Research map / Architectures and efficiency / CNN design
A Voxel Graph CNN for Object Classification With Event Cameras
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 MFRL to process adjacent and distant neighbors separately instead of shared-parameter SFRL (ablation) | EV-VGCNN · N-MNIST, N-Caltech101, N-CARS, CIFAR10-DVS, ASL-DVS | accuracy: 0.992 -> 0.994; 0.737 -> 0.748; 0.944 -> 0.953; 0.652 -> 0.670; 0.962 -> 0.983 | Helped |
| Use event-voxel rather than point-wise vertices (ablation) | EV-VGCNN · N-Caltech101 | accuracy: 0.565 -> 0.748 | Helped |
| Increase point-wise graph vertices from 2048 to 4096 (ablation) | EV-VGCNN with original-event vertices · N-Caltech101 | accuracy: 0.565 -> 0.601 | Helped |
| Increase point-wise graph vertices from 4096 to 8192 (ablation) | EV-VGCNN with original-event vertices · N-Caltech101 | accuracy: 0.601 -> 0.619 | Helped |
| Increase distant neighbors from 10 to 20 while keeping adjacent neighbors at 10 (ablation) | EV-VGCNN · N-Caltech101 | accuracy: 0.742 -> 0.751 | Helped |
| Increase adjacent neighbors from 10 to 20 while keeping distant neighbors at 15 (ablation) | EV-VGCNN · N-Caltech101 | accuracy: 0.748 -> 0.740 | Hurt |
| Use 10 adjacent and 15 distant neighbors rather than 5 adjacent and 15 distant neighbors (ablation) | EV-VGCNN · N-Caltech101 | accuracy: 0.737 -> 0.748 | Helped |
| Use 10 adjacent and 15 distant neighbors rather than 5 adjacent and 20 distant neighbors (ablation) | EV-VGCNN · N-Caltech101 | accuracy: 0.743 -> 0.748 | Helped |
| Use 10 adjacent and 15 distant neighbors rather than 20 adjacent and 5 distant neighbors (ablation) | EV-VGCNN · N-Caltech101 | accuracy: 0.730 -> 0.748 | Helped |
| Use event-voxel graph construction rather than point-wise graph construction (ablation) | EV-VGCNN · N-Caltech101 | accuracy: 0.619 -> 0.748 | Helped |
| Use event-voxel graph construction rather than point-wise graph construction (ablation) | EV-VGCNN · N-Caltech101 | GFLOPs: 0.63 -> 0.70 | Hurt |
Closest papers by meaning
- A Dynamic GCN with Cross-Representation Distillation for Event-Based Learning (AAAI 2024 · similarity 0.87)
- V2V: Scaling Event-Based Vision through Efficient Video-to-Voxel Simulation (NeurIPS 2025 · similarity 0.86)
- A Chaotic Dynamics Framework Inspired by Dorsal Stream for Event Signal Processing (ICML 2025 · similarity 0.85)
- Graph Neural Network Combining Event Stream and Periodic Aggregation for Low-Latency Event-based Vision (CVPR 2025 · similarity 0.85)
- E2HQV: High-Quality Video Generation from Event Camera via Theory-Inspired Model-Aided Deep Learning (AAAI 2024 · similarity 0.85)
- PASS: Path-selective State Space Model for Event-based Recognition (NeurIPS 2025 · similarity 0.84)
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.