AI Paper Atlas

Research map / Detection and segmentation / CNN features

Spike-driven Discrete Aggregation for Event-based Object Detection

Huaning Li, Ziming Wang, Runhao Jiang, Yan Rui, Huajin Tang

CVPR 2026 · Detection and segmentation › CNN features · approach: Spiking event-camera detection

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace spiking continuous aggregation with spiking discrete aggregation (SDA) at same 20 ms resolution (ablation)spiking YOLOX-M · Gen1mAP50:95: 41.2 -> 42.8Improved
fuse SAS coarse-timescale branch by plain addition without soft mask (ablation)SDA · Gen1mAP50:95: 42.8 -> 42.8No clear effect
add soft mask from SAS membrane potentials into SDA update (ablation)SDA-MTF · Gen1mAP50:95: 42.8 -> 43.4Improved
add gated recurrent spiking neuron (GRSN) (ablation)SDA-MTF · Gen1mAP50:95: 40.7 -> 43.4Improved
use SDA instead of Event Count, Time Surface or Voxel Cube representationspiking YOLOX-M · Gen1mAP50:95: 32.3 / 37.3 / 32.6 -> 42.8Improved
add SDA to Event Count inputYOLOX-S (ANN) · Gen1mAP50:95: 37.4 -> 45.8Improved
add MTF on top of SDA for ANN YOLOX-SYOLOX-S + SDA · Gen1mAP50:95: 45.8 -> 43.3Hurt
add SDA to Event Count inputPVT-S (ANN) · Gen1mAP50:95: 42.4 -> 45.4Improved
add SDA-MTF to Event Count inputRVT-B (ANN) · Gen1mAP50:95: 47.2 -> 48.7Improved
add MTF on top of SDA (ablation)SDA (spiking YOLOX-M) · Gen1 speed levels Lv3/Lv4mAP50:95: 47.9 -> 48.8 (Lv3); 42.5 -> 44.0 (Lv4)Improved
use SDA instead of spiking continuous aggregation under noisy events (ablation)spiking detector · Gen1 with added natural/random noisemAP50:95 drop ratioImproved
replace EAS-SNN representation with SDA at identical detection architectureEAS-SNN (spiking YOLOX-M) · Gen1mAP50:95: 37.5 -> 42.8Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy