario.
  4. [Section 5.4; Table 8] On FALL and AIR, Table 8 reports HsVT outperforming RVT by margins of only 0.004 to 0.028 mAP, and the AIR dataset is not publicly available. Without repeated runs and statistical tests, the claim that 'HsVT consistently outperforms RVT on both datasets' is not supported by the evidence presented. Moreover, because the FALL benchmark is synthetic and AIR is proprietary, the only independently checkable comparison where HsVT beats RVT is the 0.006 GEN1 gap, which is not statistically established.
minor comments (7)
  1. [Section 5.1] There is a duplicated article in 'The The Aircraft Detection Dataset'; please fix this typo.
  2. [Section 2.3] 'fast interference' should be 'fast inference' in the description of RVT.
  3. [Section 3.1] 'identifie' is a typo; it should be 'identify'.
  4. [Table 4] The column header 'FALL' followed by '1000ms 200ms AIR' is confusing; please label the columns as separate dataset/time-interval conditions. Also, the 200ms FALL mAP in Table 4 (0.490 for ATan) differs from the value 0.487 in Table 2 for the same interval, and this discrepancy should be reconciled.
  5. [Table 9] The Tiny row reports E_SNN = 0.017 mJ, but Equation (6) with SOP = 1156.00M gives approximately 1.04 mJ, matching the scale of the Small and Base rows; this looks like a typo and should be corrected.
  6. [References] The reference list includes Spikformer (Zhou et al., 2022) but the text does not cite it; please either cite it where relevant or remove it.
  7. [Section 3.2] The dataset release is only described as 'our Dropbox repository'; for a public benchmark, please provide a persistent identifier such as a DOI or an institutional repository link.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: HsVT's performance claims are evaluated against external RVT baselines and standard GEN1 data; self-citations are background only.

full rationale

The central claim is that HsVT improves event-based detection with fewer parameters. This is checked empirically against RVT (Gehrig & Scaramuzza, 2023) on GEN1 (Table 7) and against RVT on FALL/AIR (Table 8); these baselines are external, and GEN1 is a standard public benchmark. The HsVT design is not defined in terms of its own outputs: spatial features use MaxViT attention and SpikingMLP, temporal features use LSTM in the first three blocks and STFE in the fourth, and the only fitted-looking choices are reported as ablations. The FALL dt=200 ms choice is explicitly justified by the reported comparison in Table 2, i.e., honest tuning with evidence, not a hidden fitted parameter renamed as a prediction. The ablation-based selection of LIF neurons and STFE placement on the AIR test set (Tables 3-6) is a test-set selection / overfitting concern that weakens the strength of the AIR comparison, but it does not make the final mAP equal to the selection criterion by construction, so it is not circularity under the stated definition. Self-citations (Xu et al.; Shen et al.; Darwin3) are background SNN/hardware literature and are not load-bearing for the detection result, and no uniqueness theorem is imported from the authors. The lack of error bars or significance tests for the 0.006 GEN1 mAP gap is a statistical-reporting issue rather than a circular-derivation issue.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper is empirical, so the axioms capture assumptions about data validity and benchmark fairness. No free parameters in the mathematical sense are fit, but preprocessing choices (event accumulation intervals, neuron type, surrogate gradient) are selected by validation performance and affect the main results. There are no invented physical entities; STFE is an architectural component, not a new physical object.

