Research map / 3D vision / Depth and stereo
Event-Based Vision Meets Deep Learning on Steering Prediction for Self-Driving Cars
Ana I. Maqueda, Antonio Loquercio, Guillermo Gallego, Narciso García, 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 event frames rather than grayscale frames | ResNet50 steering prediction · DDD17 whole dataset | EVA: 0.300 -> 0.826 | Improved |
| Use event frames rather than grayscale frames | ResNet50 steering prediction · DDD17 whole dataset | RMSE: 8.19° -> 4.10° | Improved |
| Initialize event-based ResNet50 with ImageNet weights instead of random weights (ablation) | ResNet50 event-frame steering prediction · DDD17 whole dataset | EVA: 0.800 -> 0.826 | Improved |
| Initialize event-based ResNet50 with ImageNet weights instead of random weights (ablation) | ResNet50 event-frame steering prediction · DDD17 whole dataset | RMSE: 4.40° -> 4.10° | Improved |
| Use event-frame input instead of grayscale-difference input | ResNet50 · day | EVA: 0.653 -> 0.728 | Improved |
| Use event-frame input instead of grayscale-difference input | ResNet50 · day sun | EVA: 0.802 -> 0.805 | Improved |
| Use event-frame input instead of grayscale-difference input | ResNet50 · evening | EVA: 0.418 -> 0.602 | Improved |
| Use event-frame input instead of grayscale-difference input | ResNet50 · night | EVA: 0.621 -> 0.753 | Improved |
| Set event integration time to 10 ms rather than the best-performing 50 ms (ablation) | ResNet50 event-frame steering prediction · day sun | EVA: 0.805 -> 0.790 | Hurt |
| Set event integration time to 25 ms rather than 50 ms (ablation) | ResNet50 event-frame steering prediction · day sun | EVA: 0.805 -> 0.792 | Hurt |
| Set event integration time to 100 ms rather than 50 ms (ablation) | ResNet50 event-frame steering prediction · day sun | EVA: 0.805 -> 0.634 | Hurt |
| Set event integration time to 200 ms rather than 50 ms (ablation) | ResNet50 event-frame steering prediction · day sun | EVA: 0.805 -> 0.457 | Hurt |
Closest papers by meaning
- EventDrive: Event Cameras for Vision-Language Driving Intelligence (CVPR 2026 · similarity 0.90)
- Learning to Detect Objects with a 1 Megapixel Event Camera (NeurIPS 2020 · similarity 0.89)
- Events-To-Video: Bringing Modern Computer Vision to Event Cameras (CVPR 2019 · similarity 0.88)
- Event-Aided Time-To-Collision Estimation for Autonomous Driving (ECCV 2024 · similarity 0.87)
- EventNet: Asynchronous Recursive Event Processing (CVPR 2019 · similarity 0.87)
- PASS: Path-selective State Space Model for Event-based Recognition (NeurIPS 2025 · similarity 0.85)
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.
