YOLOR Radio

--- library_name: ultralytics pipeline_tag: object-detection tags: - yolo - yolov11 - object-detection - coco - mmwave - 6g - beamforming - vibe - yolor --- # YOLOR-radio      <table> <tr> <td width="30%" valign="top"> <img src="radio.png" alt="YOLOR-radio — example radio detection" width="100%"> </td> <td valign="top"> **YOLOR-radio** is a fine-tuned object detection model for BS identification for beam initialization to detect `radio` in one inference pass. The model is trained on imagery of **[Sivers Semiconductors](https://www.sivers-semiconductors.com/) 60 GHz mmWave Radio frontends (EVK06002)**. Part of the YOLOR detector family used for the Look Once, Beam Twice mmWave V2X beam-management pipeline (SECON 2026). </td> </tr> </table> Reference implementation for the paper: > Avhishek Biswas\*, Apala Pramanik\*, Eylem Ekici, Mehmet C. Vuran. > *"Look Once, Beam Twice: Camera-Primed Real-Time Double-Directional mmWave Beam Management for Vehicular Connectivity."* (\*equal contribution) > > arXiv: <https://doi.org/10.48550/arXiv.2605.05071> <p align="center"> <img src="overview2_updated.png" alt="VIBE five-stage camera-primed beam-management pipeline" width="90%"> </p> ## Quick links - Paper (arXiv): <https://doi.org/10.48550/arXiv.2605.05071> Code and Data: <https://github.com/UNL-CPN-Lab/Look-Once-Beam-Twice> | | | |---|---| | **Architecture** | YOLOv11x, 81-class output head (COCO 80 + 1 custom) | | **Initialization** | stock `yolo11x.pt` | | **Schedule** | 200 epochs, `cos_lr`, `close_mosaic=20`, `lr0=0.01` | | **Training data** | IndoorCOTSDataset — 3,599 train / 449 val / 451 test | | **Custom classes** | `radio` (id 80) | | **Released checkpoint** | `last.pt` (the converged final model) | ## Usage ```python from huggingface_hub import hf_hub_download from ultralytics import YOLO weights = hf_hub_download(repo_id="cpnlab/YOLOR-radio", filename="last.pt") model = YOLO(weights) results = model.predict("path/to/image.jpg", conf=0.25) results[0].show() ``` Clas...