free parameters (5)
  • Event accumulation interval for FALL, dt_fall = 200 ms
    Selected as best among 40, 200, 1000 ms based on mAP in Table 2 (0.487 vs 0.445/0.405).
  • Event accumulation interval for AIR, dt_air = 10 ms
    Set to match the 100 Hz labeling frequency; this is derived from annotation rate, not fit.
  • Event accumulation interval for GEN1, dt_GEN1 = 50 ms
    Taken from Gehrig & Scaramuzza (2023), not independently fit.
  • Spiking neuron model = LIF
    LIF outperformed IF in SpikingMLP ablation (Table 3: 0.476 vs 0.441 on FALL).
  • Surrogate gradient function = ATan (implied by Table 4)
    ATan outperforms Sigmoid for the chosen 200 ms FALL setting (0.490 vs 0.473) and on AIR (0.630 vs 0.603).
assumptions (4)
  • domain assumption ESIM produces realistic event streams that are representative of real event-camera data for object detection.
    Section 3.1 converts the frame-based Le2i dataset with ESIM; the validity of the Fall benchmark depends on this.
  • domain assumption The Le2i fall detection annotations (bounding boxes and fall timing) are accurate and usable after conversion to event frames.
    Section 3.2 adopts the Le2i labels without further validation.
  • domain assumption RVT and other baselines are fairly compared under the same training settings as HsVT.
    Section 5.2 states the authors 'maintained the consistency' of settings, but no explicit verification of equivalent training budgets for baselines is given.
  • domain assumption The theoretical energy estimation model (pJ/FLOP from Spikformer, spike operation calculation) is a valid proxy for real energy consumption.
    Appendix A computes energy with E=4.6pJ*FLOPs and E=0.9pJ*SOPs with an assumed firing rate; no hardware measurement is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hybrid Spiking Vision Transformer for Object Detection with Event Cameras." pith.science (2026). https://pith.science/paper/HBYNH6NB

@misc{pith2026250507715,
  author       = {Pith},
  title        = {Pith review of: Hybrid Spiking Vision Transformer for Object Detection with Event Cameras},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HBYNH6NB}},
  note         = {Machine review of arXiv:2505.07715}
}
read the original abstract

Event-based object detection has gained increasing attention due to its advantages such as high temporal resolution, wide dynamic range, and asynchronous address-event representation. Leveraging these advantages, Spiking Neural Networks (SNNs) have emerged as a promising approach, offering low energy consumption and rich spatiotemporal dynamics. To further enhance the performance of event-based object detection, this study proposes a novel hybrid spike vision Transformer (HsVT) model. The HsVT model integrates a spatial feature extraction module to capture local and global features, and a temporal feature extraction module to model time dependencies and long-term patterns in event sequences. This combination enables HsVT to capture spatiotemporal features, improving its capability to handle complex event-based object detection tasks. To support research in this area, we developed and publicly released The Fall Detection Dataset as a benchmark for event-based object detection tasks. This dataset, captured using an event-based camera, ensures facial privacy protection and reduces memory usage due to the event representation format. We evaluated the HsVT model on GEN1 and Fall Detection datasets across various model sizes. Experimental results demonstrate that HsVT achieves significant performance improvements in event detection with fewer parameters.

Figures

Figures reproduced from arXiv: 2505.07715 by the authors.

Figure 1
Figure 1. Event Stream Representation with Time Intervals. Each vertical line represents an event occurrence, while equidistant red lines represent the time intervals. Each event is represented by a triplet (t,⟨x, y⟩, p), denoting its spatial and temporal coordinates. where t represents the timestamp; x and y represent the pixel coordinates; p represents the polarity. The timestamp signifies the time of the event, pixel coord… view at source ↗
Figure 2
Figure 2. Architecture of the HsVT Network. The architecture of the proposed HsVT network consists of four Blocks, each incor￾porating spatial feature extraction and temporal feature extraction components. 4.3. Spatial Feature Extraction In each BLOCK, spatial feature extraction is composed of MaxViT (Tu et al., 2022) and SpikingMLP (Spiking Multilayer Perceptron), as illustrated in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. Visual comparison of attention patterns from Block-SA and Grid-SA. (a) Block-SA captures local fine-grained spatial correlations. (b) Grid-SA highlights broader, long-range dependen￾cies. (c) The overlay illustrates how these mechanisms complement each other in spatial understanding. study, LSTM serves as a crucial element for capturing both temporal dependencies and long-term dependencies within input sequences acr… view at source ↗
Figures from the paper (4 more)
Figure 3
Figure 3. Figure 3: Spatial Feature Extraction in the HsVT Architecture. This figure illustrates the process of spatial feature extraction within the HsVT architecture. The components, including Block￾SA, SpikingMLP, Grid-SA, and a second SpikingMLP, collabora￾tively work to extract spati…
Figure 5
Figure 5. Figure 5: Spiking Temporal Feature Extraction in the HsVT Archi￾tecture. This figure illustrates the process of temporal feature ex￾traction within the HsVT architecture, where the first three blocks employ the LSTM model and the final block utilizes the STFE model. This archite…
Figure 6
Figure 6. Figure 6: Visualization of Aircraft Detection Dataset. (a) 40ms Integration (b) 200ms Integration (c) 1000ms Integration [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Event Frame Visualizations at Different Temporal Inte￾gration Intervals. 5.2. Experimental settings We employ a series of optimization methods and training techniques to ensure the training effectiveness and perfor￾mance of the model. Specifically, we used the RVT mode…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 30 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    H., Zhai, A., and Kislyuk, D

