Research map / Detection and segmentation / CNN features
Deep Directly-Trained Spiking Neural Networks for Object Detection
Qiaoyi Su, Yuhong Chou, Yifan Hu, Jianing Li, Shijie Mei, Ziyang Zhang, Guoqi Li
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use directly-trained EMS-YOLO instead of ANN-SNN conversion | SNN object detection · COCO2017 val | mAP@0.5: 0.454 (Spike Calibration, T=512) -> 0.501 (EMS-YOLO, T=4) | Improved |
| use EMS-Res10 SNN instead of same-structure ANN | Gen1 detection · Gen1 | mAP@0.5:0.95: 0.247 -> 0.267 | Improved |
| replace MS/SEW residual block with full-spike EMS block (ablation) | Res18 SNN · Gen1 | energy efficiency vs ANN: 2.43x (MS) -> 4.91x (EMS) | Improved |
| replace MS/SEW residual block with EMS block (ablation) | Res18 SNN · Gen1 | mAP@0.5: 0.560/0.561 -> 0.565 | No clear effect |
| replace MS/SEW block with EMS block (ablation) | Res18 SNN · COCO2017 (reduced channels) | mAP@0.5: 0.345 -> 0.362 | Improved |
| increase depth of EMS-ResNet from 10 to 34 layers (ablation) | EMS-YOLO · Gen1 | mAP@0.5: 0.547 -> 0.590 | Improved |
| increase depth of EMS-ResNet from 10 to 34 layers (ablation) | EMS-YOLO · COCO2017 (T=1, 50 epochs) | mAP@0.5: 0.203 -> 0.335 | Improved |
| increase time steps from 1 to 7 (ablation) | EMS-Res10 · COCO2017 | mAP@0.5: 0.328 -> 0.383 | Improved |
| use spiking rate-coded detection layer instead of membrane-potential non-spiking layer (ablation) | EMS-Res18 · COCO2017 (T=3) | mAP@0.5: 0.318 -> 0.305 | Hurt |
| use spiking rate-coded detection layer instead of membrane-potential non-spiking layer (ablation) | EMS-Res18 · Gen1 (T=5) | mAP@0.5: 0.566 -> 0.565 | No clear effect |
Closest papers by meaning
- Brain-Inspired Spiking Neural Networks for Energy-Efficient Object Detection (CVPR 2025 · similarity 0.93)
- SpikingYOLOX: Improved YOLOX Object Detection with Fast Fourier Convolution and Spiking Neural Networks (AAAI 2025 · similarity 0.92)
- SFOD: Spiking Fusion Object Detector (CVPR 2024 · similarity 0.91)
- Efficient Event-Based Object Detection: A Hybrid Neural Network with Spatial and Temporal Attention (CVPR 2025 · similarity 0.90)
- CREST: An Efficient Conjointly-trained Spike-driven Framework for Event-based Object Detection Exploiting Spatiotemporal Dynamics (AAAI 2025 · similarity 0.90)
- Inference-Scale Complexity in ANN-SNN Conversion for High-Performance and Low-Power Applications (CVPR 2025 · similarity 0.89)
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.
