Pith. sign in

REVIEW 4 major objections 5 minor 51 references

Sequence-SOD: Bio-inspired Sequence-aware Spiking ObjectDetection for Event Cameras

T0 review · 4 major / 5 minor · reviewed 2026-07-30 · grok-4.5

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

Pith's one-line read Keeping spiking neurons’ memory alive across event intervals improves detection without changing the network.

desk verdict Modest but clean protocol result: keeping SNN membrane state across multi-label event intervals beats single-interval reset on the same DenseNet-SSD, without closing the big accuracy gap. read the letter →

arxiv 2607.26703 v1 pith:KHDH4YH7 submitted 2026-07-29 cs.CV

Katharina Bendig , René Schuster , Didier Stricker This is my paper
classification cs.CV
keywords EventcameraSpikingneuralnetworkObjectdetectionEvent-basedvisionMembranepotentialSequence-awaretrainingAutomotiveperception
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

Event cameras send sparse, timed spikes when brightness changes, and spiking neural networks match that signal because each neuron holds a fading internal voltage. Most spiking object detectors still cut the stream into short windows, make one prediction, and wipe that internal state. This paper shows that training and running the same detector on longer sequences—leaving membrane voltages intact between windows and resetting only between independent sequences—raises detection scores on automotive event data. The gain comes from forcing the network to keep a useful running memory of recent evidence, not from a new architecture. A reader should care because continuous streams are how these sensors are actually used, and the method keeps the energy advantage of sparse spikes while using more of the temporal signal the camera already provides.

What carries the argument

Sequence-aware training and evaluation: events are accumulated into short intervals, discretized into time steps, fed sequentially to the spiking detector, and membrane potentials are carried forward across intervals inside a sequence so detection is driven by an evolving neural state instead of independently reset windows.

What would settle it

Train the identical architecture once with membrane reset after every interval and once with state preserved across multi-label sequences of the same total length and labels, on the same Gen1 splits with matched hyperparameters, and check whether the mAP gap remains.

Watch

Extended reading notes

Core claim

Under a fixed SSD-style Spiking DenseNet, sequence-aware training and evaluation that preserves membrane potentials across consecutive event intervals with multiple label timestamps improves Gen1 mAP from 23.38 for single-interval training to 25.30 without augmentation and 26.88 with event-data augmentation, while supporting a theoretical 40 Hz prediction rate from 25 ms steps.

Load-bearing premise

The measured gain is cleanly caused by keeping the membrane state alive across intervals, rather than by other fixed choices such as interval length, how sequences are cut from the data, or mismatches with the original single-interval baseline setup.

Editorial extensions

If this is right

  • Sequence-aware training is a complementary lever to designing new SNN detector architectures.
  • The same model can support a theoretical 40 Hz box stream from 25 ms steps by sliding accumulation over carried membrane state.
  • Sparse-spiking energy advantage versus an equivalent dense network is retained under the sequence setting.
  • Unlabeled intervals can still update membrane state and improve later labeled predictions.
  • Models trained on longer sequences hold up better when tested on continuous full-length streams than single-interval-trained models.

Reading between the lines

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

  • The same state-preservation recipe could be applied to newer spiking YOLO- or Transformer-style detectors without redesigning their heads.
  • Once sequences stretch beyond a few hundred milliseconds, learnable or input-dependent membrane decay may matter as much as sequence length itself.
  • Closing more of the gap to strong recurrent non-spiking event detectors likely needs truncated backprop through time at larger sequence scales.
  • The practical payoff is largest if the 40 Hz path is realized on neuromorphic hardware on true continuous streams, not only GPU-simulated steps.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes Sequence-SOD, a sequence-aware training and evaluation protocol for fully spiking SSD-style DenseNet object detectors on event cameras. Instead of processing isolated event intervals and resetting membrane potentials after each prediction, the method feeds consecutive 125 ms intervals (each discretized into T=5 steps of 25 ms), preserves PLIF membrane state across intervals within a sequence, and applies detection loss only on labeled intervals after averaging outputs over the internal steps. On Gen1, training with strain=5 raises mAP from 23.38 (strain=1) to 25.30 without augmentation and 26.88 with geometric event augmentation, with a claimed theoretical 40 Hz prediction rate and a large estimated energy advantage versus an equivalent ANN from measured spike rates.