    Beal, J., Kim, E., Tzeng, E., Park, D. H., Zhai, A., and Kislyuk, D. Toward transformer-based object detection. arXiv preprint arXiv:2012.09958, 2020

    arXiv 2012

  3. [3]

    B., Schwaiger, F., Kreuzberg, L., and Behnke, S

    Bodden, L., Ha, D. B., Schwaiger, F., Kreuzberg, L., and Behnke, S. Spiking centernet: A distillation-boosted spiking neural network for object detection. In 2024 International Joint Conference on Neural Networks (IJCNN), pp.\ 1--9. IEEE, 2024

    work page 2024

  4. [4]

    A differentiable recurrent surface for asynchronous event-based data

    Cannici, M., Ciccone, M., Romanoni, A., and Matteucci, M. A differentiable recurrent surface for asynchronous event-based data. In Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part XX 16, pp.\ 136--152. Springer, 2020

    work page 2020

  5. [5]

    End-to-end object detection with transformers

    Carion, N., Massa, F., Synnaeve, G., Usunier, N., Kirillov, A., and Zagoruyko, S. End-to-end object detection with transformers. In European conference on computer vision, pp.\ 213--229. Springer, 2020

    2020

  6. [6]

    Optimised spatio-temporal descriptors for real-time fall detection: comparison of svm and adaboost based classification

    Charfi, I., Miteran, J., Dubois, J., Atri, M., and Tourki, R. Optimised spatio-temporal descriptors for real-time fall detection: comparison of svm and adaboost based classification. Journal of Electronic Imaging (JEI), 22 0 (4): 0 17, 2013

    work page 2013

  7. [7]

    Object detection with spiking neural networks on automotive event data

    Cordone, L., Miramond, B., and Thierion, P. Object detection with spiking neural networks on automotive event data. In 2022 International Joint Conference on Neural Networks (IJCNN), pp.\ 1--8. IEEE, 2022

    work page 2022

  8. [8]

    A large scale event-based detection dataset for automotive

    De Tournemire, P., Nitti, D., Perot, E., Migliore, D., and Sironi, A. A large scale event-based detection dataset for automotive. arXiv preprint arXiv:2001.08499, 2020

    arXiv 2001

Show all 52 references
  1. [9]

    Camera-based fall detection on real world data

    Debard, G., Karsmakers, P., Deschodt, M., Vlaeyen, E., Dejaeger, E., Milisen, K., Goedem \'e , T., Vanrumste, B., and Tuytelaars, T. Camera-based fall detection on real world data. In Outdoor and Large-Scale Real-World Scene Analysis: 15th International Workshop on Theoretical...

    2011

  2. [10]

    Sfod: Spiking fusion object detector

    Fan, Y., Zhang, W., Liu, C., Li, M., and Lu, W. Sfod: Spiking fusion object detector. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 17191--17200, 2024

    2024

  3. [11]

    Spikssd: Better extraction and fusion for object detection with spiking neuron networks

    Fan, Y., Liu, C., Li, M., and Zhang, W. Spikssd: Better extraction and fusion for object detection with spiking neuron networks. arXiv preprint arXiv:2501.15151, 2025

    2025 arXiv

  4. [12]

    Yolox: Exceeding yolo series in 2021

    Ge, Z. Yolox: Exceeding yolo series in 2021. arXiv preprint arXiv:2107.08430, 2021

