supervision 0.12.0

A set of easy-to-use utils that will come in handy in any Computer Vision project

👋 hello

We write your reusable computer vision tools. Whether you need to load your dataset from your hard drive, draw detections on an image or video, or count how many detections are in a zone. You can count on us! 🤝

💻 install

Pip install the supervision package in a 3.11>=Python>=3.8 environment.

pip install supervision[desktop]

Read more about desktop, headless and local installation in our guide.

🔥 quickstart

detections processing

>>> import supervision as sv
>>> from ultralytics import YOLO

>>> model = YOLO('yolov8s.pt')
>>> result = model(IMAGE)[0]
>>> detections = sv.Detections.from_yolov8(result)

>>> len(detections)
5

👉 more detections utils

• Easily switch inference pipeline between supported object detection / instance segmentation models

  >>> import supervision as sv
  >>> from segment_anything import sam_model_registry, SamAutomaticMaskGenerator
  
  >>> sam = sam_model_registry[MODEL_TYPE](checkpoint=CHECKPOINT_PATH).to(device=DEVICE)
  >>> mask_generator = SamAutomaticMaskGenerator(sam)
  >>> sam_result = mask_generator.generate(IMAGE)
  >>> detections = sv.Detections.from_sam(sam_result=sam_result)
  

• Advanced filtering

  >>> detections = detections[detections.class_id == 0]
  >>> detections = detections[detections.confidence > 0.5]
  >>> detections = detections[detections.area > 1000]
  

• Image annotation

  >>> import supervision as sv
  
  >>> box_annotator = sv.BoxAnnotator()
  >>> annotated_frame = box_annotator.annotate(
  ...     scene=IMAGE,
  ...     detections=detections
  ... )
  

datasets processing

>>> import supervision as sv

>>> dataset = sv.DetectionDataset.from_yolo(
...     images_directory_path='...',
...     annotations_directory_path='...',
...     data_yaml_path='...'
... )

>>> dataset.classes
['dog', 'person']

>>> len(dataset)
1000

👉 more dataset utils

• Load object detection / instance segmentation datasets in one of supported formats

  >>> dataset = sv.DetectionDataset.from_yolo(
  ...     images_directory_path='...',
  ...     annotations_directory_path='...',
  ...     data_yaml_path='...'
  ... )
  
  >>> dataset = sv.DetectionDataset.from_pascal_voc(
  ...     images_directory_path='...',
  ...     annotations_directory_path='...'
  ... )
  
  >>> dataset = sv.DetectionDataset.from_coco(
  ...     images_directory_path='...',
  ...     annotations_path='...'
  ... )
  

• Loop over dataset entries

  >>> for name, image, labels in dataset:
  ...     print(labels.xyxy)
  
  array([[404.      , 719.      , 538.      , 884.5     ],
         [155.      , 497.      , 404.      , 833.5     ],
         [ 20.154999, 347.825   , 416.125   , 915.895   ]], dtype=float32)
  

• Split dataset for training, testing and validation

  >>> train_dataset, test_dataset = dataset.split(split_ratio=0.7)
  >>> test_dataset, valid_dataset = test_dataset.split(split_ratio=0.5)
  
  >>> len(train_dataset), len(test_dataset), len(valid_dataset)
  (700, 150, 150)
  

• Merge multiple datasets together

  >>> ds_1 = sv.DetectionDataset(...)
  >>> len(ds_1)
  100
  >>> ds_1.classes
  ['dog', 'person']
  
  >>> ds_2 = sv.DetectionDataset(...)
  >>> len(ds_2)
  200
  >>> ds_2.classes
  ['cat']
  
  >>> ds_merged = sv.DetectionDataset.merge([ds_1, ds_2])
  >>> len(ds_merged)
  300
  >>> ds_merged.classes
  ['cat', 'dog', 'person']
  

• Save object detection / instance segmentation datasets in one of supported formats

  >>> dataset.as_yolo(
  ...     images_directory_path='...',
  ...     annotations_directory_path='...',
  ...     data_yaml_path='...'
  ... )
  
  >>> dataset.as_pascal_voc(
  ...     images_directory_path='...',
  ...     annotations_directory_path='...'
  ... )
  
  >>> dataset.as_coco(
  ...     images_directory_path='...',
  ...     annotations_path='...'
  ... )
  

• Convert labels between supported formats

  >>> sv.DetectionDataset.from_yolo(
  ...     images_directory_path='...',
  ...     annotations_directory_path='...',
  ...     data_yaml_path='...'
  ... ).as_pascal_voc(
  ...     images_directory_path='...',
  ...     annotations_directory_path='...'
  ... )
  

• Load classification datasets in one of supported formats

  >>> cs = sv.ClassificationDataset.from_folder_structure(
  ...     root_directory_path='...'
  ... )
  

• Save classification datasets in one of supported formats

  >>> cs.as_folder_structure(
  ...     root_directory_path='...'
  ... )
  

model evaluation

>>> import supervision as sv

>>> dataset = sv.DetectionDataset.from_yolo(...)

>>> def callback(image: np.ndarray) -> sv.Detections:
...     ...

>>> confusion_matrix = sv.ConfusionMatrix.benchmark(
...     dataset = dataset,
...     callback = callback
... )

>>> confusion_matrix.matrix
array([
    [0., 0., 0., 0.],
    [0., 1., 0., 1.],
    [0., 1., 1., 0.],
    [1., 1., 0., 0.]
])

🎬 tutorials

Accelerate Image Annotation with SAM and Grounding DINO

Created: 20 Apr 2023 | Updated: 20 Apr 2023

Discover how to speed up your image annotation process using Grounding DINO and Segment Anything Model (SAM). Learn how to convert object detection datasets into instance segmentation datasets, and see the potential of using these models to automatically annotate your datasets for real-time detectors like YOLOv8...

SAM - Segment Anything Model by Meta AI: Complete Guide

Created: 11 Apr 2023 | Updated: 11 Apr 2023

Discover the incredible potential of Meta AI's Segment Anything Model (SAM)! We dive into SAM, an efficient and promptable model for image segmentation, which has revolutionized computer vision tasks. With over 1 billion masks on 11M licensed and privacy-respecting images, SAM's zero-shot performance is often competitive with or even superior to prior fully supervised results...

📚 documentation

Visit our documentation page to learn how supervision can help you build computer vision applications faster and more reliably.

🏆 contribution

We love your input! Please see our contributing guide to get started. Thank you 🙏 to all our contributors!