Significance. If the gains are cleanly due to sequence-aware stateful training, the work is a useful complementary contribution: it isolates a training/evaluation protocol rather than a new backbone, and shows that keeping leaky membrane memory across multi-label event streams improves detection on a fixed fully spiking detector. The internal strain=1 vs strain=5 comparison, augmentation table, spike-rate figure, and CMOS-style energy accounting are concrete and reproducible in principle. The absolute accuracy remains well below strong ANN/hybrid event detectors, so impact is mainly as evidence that SNN detectors should be trained and evaluated as continuous-state systems, not as reset snapshot classifiers.

major comments (4)
  1. [Table 1; Method (Sequential Event Data Processing); Experiments (Sequence Length)] Table 1 attributes the mAP lift (23.38→25.30) to sequence-aware training with preserved membrane state, but strain=1 vs strain=5 jointly changes several factors: unrolled BPTT length, number of labeled supervision points, presence of unlabeled forward-only intervals, and whether state is reset between intervals. A load-bearing control is missing: train and test on the same multi-interval sequences while only toggling cross-interval reset vs preserve (and, ideally, a same-Δt matched reproduction of Cordone et al.). Without that, the central causal claim that evolving neural state—not longer optimization/unrolling—drives the gain is not isolated.
  2. [Tables 1–3; Implementation] All headline numbers appear to be single-run best checkpoints after 110 epochs with no seeds, error bars, or repeated trials (Tables 1–3). The no-augmentation gain is about +1.9 mAP and several augmentation deltas are <1 mAP. For a result of this magnitude, at least mean±std over multiple seeds (or a clear statement that variance was checked) is needed before the improvement can be treated as reliable.
  3. [Setup; Discussion and Limitations; Table 1] The protocol fixes Δt=125 ms and T=5 everywhere, builds sequences every 125 ms although the minimum label gap is 250 ms, and defines “full” test sequences by a 200-interval (25 s) gap heuristic, yet reports no sensitivity to these choices. The Discussion acknowledges this, but the 40 Hz claim and the sequence-construction grid are part of the method’s practical story; at minimum, one ablation over T or Δt (or intermediate strain) is needed to show the gain is not an artifact of this particular sampling alignment.
  4. [Table 3; Benchmark Comparison] Table 3 lists ODSNN at 18.9 mAP (Δt=100 ms) while the authors’ own single-interval baseline reaches 23.38 at Δt=125 ms under the “same” SSD Spiking DenseNet. The paper correctly emphasizes the internal control, but the benchmark table still invites a direct comparison that is unmatched in interval length and reimplementation details. Either align Δt/training recipe with Cordone et al. for the strain=1 row or clearly separate “reproduced baseline under our protocol” from the cited ODSNN number so the sequence-aware gain is not read against an understated external baseline.
minor comments (5)
  1. [Abstract; Introduction] Abstract and several places write “withevent-data” / missing spaces and hyphenation inconsistencies (“hightemporal”, “Articial Neural Network”). A full copy-edit pass is needed.
  2. [Fig. 2] Figure 2 caption refers to “strain=5” with a missing subscript formatting (“strain =5”), and the schematic would benefit from explicitly marking which intervals receive loss versus state-only forward passes.
  3. [Eq. (1); Network Architecture] Equation (1) uses Sout[t−1]θ for reset; briefly state whether a soft/hard reset and subtractive vs zero reset are used in the PLIF implementation, since this affects cross-interval memory.
  4. [Energy Consumption; Table 4; Fig. 4] Energy section: clarify whether spike rates in Fig. 4/Table 4 are measured under strain=5 full-sequence evaluation or stest=1, and whether BN-folded inference is assumed in the FLOP counts.
  5. [Related Work] Related Work is thorough but long relative to the empirical novelty; tightening hybrid vs fully-spiking distinctions would help the reader reach the contribution faster.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical train/test comparison on held-out Gen1 mAP under a fixed architecture.

