| # 🧠TTA-S2R: Pretrained YOLO Models Collection | |
| This repository contains the pretrained models used in the **TTA-S2R (Tidal Turbine Assembly – Sim2Real)** pipeline. | |
| All models were trained and evaluated as part of the study on **sim-to-real transfer for industrial assembly object detection**. | |
| The collection includes YOLOv8, YOLOv9, and YOLO-World models trained under three regimes: | |
| - **Controlled only** – trained purely on real controlled data captured with a collaborative robot. | |
| - **Synthetic only** – trained purely on domain-randomized synthetic data. | |
| - **Synthetic + Controlled** – fine-tuned using both sources for improved sim-to-real generalization. | |
| --- | |
| ## 🧩 Usage | |
| You can load any model directly in Python using the [Ultralytics](https://github.com/ultralytics/ultralytics) framework: | |
| ```python | |
| from ultralytics import YOLO, YOLOWorld | |
| # Example for YOLOv8 or YOLOv9 | |
| model = YOLO("best_model_8_synthetic_controlled.pt") | |
| # Example for YOLO-World | |
| model = YOLOWorld("best_model_yoloworld_synthetic_controlled.pt") | |
| results = model.predict(source="path/to/image_or_video.jpg") | |