Research map / Video / Video segmentation and tracking
Reducing the Sim-to-Real Gap for Event Cameras
Timo Stoffregen, Cedric Scheerlinck, Davide Scaramuzza, Tom Drummond, Nick Barnes, Lindsay Kleeman, Robert Mahony
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 |
|---|---|---|---|
| retrain E2VID on ESIM data with matched contrast thresholds and diverse motion | E2VID · HQF | LPIPS: 0.4609 -> 0.2562 | Improved |
| retrain E2VID on ESIM data with matched contrast thresholds and diverse motion | E2VID · IJRR | LPIPS: 0.2813 -> 0.2240 | Improved |
| retrain E2VID on ESIM data with matched contrast thresholds and diverse motion | E2VID · MVSEC | LPIPS: 0.6457 -> 0.4670 | Improved |
| retrain EV-FlowNet on simulated data instead of real MVSEC data | EV-FlowNet · HQF | FWL: 1.2045 -> 1.3540 | Improved |
| retrain EV-FlowNet on simulated data instead of real MVSEC data | EV-FlowNet · MVSEC | FWL: 1.11856 -> 1.29962 | Improved |
| train on the full range of contrast thresholds (All) rather than a single high threshold 1.50 (ablation) | E2VID · MVSEC | LPIPS: 0.52 -> 0.51 | Improved |
| use low training contrast threshold 0.2 instead of 1.00 for IJRR evaluation (ablation) | E2VID · IJRR | LPIPS: 0.29 -> 0.25 | Improved |
| use high training contrast threshold 1.00 instead of 0.2 for MVSEC evaluation (ablation) | E2VID · MVSEC | LPIPS: 0.55 -> 0.51 | Improved |
| add noise augmentation at sequence length 40 (ablation) | E2VID · HQF | LPIPS: 0.2964 -> 0.2562 | Improved |
| add noise augmentation at sequence length 40 (ablation) | E2VID · MVSEC | LPIPS: 0.5255 -> 0.4670 | Improved |
| add noise augmentation at sequence length 120 (ablation) | E2VID · HQF | LPIPS: 0.2791 -> 0.2904 | Hurt |
| increase training sequence length from 40 to 120 (ablation) | E2VID · HQF | LPIPS: 0.2964 -> 0.2791 | No clear effect |
Closest papers by meaning
- Events-To-Video: Bringing Modern Computer Vision to Event Cameras (CVPR 2019 · similarity 0.91)
- High Speed and High Dynamic Range Video with an Event Camera (TPAMI 2019 · similarity 0.91)
- Video to Events: Recycling Video Datasets for Event Cameras (CVPR 2020 · similarity 0.91)
- E2HQV: High-Quality Video Generation from Event Camera via Theory-Inspired Model-Aided Deep Learning (AAAI 2024 · similarity 0.89)
- LED: A Large-scale Real-world Paired Dataset for Event Camera Denoising (CVPR 2024 · similarity 0.88)
- Graph Neural Network Combining Event Stream and Periodic Aggregation for Low-Latency Event-based Vision (CVPR 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.
