Views
No views yet
 |
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 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). |
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)
| 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) |
1from huggingface_hub import hf_hub_download
2from ultralytics import YOLO
3
4weights = hf_hub_download(repo_id="cpnlab/YOLOR-radio", filename="last.pt")
5model = YOLO(weights)
6
7results = model.predict("path/to/image.jpg", conf=0.25)
8results[0].show()0–79 are the standard COCO
classes; 80 is radio. The model's names dict carries the same
mapping.1@inproceedings{biswas2026look,
2 title = {Look Once, Beam Twice: Camera-Primed Real-Time Double-Directional
3 mmWave Beam Management for Vehicular Connectivity},
4 author = {Biswas, Avhishek and Pramanik, Apala and Ekici, Eylem and Vuran, Mehmet C.},
5 booktitle = {Proc. IEEE SECON},
6 year = {2026}
7}