Supervision: Ultimate 2024 Vision Tool – 14.6K+ Stars on GitHub!

Supervision is an open-source computer vision tool library developed by Roboflow, designed to help developers quickly build and deploy computer vision applications. With over 14,600 stars on GitHub, this project has gained significant traction in the developer community.

The library offers a comprehensive suite of tools and interfaces for various computer vision tasks, including:

• Loading datasets
• Performing object detection
• Tracking objects
• Classification
• Annotating images and videos

Whether you need to load datasets, draw detection boxes on images or videos, or count detections in specific areas, Supervision has you covered.

Key Applications

Supervision can be utilized in a wide range of computer vision-related scenarios, including but not limited to:

1. Real-time object detection and tracking in images and videos
2. Dataset manipulation (segmentation, merging, loading, and saving)
3. Creation of custom annotations to visualize detection results (e.g., bounding boxes, labels)
4. Integration with Roboflow Inference for model inference using Roboflow API KEY
5. Educational and research purposes as a tool for learning and experimenting with computer vision

Getting Started with Supervision

System Requirements

Before installing Supervision, ensure that your Python environment is version 3.8 or higher. You can check your Python version by running python --version in your terminal or command prompt.

Installation

The simplest way to install Supervision is using pip, Python’s package manager. Open your terminal or command prompt and enter the following command:

pip install supervision

For users of conda or mamba environments, or those who prefer to install from source, please refer to the official Supervision installation guide at https://roboflow.github.io/supervision/.

Quick Start Guide

Working with Models

Supervision supports various model types, including classification, detection, and segmentation models. It also provides connectors for popular libraries such as Ultralytics, Transformers, and MMDetection.

Here’s a basic example using YOLO:

import cv2
import supervision as sv
from ultralytics import YOLO

image = cv2.imread(...)
model = YOLO('yolov8s.pt')
result = model(image)[0]
detections = sv.Detections.from_ultralytics(result)

print(len(detections))  # Outputs the number of detected objects

Using Annotators

Supervision offers highly customizable annotators for visualizing detection results:

import cv2
import supervision as sv

image = cv2.imread(...)
detections = sv.Detections(...)

bounding_box_annotator = sv.BoundingBoxAnnotator()
annotated_frame = bounding_box_annotator.annotate(
    scene=image.copy(),
    detections=detections
)

Handling Datasets

Supervision provides tools for loading, segmenting, merging, and saving datasets, supporting formats like YOLO, PASCAL VOC, and COCO:

import supervision as sv

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

print(dataset.classes)  # Outputs the list of classes in the dataset
print(len(dataset))  # Outputs the number of samples in the dataset

Learning Resources

To master Supervision, consider exploring these valuable resources:

1. Usage Guide
2. End-to-End Examples
3. Cookbooks
4. Official Documentation

Conclusion

Supervision is a powerful and versatile tool for computer vision tasks, offering a wide range of functionalities for developers and researchers alike. Its growing popularity on GitHub is a testament to its utility and effectiveness in the field. Whether you’re working on object detection, dataset manipulation, or custom annotations, Supervision provides the tools you need to streamline your computer vision workflow.

As the project continues to evolve, be sure to check the official GitHub page for the most up-to-date information on features and capabilities.