full rationale

Sequence-SOD’s central claim is an empirical performance lift (mAP 23.38 → 25.30/26.88) from sequence-aware training that preserves membrane potentials across multi-label event intervals, versus single-interval reset training on the same SSD-style Spiking DenseNet. That claim is evaluated on the external Gen1 Automotive Detection test split with standard COCO-style mAP; mAP is not defined from the training loss or from any fitted scalar that forces the reported gain. Energy figures use measured spike rates scaled by standard 45 nm CMOS MAC/AC constants and are presented as estimates, not first-principles predictions. Self-citations (e.g., the authors’ prior ICPRAM note) appear only as related-work context and do not underwrite uniqueness, forbid alternatives, or supply the measured mAP delta. Concerns about unmatched Δt versus Cordone et al., missing ablations, and single-run checkpoints are experimental-validity issues, not circular reductions of outputs to inputs. No step in the paper’s chain equates a claimed prediction to a fitted input or a self-definitional identity.

Assumptions & free parameters 6 free parameters · 6 assumptions · 1 invented entities

Load-bearing content is mostly standard event-vision and SNN practice plus experimental protocol choices. The central claim rests on domain assumptions about discretized event histograms, PLIF dynamics, surrogate-gradient training, Gen1 label timing, and hand-chosen sequence/interval hyperparameters rather than on newly postulated physical entities.

free parameters (6)
  • strain (training sequence length) = 5 (main); 1 (baseline)
    Primary experimental knob; main results use strain=5 vs 1 with no sweep over intermediate lengths.
  • Δt event interval duration = 125 ms
    Fixed accumulation window for histograms; chosen to align with Gen1 annotation spacing, not swept.
  • T internal time steps per interval = 5 (25 ms steps)
    Discretization of each interval into SNN steps; fixed by appeal to Cordone et al., not ablated here.
  • Learning rate, batch size, epochs, weight decay = lr 4e-5, bs 14, 110 epochs, wd 1e-4
    Standard training hyperparameters that affect the reported checkpoint mAP; batch size constrained by BPTT memory.
  • Full-sequence split gap threshold = 200 intervals / 25 s
    Heuristic used to define stest=full by splitting when label gaps exceed 200 intervals (25 s).
  • CMOS energy constants EMULT and EAC = 3.7 pJ / 0.9 pJ
    External 45 nm constants used to convert FLOPs/spike rates into mJ; standard but choice directly scales Table 4 ratios.
assumptions (6)
  • domain assumption Discretized PLIF/LIF membrane update with learnable leak β is an adequate temporal memory model for event detection.
    Method section adopts PLIF neurons and treats membrane potentials as the retained state across intervals.
  • domain assumption Surrogate-gradient BPTT through unrolled spike functions yields meaningful credit assignment across multi-interval sequences.
    Training relies on direct SNN training with surrogate gradients over sequence time.
  • domain assumption Event histograms with separate polarity channels and T≥5 without extra time bins are sufficient input representations.
    Eqs. 2–3 and citation to Cordone et al. justify not using finer binning.
  • domain assumption Gen1 RGB-derived boxes at 1–4 Hz, after discarding small boxes, are valid supervision for continuous event-stream detection.
    Setup filters boxes and aligns 125 ms intervals to intermittent labels; unlabeled intervals get state propagation only.
  • domain assumption After training, batch-norm parameters can be folded so inference is treated as a pure SNN for energy comparison.
    Network Architecture section claims mathematical equivalence post-absorption into convolutions.
  • standard math Standard floating-point MAC/AC energy model on 45 nm CMOS estimates relative SNN vs ANN inference cost from spike rates.
    Energy section uses conventional FLOPs counting and Horowitz constants; not a neuromorphic-chip measurement.
invented entities (1)
  • Sequence-SOD (sequence-aware training/evaluation protocol) independent evidence
    purpose: Name the practice of training/evaluating a fully spiking SSD DenseNet on multi-interval sequences while preserving membrane state within a sequence.
    Not a new physical entity; a methodological wrapper around existing SNN detector components. No independent evidence needed beyond the empirical tables.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sequence-SOD: Bio-inspired Sequence-aware Spiking ObjectDetection for Event Cameras." pith.science (2026). https://pith.science/paper/KHDH4YH7

