Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
LiDAS: Lighting-driven Dynamic Active Sensing for Nighttime Perception
Simon de Moreau, Andrei Bursuc, Hafid El Idrissi, Fabien Moutarde
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 |
|---|---|---|---|
| add dynamic learned illumination instead of low beam at equal power | LiDAS power 1 · Synthetic night test (YOLO11L) | mAP50: 36.9 -> 47.3 | Improved |
| add dynamic learned illumination instead of low beam at equal power | LiDAS power 1 · Synthetic night test (Mask2Former) | mIoU: 66.0 -> 72.8 | Improved |
| make light field scene-adaptive instead of the averaged static pattern | LiDAS power 1 · Synthetic night test (YOLO11L) | mAP50: 42.6 -> 47.3 | Improved |
| reduce power by 40% with LiDAS versus high beam | LiDAS power 0.6 · Synthetic night test (YOLO11L) | mAP50: 43.0 -> 45.9 | Improved |
| use LiDAS illumination on real unlit roads instead of low beam | LiDAS power 1 · Real car-mounted prototype, unlit roads (YOLOv8L-Worldv2) | mAP50: 15.8 -> 34.5 | Improved |
| use LiDAS illumination on real unlit roads instead of low beam | LiDAS power 1 · Real car-mounted prototype, unlit roads (Mask2Former) | mIoU: 10.3 -> 15.3 | Improved |
| add LiDAS on top of SoMA domain generalization | SoMA on real unlit roads · Real prototype, unlit roads, SoMA downstream | mAP50: 34.8 -> 40.3 | Improved |
| apply LiDAS in rain, unseen at training | LiDAS power 1 · Synthetic rain, unlit roads (YOLO11L, zero-shot) | mAP50: 29.3 -> 39.2 | Improved |
| apply LiDAS with street lighting present | LiDAS power 1 · Synthetic urban area with street lights (YOLO11L) | mAP50: 48.3 -> 47.4 | No clear effect |
| swap in unseen downstream detector YOLOv5L | LiDAS power 1 · Synthetic night, unseen downstream YOLOv5L | mAP50: 33.8 -> 44.8 | Improved |
| add previous light map input M_{t-1} and CoordConv (ablation) | LiDAS power 1 · Synthetic night ablation, LiDAS power 1 (YOLO11L) | mAP50: 33.0 -> 47.3 | Improved |
| add residual output (ablation) | LiDAS power 1 · Synthetic night ablation, LiDAS power 1 (YOLO11L) | mAP50: 46.5 -> 47.3 | Improved |
Closest papers by meaning
- Light the Night: A Multi-Condition Diffusion Framework for Unpaired Low-Light Enhancement in Autonomous Driving (CVPR 2024 · similarity 0.86)
- See through the Dark: Learning Illumination-affined Representations for Nighttime Occupancy Prediction (NeurIPS 2025 · similarity 0.85)
- The Parables of the Mustard Seed and the Yeast: Extremely Low-Budget, High-Performance Nighttime Semantic Segmentation (AAAI 2025 · similarity 0.84)
- Exploring the Common Appearance-Boundary Adaptation for Nighttime Optical Flow (ICLR 2024 · similarity 0.84)
- NightAdapter: Learning a Frequency Adapter for Generalizable Night-time Scene Segmentation (CVPR 2025 · similarity 0.83)
- UniLumos: Fast and Unified Image and Video Relighting with Physics-Plausible Feedback (NeurIPS 2025 · similarity 0.83)
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.
