Tools for computer vision projects
Developed by Roboflow, Supervision offers reusable auxiliary tools and functions for computer vision projects. This Python-based library accelerates development workflows by facilitating standard operations in processes such as object detection and tracking.
Updates
- September 18, 2026: Stars 50,140 → 50,896, latest release 0.30.4 (September 17, 2026).
- September 15, 2026: Stars 49,875 → 50,140, latest release 0.30.3 (September 14, 2026).
- September 4, 2026: Stars 49,757 → 49,875, latest release 0.30.2 (September 4, 2026).
- August 27, 2026: Stars 49,033 → 49,757, latest release 0.30.1 (August 24, 2026).
What you get
- It accelerates data loading and processing processes in computer vision projects.
- It simplifies application development by standardizing operations such as object detection and tracking.
- It provides visualization and data set management by working compatible with different model libraries.
Installation
pip install supervisionRunning it
import cv2
import supervision as sv
image = cv2.imread(...)
detections = sv.Detections(...)
box_annotator = sv.BoxAnnotator()
annotated_frame = box_annotator.annotate(scene=image.copy(), detections=detections)If you don't write code
I installed the library with the pip install supervision command in a Python 3.9 or above environment. I want to visualize object detection results and manage my dataset in my computer vision project. How can I mark object detection results on an image using the Supervision library and how can I load and convert datasets in different formats (COCO, YOLO, etc.)? Please help me create a sample workflow using the annotator and dataset helper tools provided by the library.
Related dictionary terms
Links
TreScout did not build this tool · we found it in GitHub trends and wrote it up. This page describes the repository as of 2026-06-09: The star count and our text belong to that day, the repository may have changed since. Check the repository link for the current state. This page was machine-translated from the Turkish original · the Turkish version prevails.