    2021 arXiv

  5. [13]

    and Scaramuzza, D

    Gehrig, D. and Scaramuzza, D. Low-latency automotive vision with event cameras. Nature, 629 0 (8014): 0 1034--1040, 2024

    2024

  6. [14]

    and Scaramuzza, D

    Gehrig, M. and Scaramuzza, D. Recurrent vision transformers for object detection with event cameras. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp.\ 13884--13893, 2023

    2023

  7. [15]

    Joint a-snn: Joint training of artificial and spiking neural networks via self-distillation and weight factorization

    Guo, Y., Peng, W., Chen, Y., Zhang, L., Liu, X., Huang, X., and Ma, Z. Joint a-snn: Joint training of artificial and spiking neural networks via self-distillation and weight factorization. Pattern Recognition, 142: 0 109639, 2023

    2023

  8. [16]

    Spatio-temporal aggregation transformer for object detection with neuromorphic vision sensors

    Guo, Z., Gao, J., Ma, G., and Xu, J. Spatio-temporal aggregation transformer for object detection with neuromorphic vision sensors. IEEE Sensors Journal, 2024

    2024

  9. [17]

    Lt-snn: Self-adaptive spiking neural network for event-based classification and object detection

    Hasssan, A., Meng, J., and Seo, J.-s. Lt-snn: Self-adaptive spiking neural network for event-based classification and object detection. 2023

    2023

  10. [18]

    and Schmidhuber, J

    Hochreiter, S. and Schmidhuber, J. Long short-term memory. Neural computation, 9 0 (8): 0 1735--1780, 1997

    1997

  11. [19]

    Spiking deep residual networks

    Hu, Y., Tang, H., and Pan, G. Spiking deep residual networks. IEEE Transactions on Neural Networks and Learning Systems, 34 0 (8): 0 5200--5205, 2021

    2021

  12. [20]

    Fast-snn: Fast spiking neural network by converting quantized ann

    Hu, Y., Zheng, Q., Jiang, X., and Pan, G. Fast-snn: Fast spiking neural network by converting quantized ann. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45 0 (12): 0 14546--14562, 2023

    2023

  13. [21]

    Towards event-driven object detection with off-the-shelf deep learning

    Iacono, M., Weber, S., Glover, A., and Bartolozzi, C. Towards event-driven object detection with off-the-shelf deep learning. In 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp.\ 1--9. IEEE, 2018

    2018

  14. [22]

    and Ba, J

    Kingma, D. and Ba, J. Adam: A method for stochastic optimization. arXiv: Learning,arXiv: Learning, Dec 2014

    2014

  15. [23]

    Sodformer: Streaming object detection with transformer using events and frames

    Li, D., Li, J., and Tian, Y. Sodformer: Streaming object detection with transformer using events and frames. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    2023

  16. [24]

    Asynchronous spatio-temporal memory network for continuous event-based object detection

    Li, J., Li, J., Zhu, L., Xiang, X., Huang, T., and Tian, Y. Asynchronous spatio-temporal memory network for continuous event-based object detection. IEEE Transactions on Image Processing, 31: 0 2975--2987, 2022

    2022

  17. [25]

    Graph-based asynchronous event processing for rapid object recognition

    Li, Y., Zhou, H., Yang, B., Zhang, Y., Cui, Z., Bao, H., and Zhang, G. Graph-based asynchronous event processing for rapid object recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 934--943, 2021

    2021

  18. [26]

    Motion robust high-speed light-weighted object detection with event camera

    Liu, B., Xu, C., Yang, W., Yu, H., and Yu, L. Motion robust high-speed light-weighted object detection with event camera. IEEE Transactions on Instrumentation and Measurement, 72: 0 1--13, 2023

    2023

  19. [27]

    Short-term traffic flow prediction with conv-lstm

    Liu, Y., Zheng, H., Feng, X., and Chen, Z. Short-term traffic flow prediction with conv-lstm. In 2017 9th international conference on wireless communications and signal processing (WCSP), pp.\ 1--6. IEEE, 2017

    2017

  20. [28]

    Optical flow-guided 6dof object pose tracking with an event camera

    Liu, Z., Guan, B., Shang, Y., Liang, S., Yu, Z., and Yu, Q. Optical flow-guided 6dof object pose tracking with an event camera. In Proceedings of the 32nd ACM International Conference on Multimedia, pp.\ 6501--6509, 2024 a

    2024

  21. [29]

    Line-based 6-dof object pose estimation and tracking with an event camera

    Liu, Z., Guan, B., Shang, Y., Yu, Q., and Kneip, L. Line-based 6-dof object pose estimation and tracking with an event camera. IEEE Transactions on Image Processing, 2024 b

    2024

  22. [30]

    Stereo event-based, 6-dof pose tracking for uncooperative spacecraft

    Liu, Z., Guan, B., Shang, Y., Bian, Y., Sun, P., and Yu, Q. Stereo event-based, 6-dof pose tracking for uncooperative spacecraft. IEEE Transactions on Geoscience and Remote Sensing, 2025

    2025

  23. [31]

    Darwin3: a large-scale neuromorphic chip with a novel isa and on-chip learning

    Ma, D., Jin, X., Sun, S., Li, Y., Wu, X., Hu, Y., Yang, F., Tang, H., Zhu, X., Lin, P., et al. Darwin3: a large-scale neuromorphic chip with a novel isa and on-chip learning. National Science Review, 11 0 (5): 0 nwae102, 2024

    2024

  24. [32]

    Event-based asynchronous sparse convolutional networks

    Messikommer, N., Gehrig, D., Loquercio, A., and Scaramuzza, D. Event-based asynchronous sparse convolutional networks. In European Conference on Computer Vision, pp.\ 415--431. Springer, 2020

    2020

  25. [33]

    Mixed precision training

    Micikevicius, P., Narang, S., Alben, J., Diamos, G., Elsen, E., Garcia, D., Ginsburg, B., Houston, M., Kuchaiev, O., Venkatesh, G., et al. Mixed precision training. arXiv preprint arXiv:1710.03740, 2017

    2017 arXiv

  26. [34]

    Get: group event transformer for event-based vision

    Peng, Y., Zhang, Y., Xiong, Z., Sun, X., and Wu, F. Get: group event transformer for event-based vision. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 6038--6048, 2023

    2023

  27. [35]

    Learning to detect objects with a 1 megapixel event camera

    Perot, E., De Tournemire, P., Nitti, D., Masci, J., and Sironi, A. Learning to detect objects with a 1 megapixel event camera. Advances in Neural Information Processing Systems, 33: 0 16639--16652, 2020

    2020

  28. [36]

    A biomimetic frame-free event-driven image sensor

    Posch, C. A biomimetic frame-free event-driven image sensor. Focal-Plane Sensor-Processor Chips, pp.\ 125--149, 2011

    2011

  29. [37]

    ESIM : an open event camera simulator

    Rebecq, H., Gehrig, D., and Scaramuzza, D. ESIM : an open event camera simulator. Conf. on Robotics Learning (CoRL), October 2018

    2018

  30. [38]

    Aegnn: Asynchronous event-based graph neural networks

    Schaefer, S., Gehrig, D., and Scaramuzza, D. Aegnn: Asynchronous event-based graph neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp.\ 12371--12381, 2022

    2022

  31. [39]

    and Linares-Barranco, B

    Serrano-Gotarredona, T. and Linares-Barranco, B. A 128 128 1.5\ latency 4 mw asynchronous frame-free dynamic vision sensor using transimpedance preamplifiers. IEEE Journal of Solid-State Circuits, 48 0 (3): 0 827--838, 2013

    2013

  32. [40]

    Efficient spiking neural networks with sparse selective activation for continual learning

    Shen, J., Ni, W., Xu, Q., and Tang, H. Efficient spiking neural networks with sparse selective activation for continual learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pp.\ 611--619, 2024

    2024

  33. [41]

    Improving the sparse structure learning of spiking neural networks from the view of compression efficiency

