Research map / Vision-language and multimodal / Cross-modal learning
Hidden Gems: 4D Radar Scene Flow Learning Using Cross-Modal Supervision
Read the paperCheck 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 cross-modal supervision from odometer, LiDAR, and camera instead of no modality supervision (ablation) | CMFlow · VoD Test | EPE [m]: 0.228 -> 0.141 | Helped |
| Add odometer supervision (ablation) | CMFlow · VoD Test | EPE [m]: 0.228 -> 0.161 | Helped |
| Add LiDAR supervision to odometer supervision (ablation) | CMFlow · VoD Test | EPE [m]: 0.161 -> 0.145 | Helped |
| Add camera supervision to odometer supervision (ablation) | CMFlow · VoD Test | EPE [m]: 0.161 -> 0.159 | Helped |
| Add temporal update module (ablation) | CMFlow · VoD Test | EPE [m]: 0.141 -> 0.130 | Helped |
| Add temporal update module (ablation) | CMFlow · VoD Test | AccS: 0.233 -> 0.228 | Hurt |
| Use odometer-derived moving label S^v (ablation) | CMFlow motion segmentation · VoD Test | mIoU (%): 46.9 -> 52.8 | Helped |
| Add LiDAR-derived label S^l to S^v (ablation) | CMFlow motion segmentation · VoD Test | mIoU (%): 52.8 -> 54.1 | Helped |
| Add extra unannotated training data (ablation) | CMFlow motion segmentation · VoD Test | mIoU (%): 54.1 -> 57.1 | Helped |
| Add odometer supervision (ablation) | CMFlow ego-motion estimation · VoD Test | RAE [°]: 0.336 -> 0.183 | Helped |
| Add LiDAR and camera supervision to odometer supervision (ablation) | CMFlow ego-motion estimation · VoD Test | RAE [°]: 0.183 -> 0.145 | Helped |
| Add temporal update module (ablation) | CMFlow ego-motion estimation · VoD Test | RTE [m]: 0.071 -> 0.066 | Helped |
Closest papers by meaning
- Weakly Supervised Cross-Modal Learning for 4D Radar Scene Flow Estimation (ICML 2026 · similarity 0.95)
- SCKD: Semi-Supervised Cross-Modality Knowledge Distillation for 4D Radar Object Detection (AAAI 2025 · similarity 0.85)
- V2X-Radar: A Multi-modal Dataset with 4D Radar for Cooperative Perception (NeurIPS 2025 · similarity 0.84)
- Self-Supervised Bird’s Eye View Motion Prediction with Cross-Modality Signals (AAAI 2024 · similarity 0.83)
- CRKD: Enhanced Camera-Radar Object Detection with Cross-modality Knowledge Distillation (CVPR 2024 · similarity 0.83)
- H-Flow: Self-supervised Human Scene Flow via Physics-inspired Joint Multi-modal Learning (NeurIPS 2026 · 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.