@misc{pith2026260726703,
  author       = {Pith},
  title        = {Pith review of: Sequence-SOD: Bio-inspired Sequence-aware Spiking ObjectDetection for Event Cameras},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KHDH4YH7}},
  note         = {Machine review of arXiv:2607.26703}
}
read the original abstract

Event cameras follow a retina-inspired sensing principle, reporting local intensity changes asynchronously with hightemporal resolution and a wide dynamic range. Spiking Neural Networks (SNNs) complement these sparse event streams through brain-inspired dynamics, using sparse spikes and leaky membrane potentials to integrate information over time. However, many SNN object detectors process isolated event intervals with a single label and reset the network state after each prediction, thereby underusing temporal information in continuous event streams. We introduce Sequence-SOD, a sequence-aware SNN object detector that processes extended event sequences containing labels at multiple time points. Events are accumulated into short intervals, discretized into temporal steps, and fed sequentially to an SSD-style Spiking DenseNet while preserving membrane potentials across intervals within a sequence, so that detection is driven by an evolving neural state instead of independently reset input windows. On the Gen1 Automotive Detection Dataset, sequence-aware training improves mAP from 23.38 for single-interval training to 25.30 without augmentation and to 26.88 withevent-data augmentation. The model achieves a theoretical prediction frequency of 40 Hz. Training and evaluating SNN object detectors on extended event sequences improves their ability to exploit temporal cues while preserving the energy-efficiency benefits of sparse spiking computation. The results highlight sequence-aware training as a complementary direction to architectural improvements for event-based SNN detection.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

51 extracted references · 3 linked inside Pith

  1. [1]

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

    Lichtsteiner P, Posch C, Delbruck T. A 128 × 128 120 dB 15 µs Latency Asynchronous Temporal Contrast Vision Sensor. Journal of Solid-State Circuits. 2008

    2008

  2. [2]

    Event-based vision: A survey

    Gallego G, Delbrück T, Orchard G, Bartolozzi C, Taba B, Censi A. Event-based vision: A survey. Transactions on Pattern Analy- sis and Machine Intelligence (TPAMI). 2022

    2022

  3. [3]

    Spiking Neuron Models: Single Neu - rons, Populations

    Gerstner W, Kistler WM. Spiking Neuron Models: Single Neu - rons, Populations. Plasticity: Cambridge University Press; 2002

    2002

  4. [4]

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

    Maass W. Networks of spiking neurons: The third generation of neural network models. Neural networks. 1997. Table 4 Energy consumption of an ANN and a SNN Method Network F LOP sM ult F LOP sAc Rl E (mJ) EAN N EM ethod ANN Backbone 2295.49 M 2298.21 M − 10.56 1× Heads 351.32 M 351.32 M − 1.61 1× Overall 2646.81 M 2649.53 M − 12.17 1× SNN Backbone − 2298.21...

    1997

  5. [5]

    Deep learning with spiking neurons: Opportu- nities and challenges

    Pfeiffer M, Pfeil T. Deep learning with spiking neurons: Opportu- nities and challenges. Frontiers in Neuroscience. 2018

    2018

  6. [6]

    Memory and information processing in neu - romorphic systems

    Indiveri G, Liu SC. Memory and information processing in neu - romorphic systems. In: Proceedings of the IEEE. 2015

    2015

  7. [7]

    Object detection with spik- ing neural networks on automotive event data

    Cordone L, Miramond B, Thierion P. Object detection with spik- ing neural networks on automotive event data. In: International Joint Conference on Neural Networks (IJCNN). 2022

    2022

  8. [8]

