Research map / Detection and segmentation / CNN features
Event-based Asynchronous Sparse Convolutional Networks
Nico Messikommer, Daniel Gehrig, Antonio Loquercio, Davide Scaramuzza
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 asynchronous sparse conv instead of dense conv (histogram) | VGG13 · N-Caltech101 recognition | MFLOP per event: 1621 -> 202 | Improved |
| use asynchronous sparse conv instead of dense conv (histogram) | VGG13 · N-Cars recognition | MFLOP per event: 321 -> 21.5 | Improved |
| use asynchronous sparse conv instead of dense conv (histogram) | VGG13 · N-Caltech101 recognition | accuracy: 0.761 -> 0.745 | Hurt |
| use asynchronous sparse conv instead of dense conv (queue) | VGG13 · N-Cars recognition | accuracy: 0.936 -> 0.936 | No clear effect |
| use asynchronous sparse conv instead of dense conv (histogram) | YOLO head · N-Caltech101 detection | mAP: 0.623 -> 0.643 | Improved |
| use asynchronous sparse conv instead of dense conv (histogram) | YOLO head · Gen1 Automotive detection | mAP: 0.145 -> 0.129 | Hurt |
| use asynchronous sparse conv instead of dense conv (queue) | YOLO head · Gen1 Automotive detection | mAP: 0.149 -> 0.119 | Hurt |
| use asynchronous sparse conv instead of dense conv (histogram) | YOLO head · Gen1 Automotive detection | MFLOP: 2098 -> 205 | Improved |
| use 25000 events instead of 50000 for the sparse network (ablation) | Sparse VGG13 · N-Caltech101 recognition | accuracy: 0.730 -> 0.745 | Improved |
| use 25000 events instead of 5000 for the sparse network (ablation) | Sparse VGG13 · N-Caltech101 recognition | accuracy: 0.734 -> 0.745 | Improved |
| use 25000 events instead of 64 for the sparse network (ablation) | Sparse VGG13 · N-Caltech101 recognition | accuracy: 0.247 -> 0.745 | Improved |
| use 25000 events instead of 50000 for the dense network (ablation) | Dense VGG13 · N-Caltech101 recognition | accuracy: 0.761 -> 0.766 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HATS — HATS: Histograms of Averaged Time Surfaces for Robust Event-Based Object Classification (CVPR 2018) · better on 2 of 2 shared results
- DART — DART: Distribution Aware Retinal Transform for Event-based Cameras (TPAMI 2019) · better on 1 of 1 shared result
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
- EventNet: Asynchronous Recursive Event Processing (CVPR 2019 · similarity 0.93)
- End-to-End Learning of Representations for Asynchronous Event-Based Data (ICCV 2019 · similarity 0.92)
- "FARSE-CNN: Fully Asynchronous, Recurrent and Sparse Event-Based CNN" (ECCV 2024 · similarity 0.92)
- Leveraging Asynchronous Spiking Neural Networks for Ultra Efficient Event-Based Visual Processing (AAAI 2025 · similarity 0.90)
- Graph Neural Network Combining Event Stream and Periodic Aggregation for Low-Latency Event-based Vision (CVPR 2025 · similarity 0.89)
- ALERT-Transformer: Bridging Asynchronous and Synchronous Machine Learning for Real-Time Event-based Spatio-Temporal Data (ICML 2024 · similarity 0.88)
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.
