![[2602.13314] Sim2Radar: Toward Bridging the Radar Sim-to-Real Gap with VLM-Guided Scene Reconstruction](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.13314] Sim2Radar: Toward Bridging the Radar Sim-to-Real Gap with VLM-Guided Scene Reconstruction
Summary
The paper presents Sim2Radar, a framework that generates synthetic radar data from RGB images, addressing the challenges of limited radar datasets and enhancing 3D radar perception through transfer learning.
Why It Matters
Sim2Radar is significant as it leverages existing visual data to create scalable radar datasets, which is crucial for improving radar perception in challenging environments. This innovation could lead to advancements in robotics and autonomous systems where reliable perception is essential.
Key Takeaways
- Sim2Radar synthesizes radar data from single-view RGB images.
- It improves 3D radar perception through transfer learning techniques.
- The framework utilizes physics-based models for realistic scene reconstruction.
- Results show measurable performance gains in radar object detection.
- This approach addresses the scarcity of annotated radar datasets.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.13314 (cs) [Submitted on 10 Feb 2026] Title:Sim2Radar: Toward Bridging the Radar Sim-to-Real Gap with VLM-Guided Scene Reconstruction Authors:Emily Bejerano, Federico Tondolo, Aayan Qayyum, Xiaofan Yu, Xiaofan Jiang View a PDF of the paper titled Sim2Radar: Toward Bridging the Radar Sim-to-Real Gap with VLM-Guided Scene Reconstruction, by Emily Bejerano and 4 other authors View PDF HTML (experimental) Abstract:Millimeter-wave (mmWave) radar provides reliable perception in visually degraded indoor environments (e.g., smoke, dust, and low light), but learning-based radar perception is bottlenecked by the scarcity and cost of collecting and annotating large-scale radar datasets. We present Sim2Radar, an end-to-end framework that synthesizes training radar data directly from single-view RGB images, enabling scalable data generation without manual scene modeling. Sim2Radar reconstructs a material-aware 3D scene by combining monocular depth estimation, segmentation, and vision-language reasoning to infer object materials, then simulates mmWave propagation with a configurable physics-based ray tracer using Fresnel reflection models parameterized by ITU-R electromagnetic properties. Evaluated on real-world indoor scenes, Sim2Radar improves downstream 3D radar perception via transfer learning: pre-training a radar point-cloud object detection model on synthetic data and fine-tuning on real radar yields up to +3.7...