    On the computational power of circuits of spiking neurons

    Maass W, Markram H. On the computational power of circuits of spiking neurons. Journal of Computer and System Sciences (JCSS). 2004

    2004

Show all 51 references
  1. [9]

    Rethinking the performance comparison between SNNS and ANNS

    Deng L, Wu Y , Hu X, Liang L, Ding Y , Li G. Rethinking the performance comparison between SNNS and ANNS. Neural Net- works. 2019

    2019

  2. [10]

    Direct training for spiking neural networks: Faster, larger, better

    Wu Y , Deng L, Li G, Zhu J, Xie Y , Shi LP. Direct training for spiking neural networks: Faster, larger, better. In: AAAI Confer- ence on Artificial Intelligence (AAAI). 2019

    2019

  3. [11]

    Spike-timing dependent plasticity and mutual information maximization for a spiking neuron model

    Toyoizumi T, Jp Pfister, Aihara K, Gerstner W. Spike-timing dependent plasticity and mutual information maximization for a spiking neuron model. In: Advances in Neural Information Pro - cessing Systems (NeurIPS). 2004

    2004

  4. [12]

    Conversion of continuous-valued deep networks to efficient event-driven net- works for image classification

    Rueckauer B, Lungu IA, Hu Y , Pfeiffer M, Liu SC. Conversion of continuous-valued deep networks to efficient event-driven net- works for image classification. Frontiers in Neuroscience. 2017

    2017

  5. [13]

    On the future of training spik - ing neural networks

    Bendig K, Schuster R, Stricker D. On the future of training spik - ing neural networks. In: International Conference on Pattern Rec- ognition Applications and Methods (ICPRAM). 2023

    2023

  6. [14]

    Surrogate gradient learning in spiking neural networks: Bringing the power of gradient-based optimization to spiking neural networks

    Neftci E, Mostafa H, Zenke F. Surrogate gradient learning in spiking neural networks: Bringing the power of gradient-based optimization to spiking neural networks. IEEE Signal Processing Magazine. 2019

    2019

  7. [15]

    Moving object detection and tracking by event frame from neuromorphic vision sensors

    Zhao J, Cai JiS, Zeng Z, Wang Y , Y . Moving object detection and tracking by event frame from neuromorphic vision sensors. Bio - mimetics. 2022

    2022

  8. [16]

    High-temporal-resolution object detection and tracking using images and events

    Shair ZE, Rawashdeh SA. High-temporal-resolution object detection and tracking using images and events. Journal of Imag- ing. 2022

    2022

  9. [17]

    Event-based moving object detection and tracking

    Mitrokhin A, Fermuller C, Parameshwara C, Aloimonos Y . Event-based moving object detection and tracking. In: Interna - tional Conference on Intelligent Robots and Systems (IROS). 2018

    2018

  10. [18]

    Asynchro - nous convolutional networks for object detection in neuromor - phic cameras

    Cannici M, Ciccone M, Romanoni A, Matteucci M. Asynchro - nous convolutional networks for object detection in neuromor - phic cameras. In: Conference on Computer Vision and Pattern Recognition Workshops. CVPRW; 2019

    2019

  11. [19]

    Learning to detect objects with a 1 megapixel event camera

    Perot E, de Tournemire P, Nitti D, Masci J, Sironi A. Learning to detect objects with a 1 megapixel event camera. In: Neural Infor- mation Processing Systems (NeurIPS); 2020

    2020

  12. [20]

    Asynchronous spa- tio-temporal memory network for continuous event-based object detection

    Li J, Li J, Zhu L, Xiang X, Huang T, Tian Y . Asynchronous spa- tio-temporal memory network for continuous event-based object detection. Transactions on Image Processing (TIP). 2022

    2022

  13. [21]

    Sodformer: Streaming object detection with transformer using events and frames

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

    2023

  14. [22]

    Get: Group event trans- former for event-based vision

    Peng Y , Zhang Y , Xiong Z, Sun X, Wu F. Get: Group event trans- former for event-based vision. In: International Conference on Computer Vision (ICCV). 2023

