SAHI: Slicing Aided Hyper Inference

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

Overview

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

Command	Description
predict	perform sliced/standard video/image prediction using any yolov5/mmdet/detectron2/huggingface model
predict-fiftyone	perform sliced/standard prediction using any yolov5/mmdet/detectron2/huggingface model and explore results in fiftyone app
coco slice	automatically slice COCO annotation and image files
coco fiftyone	explore multiple prediction results on your COCO dataset with fiftyone ui ordered by number of misdetections
coco evaluate	evaluate classwise COCO AP and AR for given predictions and ground truth
coco analyse	calcualate and export many error analysis plots
coco yolov5	automatically convert any COCO dataset to yolov5 format

Quick Start Examples

📜 List of publications that cite SAHI (currently 20+)

🏆 List of competition winners that used SAHI

Tutorials

Introduction to SAHI
Official paper (ICIP 2022 oral) (NEW)
Pretrained weights and ICIP 2022 paper files
Video inference support is live
Kaggle notebook
Satellite object detection
Error analysis plots & evaluation (NEW)
Interactive result visualization and inspection (NEW)
COCO dataset conversion
Slicing operation notebook
YOLOX + SAHI demo: (RECOMMENDED)
YOLOv5 + SAHI walkthrough:
MMDetection + SAHI walkthrough:
Detectron2 + SAHI walkthrough:
HuggingFace + SAHI walkthrough: (NEW)
TorchVision + SAHI walkthrough: (NEW)

Installation

Installation details:

Install sahi using pip:

pip install sahi

On Windows, Shapely needs to be installed via Conda:

conda install -c conda-forge shapely

Install your desired version of pytorch and torchvision:

conda install pytorch=1.12.1 torchvision=0.13.1 cudatoolkit=11.3 -c pytorch

Install your desired detection framework (yolov5):

pip install yolov5==6.2.3

Install your desired detection framework (mmdet):

pip install mmcv-full==1.7.0 -f https://download.openmmlab.com/mmcv/dist/cu113/torch1.11.0/index.html

pip install mmdet==2.26.0

Install your desired detection framework (detectron2):

pip install detectron2 -f https://dl.fbaipublicfiles.com/detectron2/wheels/cu113/torch1.10/index.html

Install your desired detection framework (huggingface):

pip install transformers timm

Framework Agnostic Sliced/Standard Prediction

Find detailed info on sahi predict command at cli.md.

Find detailed info on video inference at video inference tutorial.

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

Error Analysis Plots & Evaluation

Find detailed info at Error Analysis Plots & Evaluation.

Interactive Visualization & Inspection

Find detailed info at Interactive Result Visualization and Inspection.

Other utilities

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

Find detailed info on MOT utilities (ground truth dataset creation, exporting tracker metrics in mot challenge format) at mot.md.

Citation

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}
}

Contributing

sahi library currently supports all YOLOv5 models, MMDetection models, Detectron2 models, and HuggingFace object detection models. Moreover, it is easy to add new frameworks.

All you need to do is, create a new .py file under sahi/models/ folder and create a new class in that .py file that implements DetectionModel class. You can take the MMDetection wrapper or YOLOv5 wrapper as a reference.

Before opening a PR: