https://github.com/ccomkhj/sahi

Framework agnostic sliced/tiled inference + interactive ui + error analysis plots

Last synced: 6 months ago · JSON representation

Repository

Framework agnostic sliced/tiled inference + interactive ui + error analysis plots

Basic Info

Host: GitHub
Owner: ccomkhj
License: mit
Language: Python
Default Branch: main
Homepage: https://ieeexplore.ieee.org/document/9897990
Size: 65.7 MB

Statistics

Stars: 0
Watchers: 0
Forks: 0
Open Issues: 0
Releases: 0

Fork of obss/sahi

Created almost 2 years ago · Last pushed over 1 year ago

https://github.com/ccomkhj/sahi/blob/main/



  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 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.

| Command  | Description  |
|---|---|
| [predict](https://github.com/obss/sahi/blob/main/docs/cli.md#predict-command-usage)  | perform sliced/standard video/image prediction using any [ultralytics](https://github.com/ultralytics/ultralytics)/[mmdet](https://github.com/open-mmlab/mmdetection)/[detectron2](https://github.com/facebookresearch/detectron2)/[huggingface](https://huggingface.co/models?pipeline_tag=object-detection&sort=downloads)/[torchvision](https://pytorch.org/vision/stable/models.html#object-detection) model |
| [predict-fiftyone](https://github.com/obss/sahi/blob/main/docs/cli.md#predict-fiftyone-command-usage)  | perform sliced/standard prediction using any [ultralytics](https://github.com/ultralytics/ultralytics)/[mmdet](https://github.com/open-mmlab/mmdetection)/[detectron2](https://github.com/facebookresearch/detectron2)/[huggingface](https://huggingface.co/models?pipeline_tag=object-detection&sort=downloads)/[torchvision](https://pytorch.org/vision/stable/models.html#object-detection) model and explore results in [fiftyone app](https://github.com/voxel51/fiftyone) |
| [coco slice](https://github.com/obss/sahi/blob/main/docs/cli.md#coco-slice-command-usage)  | automatically slice COCO annotation and image files |
| [coco fiftyone](https://github.com/obss/sahi/blob/main/docs/cli.md#coco-fiftyone-command-usage)  | explore multiple prediction results on your COCO dataset with [fiftyone ui](https://github.com/voxel51/fiftyone) ordered by number of misdetections |
| [coco evaluate](https://github.com/obss/sahi/blob/main/docs/cli.md#coco-evaluate-command-usage)  | evaluate classwise COCO AP and AR for given predictions and ground truth |
| [coco analyse](https://github.com/obss/sahi/blob/main/docs/cli.md#coco-analyse-command-usage)  | calculate and export many error analysis plots |
| [coco yolov5](https://github.com/obss/sahi/blob/main/docs/cli.md#coco-yolov5-command-usage)  | automatically convert any COCO dataset to [ultralytics](https://github.com/ultralytics/ultralytics) format |

## Quick Start Examples

[ List of publications that cite SAHI (currently 200+)](https://scholar.google.com/scholar?hl=en&as_sdt=2005&sciodt=0,5&cites=14065474760484865747&scipsc=&q=&scisbd=1)

[ List of competition winners that used SAHI](https://github.com/obss/sahi/discussions/688)

### Tutorials

- [Introduction to SAHI](https://medium.com/codable/sahi-a-vision-library-for-performing-sliced-inference-on-large-images-small-objects-c8b086af3b80)

- [Official paper](https://ieeexplore.ieee.org/document/9897990) (ICIP 2022 oral)

- [Pretrained weights and ICIP 2022 paper files](https://github.com/fcakyon/small-object-detection-benchmark)

- [Visualizing and Evaluating SAHI predictions with FiftyOne](https://voxel51.com/blog/how-to-detect-small-objects/) (2024) (NEW)

- ['Exploring SAHI' Research Article from 'learnopencv.com'](https://learnopencv.com/slicing-aided-hyper-inference/)

- ['VIDEO TUTORIAL: Slicing Aided Hyper Inference for Small Object Detection - SAHI'](https://www.youtube.com/watch?v=UuOjJKxn-M8&t=270s) (RECOMMENDED)

- [Video inference support is live](https://github.com/obss/sahi/discussions/626)

- [Kaggle notebook](https://www.kaggle.com/remekkinas/sahi-slicing-aided-hyper-inference-yv5-and-yx)

- [Satellite object detection](https://blog.ml6.eu/how-to-detect-small-objects-in-very-large-images-70234bab0f98)

- [Error analysis plots & evaluation](https://github.com/obss/sahi/discussions/622) (RECOMMENDED)

- [Interactive result visualization and inspection](https://github.com/obss/sahi/discussions/624) (RECOMMENDED)