    2023

  15. [23]

    State space models for event cameras

    Zubic N, Gehrig M, Scaramuzza D. State space models for event cameras. In: Conference on Computer Vision and Pattern Recog- nition (CVPR). 2024

    2024

  16. [24]

    Recurrent vision transformers for object detection with event cameras

    Gehrig M, Scaramuzza D. Recurrent vision transformers for object detection with event cameras. In: Conference on Computer Vision and Pattern Recognition (CVPR). 2023

    2023

  17. [25]

    Scene adaptive sparse trans- former for event-based object detection

    Peng Y , Li H, Zhang Y , Sun X, Wu F. Scene adaptive sparse trans- former for event-based object detection. In: Conference on Com- puter Vision and Pattern Recognition (CVPR). 2024

    2024

  18. [26]

    Sparse mamba for event-based object detection

    Yang N, Wang Y , Liu Z, Li M, Zhao AnY , Smamba X. Sparse mamba for event-based object detection. In: AAAI Conference on Artificial Intelligence. 2025

    2025

  19. [27]

    Evrt-detr: Latent space adaptation of image detectors for event-based vision

    Torbunov D, Ren Y , Ghose A, Dim O, Cui Y . Evrt-detr: Latent space adaptation of image detectors for event-based vision. In: International Conference on Computer Vision (ICCV). 2025

    2025

  20. [28]

    Detrs beat yolos on real-time object detection

    Zhao Y , Lv W, Xu S, Wei J, Wang G, Dang Q. Detrs beat yolos on real-time object detection. In: Conference on Computer Vision and Pattern Recognition (CVPR). 2024

    2024

  21. [29]

    Deep directly-trained spiking neural networks for object detection

    Su Q, Chou Y , Hu Y , Li J, Mei S, Zhang Z. Deep directly-trained spiking neural networks for object detection. In: International Conference on Computer Vision (ICCV). 2023

    2023

  22. [30]

    SFOD: Spiking fusion object detector

    Fan Y , Zhang W, Liu C, Li M, Lu W. SFOD: Spiking fusion object detector. In: Conference on Computer Vision and Pattern Recog- nition (CVPR); 2024

    2024

  23. [31]

    Spik - ing centerNet: A distillation-boosted spiking neural network for object detection

    Bodden L, Schwaiger F, Ha DB, Kreuzberg L, Behnke S. Spik - ing centerNet: A distillation-boosted spiking neural network for object detection. In: International Joint Conference on Neural Networks (IJCNN). 2024

    2024

  24. [32]

    SpikingViT: A multi-scale spiking vision transformer model for event-based object detection

    Yu L, et al. SpikingViT: A multi-scale spiking vision transformer model for event-based object detection. Transactions on Cogni - tive and Developmental Systems. 2024

    2024

  25. [33]

    Automotive object detection via learning sparse events by spiking neurons

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

    2024

  26. [34]

    A hybrid SNN-ANN network for event-based object detection with spatial and temporal atten - tion

    Ahmed SH, Finkbeiner J, Neftci E. A hybrid SNN-ANN network for event-based object detection with spatial and temporal atten - tion. 2024. arXiv:2403.10173 arXiv preprint

    2024 arXiv

  27. [35]

    Hybrid spiking vision transformer for object detection with event cameras

    Xu Q, Deng J, Shen J, Chen B, Tang H, Pan G. Hybrid spiking vision transformer for object detection with event cameras. In: International Conference on Machine Learning (ICML). 2025

    2025

  28. [36]

    Integer-valued training and spike-driven inference spiking neural network for high-perfor - mance and energy-efficient object detection

    Luo X, Yao M, Chou Y , Xu B, Li G. Integer-valued training and spike-driven inference spiking neural network for high-perfor - mance and energy-efficient object detection. In: European Con- ference on Computer Vision (ECCV). 2024

    2024

  29. [37]

    Trainable spiking-yolo for low-latency and high-performance object detec- tion

    Yuan M, Zhang C, Wang Z, Liu H, Pan G, Tang H. Trainable spiking-yolo for low-latency and high-performance object detec- tion. Neural Networks. 2024

