Interactive Notebooks
May 31, 2026 ยท View on GitHub
Hands-on Jupyter notebooks demonstrating SAHI with different detection frameworks. Each notebook can be run directly in Google Colab or cloned from the demo directory on GitHub.
Inference Notebooks
| Notebook | Framework | Models | Links |
|---|---|---|---|
| Ultralytics | ultralytics | YOLOv8, YOLO11, YOLO26 |   |
| YOLOE | ultralytics | YOLOE variants | |
| YOLOv5 | yolov5 | YOLOv5 variants | |
| HuggingFace | huggingface | DETR, Deformable DETR, DETA | |
| GroundingDINO | huggingface | GroundingDINO zero-shot detection | |
| RT-DETR | rtdetr | RT-DETR variants | |
| MMDetection | mmdet | 300+ detection models | |
| Detectron2 | detectron2 | Detectron2 models | |
| TorchVision | torchvision | Faster R-CNN, RetinaNet, FCOS, SSD | |
| Roboflow | roboflow | RF-DETR | |
Utility Notebooks
| Notebook | Description | Links |
|---|---|---|
| Slicing | Image and COCO dataset slicing operations | |
Running Locally
Clone the repository and run notebooks with Jupyter:
git clone https://github.com/obss/sahi.git
cd sahi
pip install -e ".[dev]"
jupyter notebook demo/