- [COCO dataset conversion](https://medium.com/codable/convert-any-dataset-to-coco-object-detection-format-with-sahi-95349e1fe2b7)

- [Slicing operation notebook](demo/slicing.ipynb)

- `YOLOX` + `SAHI` demo:  (RECOMMENDED)

- `RT-DETR` + `SAHI` walkthrough:  (NEW)

- `YOLOv8` + `SAHI` walkthrough: 

- `DeepSparse` + `SAHI` walkthrough: 

- `HuggingFace` + `SAHI` walkthrough: 

- `YOLOv5` + `SAHI` walkthrough: 

- `MMDetection` + `SAHI` walkthrough: 

- `Detectron2` + `SAHI` walkthrough: 

- `TorchVision` + `SAHI` walkthrough: 


 




### Installation






Installation details:


- Install `sahi` using pip:

```console
pip install sahi
```

- On Windows, `Shapely` needs to be installed via Conda:

```console
conda install -c conda-forge shapely
```

- Install your desired version of pytorch and torchvision (cuda 11.3 for detectron2, cuda 11.7 for rest):

```console
conda install pytorch=1.10.2 torchvision=0.11.3 cudatoolkit=11.3 -c pytorch
```

```console
conda install pytorch=1.13.1 torchvision=0.14.1 pytorch-cuda=11.7 -c pytorch -c nvidia
```

- Install your desired detection framework (yolov5):

```console
pip install yolov5==7.0.13
```

- Install your desired detection framework (ultralytics):

```console
pip install ultralytics==8.0.207
```

- Install your desired detection framework (mmdet):

```console
pip install mim
mim install mmdet==3.0.0
```

- Install your desired detection framework (detectron2):

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

- Install your desired detection framework (huggingface):

```console
pip install transformers timm
```

- Install your desired detection framework (super-gradients):

```console
pip install super-gradients==3.3.1
```



### Framework Agnostic Sliced/Standard Prediction



Find detailed info on `sahi predict` command at [cli.md](docs/cli.md#predict-command-usage).

Find detailed info on video inference at [video inference tutorial](https://github.com/obss/sahi/discussions/626).

Find detailed info on image/dataset slicing utilities at [slicing.md](docs/slicing.md).

### Error Analysis Plots & Evaluation



Find detailed info at [Error Analysis Plots & Evaluation](https://github.com/obss/sahi/discussions/622).

### Interactive Visualization & Inspection



Find detailed info at [Interactive Result Visualization and Inspection](https://github.com/obss/sahi/discussions/624).

### Other utilities

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

Find detailed info on MOT utilities (ground truth dataset creation, exporting tracker metrics in mot challenge format) at [mot.md](docs/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](https://github.com/ultralytics/yolov5/releases), [MMDetection models](https://github.com/open-mmlab/mmdetection/blob/master/docs/en/model_zoo.md), [Detectron2 models](https://github.com/facebookresearch/detectron2/blob/main/MODEL_ZOO.md), and [HuggingFace object detection models](https://huggingface.co/models?pipeline_tag=object-detection&sort=downloads). Moreover, it is easy to add new frameworks.

All you need to do is, create a new .py file under [sahi/models/](https://github.com/obss/sahi/tree/main/sahi/models) folder and create a new class in that .py file that implements [DetectionModel class](https://github.com/obss/sahi/blob/7e48bdb6afda26f977b763abdd7d8c9c170636bd/sahi/models/base.py#L12). You can take the [MMDetection wrapper](https://github.com/obss/sahi/blob/7e48bdb6afda26f977b763abdd7d8c9c170636bd/sahi/models/mmdet.py#L18) or [YOLOv5 wrapper](https://github.com/obss/sahi/blob/7e48bdb6afda26f977b763abdd7d8c9c170636bd/sahi/models/yolov5.py#L17) as a reference.

Before opening a PR:

- Install required development packages:

```bash
pip install -e ."[dev]"
```

- Reformat with black and isort:

```bash
python -m scripts.run_code_style format
```

## Contributors



Fatih Cagatay Akyon

Sinan Onur Altinuc

Devrim Cavusoglu

Cemil Cengiz

Ogulcan Eryuksel

Kadir Nar

Burak Maden

Pushpak Bhoge

M. Can V.

Christoffer Edlund

Ishwor

Mehmet Ecevit

Kadir Sahin

Wey

Youngjae

Alzbeta Tureckova

So Uchida

Yonghye Kwon

Neville

Janne Myr

Christoffer Edlund

Ilker Manap

Nguyn Th An

Wei Ji

Aynur Susuz

Pranav Durai

Lakshay Mehra

Karl-Joan Alesma

Jacob Marks

William Lung

Amogh Dhaliwal