    2024

  30. [38]

    Object detection with spiking neural networks on event videos for driving sce - narios

    Gao S, Tao Y , Gruel A, Saïghi S, Chang CH. Object detection with spiking neural networks on event videos for driving sce - narios. Transactions on Cognitive and Developmental Systems. 2026

    2026

  31. [39]

    SSD: Single shot multibox detector

    Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu CY , et al. SSD: Single shot multibox detector. In: European Conference on Computer Vision (ECCV); 2016

    2016

  32. [40]

    Densely con - nected convolutional networks

    Huang G, Liu Z, Maaten LVD, Weinberger KQ. Densely con - nected convolutional networks. In: Conference on Computer Vision and Pattern Recognition (CVPR); 2017

    2017

  33. [41]

    Incorpo- rating learnable membrane time constant to enhance learning of spiking neural networks

    Fang W, Yu Z, Chen Y , Masquelier T, Huang T, Tian Y . Incorpo- rating learnable membrane time constant to enhance learning of spiking neural networks. In: International Conference on Com - puter Vision (ICCV). 2021

    2021

  34. [42]

    Quanti- zation and training of neural networks for efficient integer-arith- metic-only inference

    Jacob B, Kligys S, Chen B, Zhu M, Tang M, Howard A. Quanti- zation and training of neural networks for efficient integer-arith- metic-only inference. In: Conference on Computer Vision and Pattern Recognition (CVPR). 2018

    2018

  35. [43]

    A Large Scale Event-based Detection Dataset for Automotive

    de Tournemire P, Nitti D, Perot E, Migliore D, Sironi A. A Large Scale Event-based Detection Dataset for Automotive. 2020. arXiv:2001.08499

    2020 arXiv

  36. [44]

    Focal loss for dense object detection

    Lin TY , Goyal P, Girshick R, He K, Dollar P. Focal loss for dense object detection. In: Transactions on Pattern Analysis and Machine Intelligence (T-PAMI). 2020

    2020

  37. [45]

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

    Iacono M, Weber S, Glover A, Bartolozzi C. Towards event-driven object detection with off-the-shelf deep learning. In: International Conference on Intelligent Robots and Systems (IROS). 2018. 1 3 Page 13 of 13 91 Cognitive Computation (2026) 18:91

    2018

  38. [46]

    Pseudo-labels for supervised learning on dynamic vision sensor data, applied to object detection under ego-motion

    Chen Nfy. Pseudo-labels for supervised learning on dynamic vision sensor data, applied to object detection under ego-motion. In: Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). 2018

    2018

  39. [47]

    YOLOv3: An incremental improvement

    Redmon J, Farhadi A. YOLOv3: An incremental improvement

  40. [48]

    Mixed frame-/event-driven fast pedestrian detection

    Jiang Z, Xia P, Huang K, Stechele W, Chen G, Bing Z. Mixed frame-/event-driven fast pedestrian detection. In: International Conference on Robotics and Automation (ICRA). 2019

    2019

  41. [49]

    YOLOX: Exceeding YOLO series in 2021

    Ge Z, Liu S, Wang F, Li Z, Sun J. YOLOX: Exceeding YOLO series in 2021. 2021. arXiv:2107.08430

    2021 arXiv

  42. [50]

    Beyond classification: directly train- ing spiking neural networks for semantic segmentation

    Kim Y , Chough J, Panda P . Beyond classification: directly train- ing spiking neural networks for semantic segmentation. Neuro - morphic Computing and Engineering (NCE). 2022

    2022

  43. [51]

    1.1 Computing’s energy problem (and what we can do about it)

    Horowitz M. 1.1 Computing’s energy problem (and what we can do about it). In: International Solid-State Circuits Conference Digest of Technical Papers (ISSCC); 2014. Publisher's Note Springer Nature remains neutral with regard to juris- dictional claims in published maps and i...

    2014

Pith tools

Reviewed July 30, 2026 · model on record in the stance chip above.