GUV-Detection CV Weights

Trained detector weights for the GUV detection benchmark (YOLOv11n vs RT-DETRv3-R50),accompanying github.com/josephlaurienzo-sudo/guv-detection. v3.0 updates the two deployment files: YOLOv11_weights.pt RT-DETRv3_weights.tar Deployment weights — these two are all that is needed to run inference: YOLOv11_weights.pt — YOLOv11n deployment model (the selected detector) RT-DETRv3_weights.tar — RT-DETRv3-R50, 300-epoch converged, exported PaddleDetection inference model (model.json + model.pdiparams + infer_cfg.yml) Cross-validation weights — only needed to reproduce the 5-fold spatial CV results: yolo11n_cv_fold0..4.pt — per-fold YOLOv11n weights rtdetrv3_r50_cv_fold0..4.tar — per-fold RT-DETRv3-R50 (300ep) exported inference models Each .tar extracts to a PaddleDetection inference directory. Point the RT-DETR inferencescript's --model-dir at the extracted directory; the .pt files load directly via --weights.Unpack the deployment weights into the repository as: model_weights/YOLOv11_weights/YOLOv11_weights.pt model_weights/RT-DETRv3_weights/{model.json, model.pdiparams, infer_cfg.yml} Trained on a 5x air-objective GUV fluorescence dataset; see the repository for the fullreproduction pipeline.