Research map / Architectures and efficiency / Spiking networks
Event Transformer$^+$. A Multi-Purpose Solution for Efficient Event Data Processing
Alberto Sabater, Luis Montesano, Ana C. Murillo
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 |
|---|---|---|---|
| replace EvT with EvT+ (finer FIFO patch representation and cross-attention backbone) | EvT+ · DVS128 Gesture (10 classes) | accuracy: 98.46 -> 99.24 | Improved |
| replace EvT with EvT+ | EvT+ · DVS128 Gesture (11 classes) | accuracy: 96.20 -> 97.57 | Improved |
| replace EvT with EvT+ | EvT+ · SL-Animals-DVS (3 sets) | accuracy: 87.45 -> 92.34 | Improved |
| replace EvT with EvT+ | EvT+ · SL-Animals-DVS (4 sets) | accuracy: 88.12 -> 94.39 | Improved |
| use 3D-CNN with offline voting instead of online EvT+ | DVS128 Gesture · DVS128 Gesture (11 classes) | accuracy: 97.57 -> 99.62 | Improved |
| add grayscale image input to event-only EvT+ | EvT+ · MVSEC outdoor night 1 (30 m cutoff) | mean absolute depth error (m): 2.96 -> 2.88 | Improved |
| add grayscale image input to event-only EvT+ | EvT+ · MVSEC outdoor day 1 (30 m cutoff) | mean absolute depth error (m): 2.32 -> 2.36 | No clear effect |
| add grayscale image input to event-only EvT+ | EvT+ · MVSEC outdoor night 3 (30 m cutoff) | mean absolute depth error (m): 2.80 -> 2.77 | Improved |
| use EvT+ instead of RAM Net with pre-training | MVSEC depth · MVSEC outdoor night 1 (30 m cutoff) | mean absolute depth error (m): 3.82 -> 2.88 | Improved |
| use bigger patch size and fewer latent vectors in EvT+ than EvT | EvT+ · DVS128 Gesture (efficiency) | FLOPs per time-window (G): 0.08 -> 0.03 | Improved |
| filter out non-activated patches (exploit sparsity) (ablation) | EvT+ · DVS128 Gesture (efficiency) | FLOPs per time-window (G): 0.07 -> 0.03 | Improved |
| filter out non-activated patches (exploit sparsity) (ablation) | EvT+ · MVSEC (efficiency) | FLOPs per time-window (G): 3.96 -> 2.94 | Improved |
Closest papers by meaning
- Motion-adaptive Transformer for Event-based Image Deblurring (AAAI 2025 · similarity 0.88)
- Scene Adaptive Sparse Transformer for Event-based Object Detection (CVPR 2024 · similarity 0.88)
- EventPillars: Pillar-based Efficient Representations for Event Data (AAAI 2025 · similarity 0.88)
- Event-Driven Dynamic Scene Depth Completion (NeurIPS 2025 · similarity 0.87)
- EGSST: Event-based Graph Spatiotemporal Sensitive Transformer for Object Detection (NeurIPS 2024 · similarity 0.87)
- FlexEvent: Towards Flexible Event-Frame Object Detection at Varying Operational Frequencies (NeurIPS 2025 · similarity 0.87)
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.
