Repository files navigation

SAHI: Slicing Aided Hyper Inference

A lightweight vision library for performing large scale object detection & instance segmentation

Object detection and instance segmentation are by far the most important applications in Computer Vision. However, the detection of small objects and inference on large images still need to be improved in practical usage. Here comes the SAHI to help developers overcome these real-world problems with many vision utilities.

📜 List of publications that cite SAHI (currently 300+)

🏆 List of competition winners that used SAHI

• Introduction to SAHI

• Official paper (ICIP 2022 oral)

• Pretrained weights and ICIP 2022 paper files

• 2025 Video Tutorial (RECOMMENDED)

• Visualizing and Evaluating SAHI predictions with FiftyOne

• 'Exploring SAHI' Research Article from 'learnopencv.com'

• Slicing Aided Hyper Inference Explained by Encord

• 'VIDEO TUTORIAL: Slicing Aided Hyper Inference for Small Object Detection - SAHI'

• Video inference support is live

• Kaggle notebook

• Satellite object detection

• Error analysis plots & evaluation (RECOMMENDED)

• Interactive result visualization and inspection (RECOMMENDED)

• COCO dataset conversion

• Slicing operation notebook

• YOLOX +SAHI demo:

• YOLO12 +SAHI walkthrough: (NEW)

• YOLO11-OBB +SAHI walkthrough: (NEW)

• YOLO11 +SAHI walkthrough:

• RT-DETR v2 +SAHI walkthrough: (NEW)

• RT-DETR +SAHI walkthrough: (NEW)

• DeepSparse +SAHI walkthrough:

• HuggingFace +SAHI walkthrough:

• YOLOv5 +SAHI walkthrough:

• MMDetection +SAHI walkthrough:

• TorchVision +SAHI walkthrough:

pip install sahi

conda install -c conda-forge shapely

pip install torch==2.6.0 torchvision==0.21.0 --index-url https://download.pytorch.org/whl/cu126

pip install torch==2.1.2 torchvision==0.16.2 --index-url https://download.pytorch.org/whl/cu121

pip install yolov5==7.0.14 sahi==0.11.21

pip install ultralytics>=8.3.86

pip install mimmim install mmdet==3.3.0

pip install transformers>=4.42.0 timm

Find detailed info onsahi predict command atcli.md.

Find detailed info on video inference atvideo inference tutorial.

Find detailed info on image/dataset slicing utilities atslicing.md.

Find detailed info atError Analysis Plots & Evaluation.

Find detailed info atInteractive Result Visualization and Inspection.

Find detailed info on COCO utilities (yolov5 conversion, slicing, subsampling, filtering, merging, splitting) atcoco.md.

If you use this package in your work, please cite it as:

@article{akyon2022sahi,title={Slicing Aided Hyper Inference and Fine-tuning for Small Object Detection},author={Akyon, Fatih Cagatay and Altinuc, Sinan Onur and Temizel, Alptekin},journal={2022 IEEE International Conference on Image Processing (ICIP)},doi={10.1109/ICIP46576.2022.9897990},pages={966-970},year={2022}}

@software{obss2021sahi,author       ={Akyon, Fatih Cagatay and Cengiz, Cemil and Altinuc, Sinan Onur and Cavusoglu, Devrim and Sahin, Kadir and Eryuksel, Ogulcan},title        ={{SAHI: A lightweight vision library for performing large scale object detection and instance segmentation}},month        = nov,year         =2021,publisher    ={Zenodo},doi          ={10.5281/zenodo.5718950},url          ={https://doi.org/10.5281/zenodo.5718950}}

sahi library currently supports allUltralytics (YOLOv8/v10/v11/RTDETR) models,MMDetection models,Detectron2 models, andHuggingFace object detection models. Moreover, it is easy to add new frameworks.

All you need to do is, create a new .py file undersahi/models/ folder and create a new class in that .py file that implementsDetectionModel class. You can take theMMDetection wrapper orYOLOv5 wrapper as a reference.

• Install theuv package manager on your system.
• Installpre-commit hooks withuv run pre-commit install.