    Shen, J., Xu, Q., Pan, G., and Chen, B. Improving the sparse structure learning of spiking neural networks from the view of compression efficiency. arXiv preprint arXiv:2502.13572, 2025

    2025 arXiv

  34. [42]

    Smith, L. N. and Topin, N. Super-convergence: Very fast training of neural networks using large learning rates. In Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications, May 2019. doi:10.1117/12.2520589. URL http://dx.doi.org/10.1117/12.2520589

    2019 doi

  35. [43]

    Deep directly-trained spiking neural networks for object detection

    Su, Q., Chou, Y., Hu, Y., Li, J., Mei, S., Zhang, Z., and Li, G. Deep directly-trained spiking neural networks for object detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 6555--6565, 2023

    2023

  36. [44]

    Maxvit: Multi-axis vision transformer

    Tu, Z., Talebi, H., Zhang, H., Yang, F., Milanfar, P., Bovik, A., and Li, Y. Maxvit: Multi-axis vision transformer. In European conference on computer vision, pp.\ 459--479. Springer, 2022

    2022

  37. [45]

    Eas-snn: End-to-end adaptive sampling and representation for event-based detection with recurrent spiking neural networks

    Wang, Z., Wang, Z., Li, H., Qin, L., Jiang, R., Ma, D., and Tang, H. Eas-snn: End-to-end adaptive sampling and representation for event-based detection with recurrent spiking neural networks. In European Conference on Computer Vision, pp.\ 310--328. Springer, 2024

    2024

  38. [46]

    Enhancing adaptive history reserving by spiking convolutional block attention module in recurrent neural networks

    Xu, Q., Gao, Y., Shen, J., Li, Y., Ran, X., Tang, H., and Pan, G. Enhancing adaptive history reserving by spiking convolutional block attention module in recurrent neural networks. Advances in Neural Information Processing Systems, 36: 0 58890--58901, 2023

    2023

  39. [47]

    Reversing structural pattern learning with biologically inspired knowledge distillation for spiking neural networks

    Xu, Q., Li, Y., Fang, X., Shen, J., Zhang, Q., and Pan, G. Reversing structural pattern learning with biologically inspired knowledge distillation for spiking neural networks. In Proceedings of the 32nd ACM International Conference on Multimedia, pp.\ 3431--3439, 2024 a

    2024

  40. [48]

    K., Pan, G., and Zhang, Q

    Xu, Q., Liu, S., Ran, X., Li, Y., Shen, J., Tang, H., Liu, J. K., Pan, G., and Zhang, Q. Robust sensory information reconstruction and classification with augmented spikes. IEEE Transactions on Neural Networks and Learning Systems, 2024 b

    2024

  41. [49]

    Spikingvit: A multiscale spiking vision transformer model for event-based object detection

    Yu, L., Chen, H., Wang, Z., Zhan, S., Shao, J., Liu, Q., and Xu, S. Spikingvit: A multiscale spiking vision transformer model for event-based object detection. IEEE Transactions on Cognitive and Developmental Systems, 17 0 (1): 0 130--146, 2025. doi:10.1109/TCDS.2024.3422873

    2025

  42. [50]

    Automotive object detection via learning sparse events by spiking neurons

    Zhang, H., Li, Y., Leng, L., Che, K., Liu, Q., Guo, Q., Liao, J., and Cheng, R. Automotive object detection via learning sparse events by spiking neurons. IEEE Transactions on Cognitive and Developmental Systems, 2024

    2024

  43. [51]

    Spikformer: When spiking neural network meets transformer

    Zhou, Z., Zhu, Y., He, C., Wang, Y., Yan, S., Tian, Y., and Yuan, L. Spikformer: When spiking neural network meets transformer. arXiv preprint arXiv:2209.15425, 2022

    2022 arXiv

  44. [52]

    From chaos comes order: Ordering event representations for object recognition and detection

    Zubi \'c , N., Gehrig, D., Gehrig, M., and Scaramuzza, D. From chaos comes order: Ordering event representations for object recognition and detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 12846--12856, 2023

    2023

Pith tools

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